Today on Decoder, we’ve got the first of a two-part series on the future of business, and I’m talking with Jonathan Kanter, the former antitrust chief for the US Department of Justice in the Biden administration. These days, he’s both a professor of law at WashU and professor of technology policy at Carnegie Mellon.

The biggest story in tech right now is the spiraling debate about AI safety and regulation. Researchers at the big AI labs including Anthropic and Google DeepMind have quit in noisy ways, saying the models pose real threats and safety isn’t being taken seriously across the industry. Other researchers have said the chance of AI killing us all is greater than 10 percent, and the CEOs of all these companies have issued various calls to slow down development and develop regulation, including asking for antitrust exemptions so they can all coordinate on safety issues.

That is a lot of ideas, and I am especially curious about the antitrust piece, which has led to accusations that these companies are seeking regulatory capture, attempting to form a cartel, and even somehow finding a way out from investor pressure ahead of their IPOs. 

Verge subscribers, don’t forget you get exclusive access to ad-free Decoder wherever you get your podcasts. Head here. Not a subscriber? You can sign up here.

So to make sense of it all, I called Jonathan Kanter. 

Jonathan brought huge antitrust cases against Google, Apple, and Ticketmaster during his time as antitrust chief, and along with Lina Khan, led what amounted to a full reboot of American antitrust policy. As you’ll hear him say, he won those cases against Google and Ticketmaster, and that case against Apple is still live in the Trump DOJ, surprising many. 

That reboot has also led to some surprising alliances in the AI debate — avowed libertarian and former Trump AI czar David Sacks has been approvingly retweeting Lina Khan saying there’s no need for an antitrust exemption.

This is all a lot, and you’ll hear Jonathan and me really get into the weeds of how all this regulation might work, and how it might interact with antitrust law, competition policy, and our relationship with China.

Okay: Jonathan Kanter, former DOJ antitrust chief, on AI safety and regulation. Here we go.

This interview has been lightly edited for length and clarity.

Jonathan Kanter, you’re the former head of the DOJ Antitrust Division in the Biden administration and a current professor of law at WashU and professor of technology policy at Carnegie Mellon. Welcome back to Decoder.

Always a pleasure to be back.

This is great. You’re a private citizen now. You don’t have to give me the politician answers.

I can say whatever I want. I feel so free.

I’m very excited about this.

I plan to as well.

I have a million questions about the state of AI regulation in America right now. There is a swirl of ideas about how we should or should not regulate these companies.

There’s the Trump administration whose posture seems to be “no,” which is fascinating. You rarely see the federal government just decline to participate in the way that the Trump administration appears to be declining to participate. And then there’s an incredible backbone of antitrust concern floating throughout all of this. Are these companies trying to form a cartel? Are they aware that maybe their IPOs aren’t going to go the way they want to, so they’re seeking regulatory capture in some way that is good or bad? Are they trying to foreclose cheaper competition from open-weight Chinese models?

Give me your sense of just the state of things right now, and then I want to dive into the specifics with you.

The state of the world right now is like we’ve invented cars and trucks, but we have no lines on the road, no traffic lights and no stop signs and no speed limits. We’ve invented this incredibly powerful technology that’s transforming how we live our lives, how we conduct business, how we socialize, how we seek and generate and consume information. We need some basic rules of the road. I think about it on two fronts. What are the things that companies should be doing? And what are things the government should be doing? There are responsibilities on both ends. I’m happy to dive into both.

Let’s just start with the very notion that asking for regulation necessarily forms a cartel. All the big frontier AI companies — OpenAI, Anthropic, Google DeepMind, even Elon Musk — do not necessarily like each other. These are people that do not necessarily like each other. They are saying in concert that they should slow down. The phrase that Dario Amodei from Anthropic is using is to “pace the frontier.”

My read of this is that they simply do not trust each other, that they are in a prisoner’s dilemma and they need an outside force, most likely the government, to enforce some kind of policy framework on them that they have to agree with. Is that a fair read of just the basics here?

Kind of. Let me give you my interpretation from the most generous to the most cynical, neither of which, TL;DR, should result in any sort of antitrust exemption. 

The most generous interpretation is that they’re afraid of the pace of innovation and the lack of regulation means they don’t see any lines on the road. They don’t know where to drive and they’re worried about driving off the road or hitting somebody else. They’re begging for the government to step in and do something. Probably there’s a modicum of truth to that. I believe they believe it could destroy humanity. I’m not sure that that doomsday scenario is accurate, but I believe they believe that. That’s the more generous version.

The more cynical version is that they’re spending so much money, they’re hemorrhaging cash and they can’t keep this up, but if they pull back, it’s going to affect their valuation as part of the IPO. So they would like somebody to give them all permission to stop spending so much money and slow the pace of innovation so that they don’t have to compete as hard so they can figure out their economics before they go public.

Does an agreement among all the top players in a field like this automatically form a cartel? On that cynical side of the interpretation, there’s a lot of, “You just want regulatory capture. You want to make a cartel. You want to foreclose competition.”

That is a risk certainly of that kind of arrangement. Let’s walk through what’s necessary and what’s not.

Let me just be very clear, these companies do not need to coordinate in order to deliver safe and secure products to society. Think about this: imagine Boeing and Airbus. Boeing fears doors falling off the plane. It wasn’t Airbus’s fault and they didn’t need to slow down innovation with Airbus in order to prevent those doors from flying off the plane. They needed to figure out how to build planes that were safer. If you build cars that explode while you’re driving, it’s not the other car company’s fault and you don’t need to come together and figure out how to solve those problems. You need to figure out what went wrong in your manufacturing process. If you create products that create AI agents that break into somebody else’s and hack somebody else’s technology, it is no different, in my mind, than having an employee go do that.

You have a responsibility to build safe and secure products and companies should be held liable when their agents, whether those agents are digital and binary or whether those agents are AI or whether those agents are humans. If they start doing bad things, the company that employs them can and should be held responsible. Those are the basic rules of the road. There are areas where there is legitimate collaboration in the context of safety and security. For example, having a clearinghouse to properly share threats, malicious bots, or other things that you can create a repository for so that everyone can build safer products. Those are the kinds of collaboration that happen in other industries and you don’t need an antitrust exemption because the antitrust laws don’t prohibit that kind of work. If two companies say, “We’re competing too hard and we need to slow down,” yes, that could implicate the antitrust laws, but that’s not what we should be doing here.

We should be innovating, but companies should have the responsibility to build safe and secure products and the government should be responsible for putting lines on the road that say, “If you don’t build a safe and secure product, here’s what the consequences are. If you don’t build a safe and secure product, here’s what liability might look like. If you don’t build a safe and secure product, then perhaps there should be robot jail for your agents and it should be taken off the market.”

I want to dig into robot jail specifically. in the amount of pressure we’re putting on what feels like product liability law. There’s something there that I want to talk about more specifically, but my wife’s a lawyer. I told her I was talking to you this morning and I tried to explain that there are fault lines here and she just shot me that look and said, “This is just a nightmare law school hypothetical.” What you’re describing is, “Does tort law create a regulatory framework for these companies?” And somewhere in there is, “If my agent goes and hacks you, my directive is that Anthropic’s at fault. Are we going to create some kind of pass-through liability? Does that happen already or not? Do we need another law or clarification law?” What’s your view on this? If I direct my Meta Muse agent to go attack you and it does it, is that Meta’s fault?

Both/and. Companies employ people and they build technology. When those people and those technologies do things that are problematic, including stealing or breaking into someone else’s property on behalf of their employer, there’s liability and that can be products liability, but it could be more than products liability. You’re just people doing bad things or companies creating technology that’s doing bad things. And the products liability framework actually did work most recently in the case against Meta for Facebook and Instagram and child safety and security and mental health.

Wait, hold on. That is really fascinating to me. This is a thing I’m really interested in talking about, because it took a decade for it to work.

That’s the problem! This is where the government needs to step in. Companies have an obligation today to build safe and secure products and the pace of innovation isn’t an excuse not to go do that. Let’s be very clear about that. At the same time, we’ve invented cars and trucks and trains and we have no crossings, we have no lights, we have no lines on the road. Congress is completely incapable of pretty much doing anything. While these tech companies are out there saying that they want regulation, they’re actively lobbying against it and have been for a very long time. The whole thing is a complete mess and it’s a symptom of our system, which is a complete mess. And frankly, if the political scientist in me would trace that back to things like gerrymandering and Citizens United where our process has gotten out of control and Congress, which should be doing its basic job of putting basic safety standards and responsibilities in place, hasn’t done that.

That having been said, the lack of action by Congress is not an excuse to deliver dangerous products. If you believe your product is going to destroy humanity, then don’t build it. If you believe your product is going to break into someone else’s stuff, then go back to the drawing board and figure out how you build it in a way that’s safer. 

Guess what? That’s innovation. These are innovation problems, not just regulatory problems. We need to do a better job incentivizing our companies to build safer and secure products. When they do that, those products will perform better and we will see better innovation. Right now, innovation is asymmetric because it’s done without the proper regard for safety and security, just like we saw in social media. As a result, now we have products that are really good and really effective at letting people communicate, but really bad when it comes to childhood addiction and mental health. Perhaps we should learn our lesson and encourage people to build safer, more secure products from the get-go, at the outset of this technology.

Let me make the comparison to social media regulation a little more tightly. We just had the big Meta case; that is a products liability case. Their products were designed in a way that caused harm to teenagers. Meta settled the case because I think Mark Zuckerberg would prefer to build AI than to care about moderating Facebook.

But they did lose one to a jury.

They usually lose one to a jury, and this case wasn’t going well, but fundamentally they settled. They gave up. There’s a view that if they had fought and appealed, they would have won on some first amendment grounds, that the product liability case maybe wasn’t as strong as we all thought it was. And the other reason I’m saying that is Meta gave in to the regulator and their posture at the end was a settlement. They gave in and they accepted some regulation. It took a long time for the product liability version of this to get a company like Meta to concede some regulation, to concede some controls—

To get a company like Meta to concede some regulation, to concede some controls. If we’re saying today, “OpenAI, you are probably liable for what ChatGPT does. Anthropic, you are probably liable for what Claude does,” isn’t this the moment to actually tighten the screw and pass the regulation that says, “You are definitely liable”?

Yes. This is the job of Congress, which is to clarify the law. But I think the environment today is noticeably different than the environment was 15–20 years ago. 15–20 years ago, we had Section 230, which actually granted massive amounts of immunity to online companies, which actually made it harder to sue them and gave them more flexibility to innovate without regard for consequences. We had — and this is not partisan because it can go back to the Obama administration — a very hands off, “Oh my goodness, our economy depends on these wonderful companies from Silicon Valley. Let’s let them do whatever they want and essentially allow them to set up camp at the White House and in Congress and tell us how to write or not write our rules.”

The results of that have eroded or destroyed any trust the public has in these technology companies. One of the reasons we’re having this conversation a lot earlier in the process now is because people are saying, “Fool me once, shame on you. Fool me twice, shame on me.” The good news is there’s greater public awareness and appreciation for the problems and a desire to do something meaningful. The problem is we have a system in Congress because of things like Citizens United, because of things like gerrymandering that have a government that is not responsive to the needs of the people. That is the impasse that we’re all confronting. We’re going to have to, in the meantime, rely on things like products liability or others to help get us some semblance of structure.

But the other piece is the states. The states have actually done a better job than the federal government. You can rewind back six months or a year, and we had people in Congress, without imposing any affirmative rules, saying we should put a 10-year moratorium on the ability of states to impose their own rules and regulations at a time when Congress had no affirmative solutions of its own. So, hopefully that moment has passed and the public is saying — and the companies are saying — “Let’s sit down and figure out something sensible.”

The polling data is clear that the American public does not trust these companies, does not like these products, and is afraid of the consequences to come, even if those consequences are hazy in the long term. They seem not so hazy in the short term.

Right. Some of this is that the companies have done a terrible job at messaging. They have effectively said, “We’re going to destroy all your jobs and potentially destroy humanity, but we’re going to IPO for a massive valuation and we’re going to set up camp in your local communities with data centers, which by the way, don’t employ people and drive up the cost of power. But, it’s okay. China.”

I want to come to China in one second, because that feels like another very important piece of the puzzle. But you mentioned a cast of characters here, you mentioned Obama. There’s a political realignment inside this debate that I find confounding. David Sacks, who was Trump’s AI czar, is retweeting your former counterpart at the FTC, Lina Khan, saying, “The existing antitrust laws are fine. You don’t need an exemption. The existing laws” — like you’re saying — “will cover you. They will make it so you can coordinate on security things and you’ll be liable for what you’re doing. And if you want to stop, just stop.” That is an absolutely shocking kind of political alignment to me.

Where do you think that’s coming from?

Actually, it’s a realignment of the realignment. Going back to Trump 1, we started seeing the kind of ends of the horseshoe starting to meet where folks left, right, and center were all concerned about the power of big tech. When people like Lina and I came into office, we had a mandate and a lot of public support, frankly, to be aggressive in enforcing the law against big tech. Then came the inevitable backlash and the Trump administration in the second incarnation was a lot more friendly with those companies rather than the first, and then it became cozy again. Now I think we’re seeing that break once more.

But if I was to describe to you a day in which David Sacks is approvingly retweeting Lina Khan, while Bloomberg columnist Joe Weisenthal is saying “All these people are just saying Lina Khan was right about the power of cartels.” And hours later, Bernie Sanders and Steve Bannon are sharing a stage in DC calling for a dead stop of AI development. That doesn’t look like anything that we’ve seen before.

No, no. Although again, I will say this. When I went up before Congress, I had support from Senator Mike Lee and Chuck Grassley and Josh Hawley and Matt Gaetz asked me if I was tired of winning at one of my oversight hearings. And then I had a bunch of Democrats, many of whom supported me, some of whom were very adherent to the big tech companies, who would criticize me. These signs of this stuff being all over the map have been around for a while. The fight between little tech and big tech — this is the part that I think a lot of folks don’t see from the outside, but the biggest antagonists of the big tech companies are the other tech companies.

I did a speech before the Chamber of Commerce when I was in office and I brought a whole box of fake mustache and glasses. And I said, “I’m here to say thank you because all your members are coming in and helping us figure out how to bring our cases. And I know you don’t like the reputational hit of that. So here are some disguises, you can put them on next time you come in.” But the truth of the matter is most people hate the monopolist unless you’re the monopolist and companies want the opportunity to compete on the merits and win or lose on the merits of their own innovations and their own business acumen.

A lot of this stuff actually has wide-range appeal. The politics of it get kind of weird and funky like you’re describing. But I think if you look at the data center issue, for example, it’s not left or right. It’s all of the above. Part of this is, to me, a situation in which we’re actually more in a healthier place, as unhealthy as the whole thing is, than we were during the internet and social media revolution where even criticizing those companies or saying we needed something was seen as a very fringe position.

The truth of the matter is people want some lines on the road. They don’t want to stop innovation. They don’t want to stop progress. They’re not saying we have to burn the whole thing down, although Bernie kind of wants a complete stoppage. I don’t think that is realistic. We don’t want to stop innovation. We just want to have innovation result in safe and secure products that make us better as a society. That requires some correction in terms of the course as opposed to the path we’re on right now.

I do think one notable difference between now and what happened with social media companies is that the people who ran Twitter and Facebook and YouTube did not run around saying they might kill everyone. They ran around with the banner of the righteous, they were going to bring democracy to the Middle East with the power of social media. These AI companies obviously think that they’re going to do something great. They’re going to cure all diseases, they’re going to make life better. We’re going to live in luxury communism or whatever that thing is. But they’re also very clear that unrestrained competition might lead to really enormous negative outcomes.

One of the things they’re constantly talking about is China. They can all coordinate in the United States, but the United States needs to win because there’s the specter that Chinese models might overtake them and then the Chinese government will be in control of AGI or whatever thing that is scary on the horizon.

What do you make of that? Is that just a convenient boogeyman? Is that real? Is that something competition law can address?

It’s largely a convenient boogeyman. We’ve been dealing with these kinds of issues. I was thinking back to the breakup of AT&T. At the time, AT&T made the argument to the Reagan administration that it should not be broken up because in order to win the Cold War, we needed a national telecommunications network that was nice and strong. And the Reagan administration rejected that argument.

The fact of the matter is the way to compete against China is to lean into what makes us great as a country. What makes us great as a country is we care about our people. Also, just let’s be very clear, China is not doing what we’re doing. China limits the way in which kids can interact with technology. China censors what people see in its borders. China actually manages and centrally controls a lot of its corporations through state owned enterprises and influenced enterprises. This idea that the only way to compete against China is to let us deliver unsafe products is kind of ridiculous.

The way we win, whatever that means, against China is to make sure that we are delivering great innovations and products that benefit society, that we are protecting people from harmful uses of those products, and then we’re allowing a free market to function within those constraints. That’s how we beat China. As I was saying before with social media, the lack of rules and the lack of accountability resulted in a lack of innovation in how to build these products at the outset in a way that’s more safe and that’s more secure and better for, let’s say, mental health of people.

It is not a bad thing and it’s not bad for innovation to require people to deliver safe and secure products. It is the responsibility of the country to make sure that our companies are innovating within the constraints, within the boundaries of a free market, but doing so in a way that is beneficial.

Let me try to split the difference between your former role as antitrust chief and your current role as law professor. Give people a framework to understand how competition law domestically should interact with foreign policy. We are in some sort of national competition with China across a number of dimensions.

There is a fear that if Chinese tech companies flood America with cheap-to-run, open-weight models, then all this collapses and the United States economy collapses. There’s a lot there, between how we might deal with China as a foreign adversary, or at least a geopolitical rival and our competition policy here. What’s the best framework to understand?

The best framework is that we should promote dynamic aggressive competition inside the borders of the United States. We want companies, including foreign companies usually, but certainly domestic companies to compete hard because our system is a market-based system. It’s one where people innovate, and the opportunity to be successful encourages people to try new things, invest money and capital and make great products. All that is good. The idea that we need monopolies at home in order to compete abroad is antithetical to our way of life. The Boston Tea Party was founded as a revolution against the British monopoly over the necessities of life. This goes back to the founding of our country. Our freedom is in part in economic freedom and we need to make sure we promote that. Saying that we should tolerate monopolies domestically in order to compete abroad is wrong.

There are legitimate questions, some of which this administration is raising, about how we deal with foreign nations that try to dump cheap products below cost and things like that. I was reading with my law students the other day a case from the 80s where the allegations were that companies in Japan were cartelizing television sets to sell them at a higher price so they can cross-subsidize below-cost pricing in the United States, in order to drive out the domestic US manufacturers. That was the height of the Chicago school libertarian movement. The Supreme Court actually used some cockamamie version of economics to suggest that wasn’t an antitrust violation, and set the stage for what happened next.

There are legitimate questions about what we need to do to make sure that there is fair competition, and a fair playing field, that we are protecting the ability of people domestically to compete successfully. But the way to deal with that question is not to allow monopolization of our markets domestically in order to compete abroad. That’s just an excuse. That is leaning away from rather than into what makes us really special as a nation.

I can already hear what our most skeptical members of the audience are saying. They are screaming, probably in their cars, that it’s already a monopoly, that Nvidia has circular financed, all of these companies are already coordinated, product liability theories of harm will just fall on their face because there’s no way OpenAI would ever sue Hugging Face, because Nvidia just bought Hugging Face and Nvidia’s the biggest investor in OpenAI. Do you need to go break that up? Do you need to go pull that all apart?

Somebody should be looking at these issues. This is what we were starting to do when we were in office. We were saying, “There are some lines on the road now and you can’t go buy or invest in any company you want. We want competition to mean competition and that might be a little frustrating on some ends, but so long as you’re within those boundaries, go for it. Go compete as hard as you can.” This administration has taken the position that we’re going to allow a small number of companies to reach a level of dominance, or oligopoly, or monopoly so that we can beat China, whatever beating China means. And that’s the piece that I don’t quite understand. What does beating China mean?

I get the idea that if China has these advanced bots that are violating and infiltrating our companies, and our government servers and the like, well fine, that’s a national security issue and we should deal with that. That’s a geopolitical issue. We should deal with that. That’s an international relations issue. We should deal with that. But the way to deal with that is not having monopolies domestically, especially when those monopolies are adherent to shareholders rather than the best interest of the public.

This is a big debate across the United States government right now. The Trump administration has taken direct investment in a bunch of American companies. They’ve talked about setting up a sovereign wealth fund based in crypto. There’s more coordination between the government and private industry in a Republican administration than you would ever expect in history.

Owning the means of production. What does that sound like to you?

And then they call everyone else communists, which is great. Again, the realignment is wild stuff that’s happening out there.

Compare owning a grocery store, a couple of grocery stores in New York to actually investing in some of the most significant technological infrastructure of our nation.

Well, the grocery stores are a huge threat to the American way of life.

There’s a piece of the puzzle here where the Trump administration’s approach is national champions. They’re just not saying it maybe as explicitly as you are. Do you think that approach can be unwound? Do you think competition law is compatible with that approach?

It is harder to unwind it than it is to build it the right way from the get go. You hit the nail on the head. National champions. China is a country of national champions. The United States is not a country of national champions. The United States is a company where anyone could grow up to be a champion. That’s the American dream. Our promise, our way of life is based on a fundamentally different idea and ideal. The more we diverge and deviate from that, the worse off we are. I do think that is a bad policy, and it will be harder to undo the more it calcifies.

Again, this is not anti-business or anti-market, just to be very clear. I’m a capitalist. I’m a card-carrying capitalist. I’m proud to say I am. The world that I want is one where everyone can be investing, everyone can be building a company, and that there are rules around the edges in terms of monopolizing markets. 

But within that, everyone should compete as hard as they can, build, grow, and do all sorts of amazing things. That is a better version of our country, one that got us to the place of economic power and dominance globally. We are more vulnerable when we have central points of failure, and we are less free when we have a small number of companies that are telling us how to live our lives. I’m for the 1,000 flowers blooming inside the borders of the United States.

It is just remarkable to me. We’re weeks away from the midterm elections. That is something a traditional conservative from the 90s would have said about the power of the free markets. What do you make of this moment? If you had to describe the policy positions of conservatives, and if you had to describe the policy positions of the median Republican and the median Democrat headed into the midterms around this issue, are they compatible? Are they at war? Is it a total scramble?

It’s a total scramble. I’ll take one issue. I don’t think what I articulated is that of a conservative from the 90s. I think the conservatives of the 90s were about antitrust and competition policy, and all of these ideas that were very central to the development and growth of our country, from the 40s to the 70s, should be thrown away and we should let industry consolidate. And we did, we saw industry consolidate. That is part of the reason why we have a lot of the problems that we have today, including the K-shaped economy, and the sense of opportunity not being widely available.

In terms of your question, it is a scramble. The plates are shifting and where everything ends up is still unclear. We are living through the transformation of our politics and our ideals, and Republicans, Democrats, conservatives, and progressives and all these other different flavors will not mean in 15 years what they meant 15 years ago. It’s a little bit unclear how that’s all going to play out, but we are literally living through that transformation right now, which is kind of remarkable. It’s an interesting thing to observe as somebody who likes to study markets, technology, and policy. Again, not all of this is bad. We do need some realignment. We need to take an inward look at the way we’re governing our society, the way we’re encouraging our markets to structure, and think about whether there are things we should do differently.

But what we are living through is two things that are happening. One is from the 80s to probably the 2010s, we were living through the Reagan era of neoliberal politics and policy. The left and right have both rebelled against that. That nail in the coffin is being hammered in as we speak. The other thing that’s happening that is even more fascinating is we have been living in the New Deal era for nearly 100 years. That era is over, and what comes next is a jump ball, but it’s over in terms of the role of government, the role of policy, the role of our international relations. All of that is being rebuilt or stripped down to the studs, including our government agencies themselves, our regulatory structure.

The way this AI issue would’ve been handled 50 years ago is we would’ve created a Department for AI Safety or something, and they would’ve promulgated rules and they would had folks looking at it. And that might not have been the right approach, but that’s what would’ve been the approach. The Supreme Court has essentially thrown that out the window. It’s no longer what we have. Now we have an executive branch that basically decides on its whims what it wants to do on a daily basis. The whole thing is just completely a mess.

Trump’s approach to this is, “You don’t need a regulator, you need a great president, and that’s me.” I saw that and I thought, “But what you would do is regulate them. So you should just hire some people to do it. You should just delegate, like any boss would do.”

This is the irony. A good friend of mine, Dave Lawrence, wrote a piece in the New York Times a few weeks ago saying just this, which is that the Supreme Court has said, “We don’t want agencies and regulators promulgating rules and answering major questions, which is supposed to be the purview of Congress.” But instead they’ve essentially allowed the executive branch to do it without any process, without any input, and essentially with very little structure, which does lead to regulatory capture, which does lead to inadequate or inequitable outcomes. That’s what we’re confronting right now.

Again, this is all consistent with what I was saying a few moments ago, which is kind of ripping down to the studs of the New Deal era. The government infrastructure that we relied on to figure out how to deal with all these things has essentially been destroyed. Now we don’t really have the resources or the structure in place to deal with these kinds of existential problems. Instead, we have Congress, and Congress is largely ineffective because of things like Citizens United and gerrymandering, which have essentially made all politics national and all politics tribal.

Let me ask you two questions to wrap this up. You mentioned some eras. You, along with Lina Khan, brought forth a new era in antitrust. Some people called it hipster antitrust. The academics called it neo-Brandeisian antitrust.

Just look at my hairline. I’m far from a hipster. You’re much cooler than I am.

But the idea that we would bring back aggressive antitrust enforcement on theories of the law that weren’t simply related to consumer prices going up or down, that was new. That was something you brought forth. You brought a lot of cases against [companies like] Apple.

Successfully. We won our cases. Our cases on the law won. We beat Google in court twice. We survived our motion to dismiss on Apple. We put together a Ticketmaster case that even when the Trump administration tried to settle it, actually won in front of a jury in court. The stuff we did worked. We brought cases on algorithmic pricing in healthcare. We fought for the public because the public wanted someone to fight for them.

The settlements in the Google cases did not achieve the results you were advocating.

Not the settlements, the remedies. Which I did not preside over. I presided over the victories. We handed the baton over to the next administration. The remedies were admittedly disappointing. Could we have achieved a different outcome? I don’t know. I would’ve liked the opportunity to try, but we didn’t. So that’s disappointing. But it doesn’t change the fact that we changed the game. We changed the approach, and the state of antitrust law and the state of competition policy was a heck of a lot better when we left it than when we started.

Do you think that that era persists? There are some people, especially in tech policy, who would say, “This was all a failure. We should call this what it was. It was a failure. It did not result in structural change and we should move on from this”?

Not at all. It’s not a failure. Did we finish the job? No. There’s a lot more to go. It relates to the challenge that we were talking about earlier, which is that it’s harder to address something after it calcifies. We were coming in after these markets had hardened, trying to figure out how to create competition.

I’ve always been fond of saying that what you really want to do is you want to protect the next inflection point. Especially in technology and platforms, what comes next is likely to be highly disruptive, and competition has its greatest promise at that moment. That’s when incumbents tend to be most aggressive in trying to prevent upstarts and others from competing.

I do think the conversation we’re having around competition, this conversation today, the fact that folks on the left and right are already concerned about monopolies or oligopolies is a sign that these issues are more easily discussed today than they were six or seven years ago.

It’s funny. The last time that you were on, I believe we were talking about the case you brought against Apple, and you made the same point, right? Right before everything changes, that’s when the monopolist gets the most aggressive. With Apple, it’s the App Store. With Google, it’s tying all of its products together in various different ways.

Here you are again, and we’re talking about what would be the challengers. We’re talking about the AI companies that might disrupt Apple’s interface. Jony Ive has gone to make hardware devices at OpenAI and Apple is suing them. There’s a lot going on here. But what strikes me is the conversation we’re having today is about the challengers, not the monopolists.

The challengers of yesterday are often the monopolists of today. I’m not against the challengers, let me be very clear. This is why I’m against the antitrust exemption. I want them to keep pushing hard. I want them to keep competing. But I also would love to see thriving domestic open-source or open-weights models to compete. We want to promote more competition, not less. That’s really what we should be focused on: making sure that the lessons we learned from the last go-around don’t translate to this one. Competition is not about weakening the large player, it’s about making sure that the smaller players have an opportunity to compete as well.

Do you think it’s surprising the Trump administration hasn’t settled the Apple case? That is the one that’s ongoing. Every day I think that it’s weird that they can’t buy their way out of that one.

It is surprising. The guy showed up with gold bars to the White House. If that doesn’t do it, what will?

Do you think it’s just because they still think there’s something there, or is it just leverage in the background?

I have no idea. It’s a strong case. It’s a good case.

You’re not a politician anymore. You can just say it. Say what you think. [Laughs]

I told you the guy showed up with gold bars! [Laughs] I mentioned before that for every monopolist, you have dozens of other companies, if not hundreds, that are upset with the monopolist. Unless we fully have embraced national champions, there are a lot of companies that are frustrated, as we put in our complaint, with the tactics of Apple to control that market and exploit that market. There are countervailing forces on the other side, if I had to guess.

I want to wrap this up with a big sweep. We’ve talked about a lot of issues. There’s a lot of directions in which you can come at how we should regulate AI at this moment. If you had to make your list of prescriptions, if you were the policymaker, how would you organize them and what would you say?

I would start with clarity around consequences for bad things. If you build dangerous products, if your door flies off your plane while it’s in the air, if your AI breaks into someone else’s company, you should be held responsible. You have an obligation to build products that don’t do that. I think that is a very good logical place to start and it is an extension of the products liability style framework. That is a good place.

We need to enumerate, and Congress should step in and say, “Here are the things we value as a society. We value mental health. We value competition. We value authenticity of information. We value intellectual property and copyright of content owners. These are the kinds of things we value and we need to make sure that those values are embedded in the lines on the road and the stop signs and the traffic lights so that everyone has clarity at the outset what’s okay and what’s not.”

Maybe we’ll get a Congress that can do that in the next term. We’ll see.

I hope so. I hope so.

We’re going to have to have you back soon, Professor Kanter. Thank you so much for being on Decoder.

My pleasure.

This interview has been lightly edited for length and clarity. 

Nick Statt: Hello and welcome to Decoder, Nilay’s show about big ideas and other problems. This is Nick Statt, senior producer. And I’m joined by our brand-new supervising producer, Greg Ott.

Greg Ott: Good day, everyone. And Hi, Nilay. Nilay is here too. He is the person who hosts the show, and his name is also in the show. So it makes sense.

Nilay Patel: It’s true. We don’t consistently say my name in the show enough. We should do it all the time. I should do it. Welcome to Decoder with Nilay Patel.

GO: Like Nick was saying, this is Nilay’s show and we are doing a mailbag episode. This is where we go through all the feedback. Because you listen to the end of every episode, we know you do. We do mention that we go through every single email that we get. Nick, you can verify that. We get a ton of email, and we do read through all of it. Is that true?

NS: We now get YouTube comments. We get many, many Verge comments. We get Spotify comments and posts on Bluesky, Threads, and elsewhere. So yeah, we get a lot of feedback.

I do feel like promising we’ll read all the YouTube comments is an emotional commitment that we’re making, but we do read everything.

GO: We do.

NS: I do read the Spotify comments. I read every one because those are specifically pretty pointed.

GO: Yeah. They’re kind of like uncut gems. Because they’re trying to build this little community, somebody might read this because it only gets like six comments as opposed to 600 for a YouTube video.

NS: We last did a mailbag episode in April, and we said we were going to do more of these. So we’re back again. This is September. We’ve had a few months. We’ve had a lot of really big interviews. So yeah, let’s just get into it.

GO: We’re going to kick things off with the “software brain” video you put out. This was one of the biggest things that people have commented on. It is far and away the most talked-about episode in the past six months. Hundreds of comments on The Verge website, hundreds of comments on YouTube, single to double-digit comments on Spotify, and lots and lots of emails. 

We want to start there because a lot of the feedback was positive and people actually rarely email us a ton of negative feedback. We get a lot of, “Attaboy.” Some guy I knew in high school just emailed me — Hi Dean — saying how much he liked our Kathy Hochul episode. So people like the show, but for this one, we got some negative stuff on The Verge that we’re going to get to in a second.

One thing we wanted to talk about though is something a reader brought up about the idea of the Nilay rants. And of you effectively pulling out your inner Dennis Miller and really giving us your take on something. For this one, one listener wrote, “Just listened to ‘People do not yearn for automation,’ and all I can say is the people do yearn for more Nilay rants.” 

So we want to know, what is the origin of this type of rant? And then how much of the show’s DNA do you think is built around this idea of you giving us the rant, and the idea of you giving us your hot take on something really, really well thought out, as opposed to the interviews we usually do?

I feel like I should make another riff on the “People do not yearn for automation,” but as Dennis Miller — just full rant.

GO: We’re going to have to pull out some really good old references.

The power of AI means Dennis Miller can deliver software right now.

GO: “Let me tell you something, babe. Sam Altman, the last time he was sitting in a Whataburger…”

Right. We could just do it. It’d be great. I’m sure we wouldn’t get sued for that.

They’re just essays. I know they’re rants and they’re delivered on YouTube and they’re very pointed in particular ways. But that’s just what I would’ve written as a column 10 years ago before we had podcasts this way. It has been very obvious to everyone that you can’t just write a column and expect to have an impact. You have to stare at the camera and do it. We’re very good at publishing that as a column and then having a transcript and polishing it and delivering it all the other ways. 

That’s the origin of it: For all the podcasting I do and as little writing as I do anymore, in my heart, I’m still a writer. Taking all of the reporting and all of the ideas that we hear and all the conversations we have and synthesizing them into ideas that you can grapple with is the point. It is really the point of running a magazine.

That’s where that comes from. Now, can I do it as often as I want to do it? I obviously can’t. The reason I did “software brain” at that moment is that the excitement around AI is so high. The hype is so high and it’s all connected to how good AI systems are at writing software and manipulating databases. But the reason that I thought, “I got to make the time to do ‘software brain’ right now,” is that I could see the hype cycle around AI failing to contend with the lack of verifiability in every domain except software. That’s getting a little blurry now.

Lots of companies are just databases. Software engineering is verifiable, you can run it through the compiler and see if it works. AI has obviously gotten really good at making software. If you believe the world is just software, that means AI has gotten good, it’s done. You can see the labs are talking about how they’ve achieved AGI or they will very soon and it is very much connected to the ability of AI to write software. I’m just not discounting that that’s important. I’m just saying once you get outside of software, verifiability gets way harder.

You can see it in the episode we did with Verge reporter Robert Hart about AI and math. There’s an amount of verifiability in math and you can burn a lot of tokens, we don’t know how many, and solve some unsolved math problems. There’s probably verifiability in various other science domains, so it’s really interesting. Is there verifiability in generating novel drugs? No, you got to inject a bunch of people with novel drugs. What are we going to do?

GO: That’s why they got so many rats in cages somewhere, it has to be done.

Right. There’s just some other thing that’s going to happen in the next 10 domains that is going to be really, really challenging. That’s where I wanted to put the marker down and say, “I have been swimming in this water for a long time with these people and the power of software is not to be denied, it has remade our world. But if that’s the only framework or you think that is a universal framework, you’re going to run into these walls.” 

There are lots more of those we can do, but I want them to hit, I only want to write hit singles. Importantly, they’re all born of the reporting. That’s always been the dynamic on The Verge. Certainly the dynamic on Decoder is the data can get to your head.

The episodes where I just talk sometimes outperform the episodes with CEOs. That’s not good. That’s a bad incentive loop for me because I want everything to come out of the fact that I’ve talked to all of the people and you can go check my work. You can go see what those conversations are like and you can see that they are genuine and honest in all the things they should be.

That’s where I feel like I not only have the depth and the foundation to write something like “software brain,” but I can send it to those people and say, “This is how I feel about it.” They come on the show and they talk to me about Google Zero or “software brain” and all that stuff, and that conversation gets much bigger than me just ranting.

NS: My reaction to that, Nilay, is that you feel like your thesis for “software brain” has borne out over the last few months. We did get feedback from people who disagreed with some of the premises.

It wasn’t that they’re saying people aren’t unhappy with AI, but they think that we’re wrong about why they’re unhappy with AI. One reader, Josh Vanderberg, wrote in to say that he thinks it’s not that people hate AI as a technology, but rather that they don’t like their current economic situation and they’re projecting that onto AI. He’s pretty sure that we’ll eventually embrace natural language as the superior way of doing things on a computer.

He added, “All modern computer interfaces suck hard. We’re forced to map our fluid non-deterministic creative selves into hierarchical menu systems. No matter how intuitive and discoverable, we constantly fail to find the tools that we need.” He uses Canva’s CEO as an example of somebody that deeply understands the problem and they realize that AI can provide the user interface we’ve all been yearning for. He goes on to say, “People definitely yearn to just tell Photoshop what they want done in words and forget everything they ever learned about that horrible user interface.” 

We had a few other similar comments like, “People do not yearn for automation, they yearn for cheap stuff. And when AI gives them cheap stuff, they’ll be happy.” A third person wrote, “’Software brain’ misses three things regular people actually care about: control, agency, and meaning. We like doing things. The doing is where the meaning comes from.” 

I’m curious just from those particular comments and also the other feedback you got from “software brain,” and how you respond to those criticisms?

Let’s take them in turn. This is a very thoughtful comment, Josh. I don’t disagree that all modern computer interfaces suck hard. I’ve used Liquid Glass. I agree with you. I refuse to upgrade this computer to macOS Tahoe. I’m just not doing it. 

There’s some confusing thing happening in user interfaces, probably related to how much all of our devices have turned into shopping malls or SaaS products or whatever it is. There’s something broken there and I agree that that’s bad. I disagree that natural language is a perfect substitute for everything or can become a universal interface.

I say this as somebody who is writing a Decoder book based on all the interviews on the show. I’ve used Wispr Flow to get as much out of my head onto the page as I can. That is incredibly powerful and you can see why people like it. I also have to edit every single word it generates because it insists on formatting my thoughts in ways that I find completely bananas. All these systems say, “You think in bullets.” I say, “I am not thinking in bullets right now. I’m just pausing between sentences.”

There’s something there where you can take the natural language input and what we as people are reasonably good at doing is taking everything else with it, the context of our body language, the tone of our voice, and creating other meaning out of it. Computers are really bad at that. Maybe they’re going to get better at it, but they’re currently very, very bad at it. I’m using my Wispr Flow example. It’s just one that you can see how bad it is at that.

It’s not just you’re being creative to a computer or talking to a computer that’s going to do what you say and you don’t have to push all the buttons, it’s lossy. It’s lossy in a very important way that using a mouse to click on a button is not lossy at all. It is very, very directed and it will do exactly the thing that you want it to do. Ordering from Grubhub with your finger on a phone screen is 100 percent accurate — unless you’re drunk, which I’ve been. 

But talking to an agent to have it to go do something is lossy. It is error-prone in very specific ways. These things are going to exist together. That, to me, is the balance that we’re going to have to strike, that all of the hype around AI assumes that what I want is another person instead of a specific thing to happen. You can talk to lots of designers or professionals who use these tools and they will all complain about these tools. 

We use this app called Riverside to produce the show several times a week. Every single person on this team can issue a PhD thesis in all of the problems with Riverside because it’s the tool we use and it sucks in very specific ways. It is also still the tool we use every single week to make this show. If anybody uses Photoshop or Pro Tools… the books that can be written about people’s feelings about Pro Tools could fill a library, but they’re the tools because they get the job done in specific ways.

Making that lossy or making that non-deterministic is actually going to make people less effective. It will democratize access for a lot of people, I think, and that is a huge opportunity. I don’t think that opportunity outweighs the cost of making things more chaotic in specific ways.

The cheap stuff piece: yeah, I think people like cheap stuff. There’s a reason TVs are permanent surveillance devices and everyone has them in their home because they cost $500 and they’re just watching you all the time. But I actually think the industry is coming around to my point of view that the products aren’t good enough. 

I know this because our friend Alex Heath just interviewed Sam Altman and he asked him why people don’t like AI. Altman’s response was that he thinks that the way to win hearts and minds is to deliver value. We actually have a clip, you can run it.

Sam Altman: Generally speaking, I think the right way to get people to like something is to deliver them value… Probably if a lot of people use [agents] — which they will over time — and understand that it’s not actually fusing and destroying huge amounts of water or whatever, then they’ll be more excited.

I watched that, and I think Alex did a good job talking to Altman. My view of that is it’s a concession that Sam Altman knows ChatGPT is not delivering enough value to consumers. These free versions are not communicating enough what they could do or they’re not doing enough or they’re hallucinating too much and that’s why consumers dislike them.

I continue to believe that talking about how everything can be turned into a database that can then be automated by agents is not going to win anyone any hearts and minds. Maybe that’s the technology that creates products people love, but the technology itself is fundamentally uninspiring. It’s the products, it’s the experiences you have that change minds, and “software brain” is inherently alienating, in my opinion.

The last one, “’Software brain’ misses three things people care about: control, agency, and meaning.” Yes, I agree with that. I would connect that actually to buttons in cars. We’ve gone through this horrible period where all of the controls in cars became virtualized onto a screen. There’s been furious pushback against it and all the buttons are back. I think people want deterministic actions, even from computer-like things. 

This is maybe the single best market-driven evidence that you can’t just virtualize everything. You actually have to give real experiences to people that do the things that are printed on the label of the button exactly when you push the button. That is control and agency. I don’t know about “meaning” in changing an HVAC in a car, but it’s definitely control and agency. Meaning I think comes from having to struggle through doing the work yourself, not just letting AI do it for you.

So I agree with all these comments in different ways. I just think the industry has gotten completely lost in the sauce on what you can do if you turn everyone into a database, which is what I call “software brain,” and they’ve missed the human connection. A really good example — I’ll give you a little preview of an upcoming episode — we had the CEO of Cloudflare on, and he just laid off a bunch of people at his company because he says AI can do a bunch of those jobs.

His whole thing is he can identify rising talent in the company by having the AI look at all the things they’re doing. Then he thinks that’s good because he can make a phone call and say, “Hey, you’re a young superstar.” Maybe that is a good thing. In order to accomplish that goal, every single thing that happens to that company needs to happen in a database for an AI to review and then communicate to him. That’s the preview. We had a whole conversation about that because I think that is one of the more fascinating ideas in all of business right now.

GO: Boy, I hope that doesn’t roll out to The Verge. I’ll have to get one of those mouse movers or something. You’re going to have to start figuring out what I’m doing all day.

We are maybe insistently human. If we could automate Riverside, I think that would be good.

GO: Yeah, good luck. They did roll out the option to build an AI twin of you. The features we’ve been asking for — not better microphone connections, not better uploading. No, we need to duplicate the people we’re speaking to.

Yeah. Can we turn off the auto level control? No. But AI twins are here.

GO: There’s actually a request for an essay-style thing, a video essay that is a riff on the “software brain,” titled “People Do Not Yearn for Surveillance,” back to the whole watching-what-you’re-doing-at-work type of thing. 

Is that something that’s in your head right now, especially with all the Flock cameras everywhere, all the controversy around that? Maybe with a “software brain” format or framing or just a bigger picture thing? Is this in your mind right now?

It is a little surprising to me. I’ll candidly admit, I wasn’t expecting it to be as bipartisan as it is. Some of that is just that it’s an election year and people are angry about cameras, so all the politicians are trying to channel that anger, maybe the same way they’re trying to channel anger about data centers.

But we’ve seen there’s action in red states to ban surveillance as well, and there’s action in more Republican communities to take down the cameras, which I think is fascinating. It’s not just limited to brands like Flock, it’s the idea of these cameras watching. 

The piece of the puzzle, the reason that it caught me by surprise, is that for better or worse, my formative political experiences were 9/11 and the war in Iraq. That’s when I was in college. And what did we do? We passed the Patriot Act. We militarized the police. We started doing warrantless wiretaps. We ran our way all the way up to the Snowden revelations and then we did nothing about them. Literally none of that stuff caused this amount of panic. That’s just my framework.

This is a shock to that system. It’s surprising to me that people care now. I think it’s because the cameras are visible. The understanding that there’s an AI chatbot that you can talk to that can just go generate an answer and the cops now have it for cameras that are tracking you everywhere. And you can literally see the cameras and you can connect it to an experience you’ve had.

There’s Flock, who we are trying to get on the show, and their CEO originally tried to do a bunch of MAGA-coded messaging around this stuff. He would call his critics terrorists and that didn’t work. All that is new if you are born of the things that I was born of. All of these are criticisms of the Patriot Act in 2001, and we just did it anyway. I’m curious for the feedback from the audience on what they think shifted that made a bunch of stuff that we were happy to do in the past feel like one of the more potent political issues of our time?

GO: I tie this to the Meta glasses right now with the way those cameras are being received. Compare that to how Google Glass rolled out. If you refer back to the Patriot Act days and your formative years of tech, that’s back when cellphone cameras were seen as a joke. “Wait, I’m going to take a photo on my phone?” Right now the only thing I would take a photo with is my phone because it’s the best camera I will ever have with me. 

I find that to be a really interesting evolution, especially tied to the AI of it all. When it’s like, “No, no, these cameras now can connect to this network and have all this extra information.” It’s not just some file living on your device that you can put somewhere with your discretion. No, it’s out of your hands now. I wonder if part of that’s the anxiety around things like the Flock cameras, the glasses cameras, and all that.

I definitely think there’s some amount of anxiety around “I’m being recorded without my consent.” There’s a lot of anxiety around the intent of the people particularly wearing Meta Glasses. There’s a reason they’re called pervert glasses. Some of the biggest influencers in the world have posted videos complaining that they get stalked at the airport and people wearing Meta Glasses film them coming off of planes. This is very bad. 

It was Olivia Dunne who posted a video about that. She specifically called out Meta Glasses, which is not a great rep for those products to have. There’s something about that. There’s something about how intrusive the phones are and the cameras are and the idea that everything is recorded and meant to be shared, but we’ve been living in that for a long time. Our thesis there is that there’s a button on a phone and you pushed it and you kicked off this line of huge consequences because the camera exists in your pocket, like you’re saying, Greg.

I’m just saying there’s something that happened recently where the very notion of the police having a network of cameras that are all connected to each other and they can do an AI-powered search has just flipped over. Now people say, “We’re going to go and cover them up with spray foam or we’re going to tear them down.” Or 404 Media, which is excellent, has done so many videos finding clips of people at town hall meetings yelling about Flock cameras and being shut down by their own elected officials.

There’s something there that’s new. We’re obviously going to cover it. I’m not saying it’s so surprising we can’t engage with it. I’m saying from my worldview, the dirt that I was raised in, the thing is the same as it ever was, and that’s why it was surprising to me. We could have had this conversation about the Patriot Act. We could have had this conversation about the existence of the TSA, or any number of things, and we didn’t. We could have had this conversation about the Snowden revelations. Something flipped and to me it’s cause for investigation and also reflection.

NS: That’s a good time to switch gears to talk about our most popular episode of the year. That was, of course, our sit-down with Google CEO, Sundar Pichai. This was our fifth year in a row talking to Pichai at Google I/O, and it always generates a lot of feedback. 

This year we had commenters write in to say, “There seems to be an underlying answer of, well, we have to do all of this even if it’s bad or irresponsible.” And that person also said, “It’s a bit odd talking to him without getting some stronger concessions from Sundar.” Another said, “This interview is great, but I think Sundar got away with not having to answer the root issue of Google Zero, which is in this example, how Conde Nast were seeing a reduced click-through rate because of AI summaries.”

The question here for you, Nilay, is: Do you feel like Google is getting away with not being held responsible for what it’s doing to the web with AI? And do you think Sundar Pichai has a sense of Google’s culpability for the swath of destruction that’s being left in the wake of the AI era, and do you think he’s ever going to address or acknowledge this? Or is he still going to disagree with the premise five years from now?

That’s a lot of questions. I’m going to just issue the caveat that comes up in every single mailbag episode. Sometimes it feels to me like the audience is mad that I don’t arrest the CEOs on Decoder.

GO: [Laughs] We’ll arrest them over Riverside. Yeah, that’s a future we don’t have.

Maybe one day. Well, I was in person with Pichai. We could have cuffed him.

GO: That’s true. That’s right.

Maybe one day I’ll be granted this power. I’ll be deputized by the internet police to arrest the CEOs. But all I can do is ask the questions, and it has been five years of talking to Pichai. I hope he comes back next year. It’s remarkable that he’s the one who comes and takes the heat. We ask them all, and I can’t make them show up. They have to decide they want to do it, and he decides he wants to do it. I appreciate that he shows up and takes the heat and that he’s willing to do it.

Let’s use the Conde Nast example. I asked that question before a bunch of publishers started saying to other media outlets that they were going to stop letting Google index their sites. That’s the last turn. When they are so clear that it’s not worth it for their content to be stolen by AI because they’re getting no search traffic, that they just turn off the Google indexer, the Google crawler entirely. 

My conversation with Pichai would have been different if I had had that in my back pocket. I’m looking at these reports from various big publishers like USA Today and People Inc. And they’re all the way at, “We’re going to stop letting you index our sites because the traffic isn’t coming.” I don’t know what he would have said. I just think that answer might have been different or that the tenor of that moment might have been different.

What I have been taking from Sundar Pichai these past five years, and which I agree with, is the sense that these publishers got addicted to Google and they didn’t do anything about it. The shape of that conversation or his frustration, if you ever sense the flash of frustration that I sometimes sense is, “The web is really big. Google can measure the web. They have Chrome, they have all this data about the web. And the web is getting bigger and some parts of the web are really, really vibrant.” 

If you listen to Amy Lanzi, the CEO of Digitas, when she was on Decoder, she said, “We are getting more calls from our clients, more requests to make webpages and landing pages than ever before.” So some part of the web is growing and Google has that data. I think Pichai is saying, “These publishers are constantly complaining that I’m taking their business away and they did nothing else about it.” Every time I talk to a publisher, I ask them, “Why didn’t you do anything about it?”

I agree with that. There’s some sense that Google bears the responsibility for making the people who built a business on their platforms successful. Every platform eventually kills every business that’s built on it. YouTube will eventually kill YouTubers. Instagram will eventually kill Instagram. If you’re a Decoder listener, you know I think this. I’m always asking creators what they’re going to do about platforms and platforms when they’re going to kill creators. The web is just a platform for Google. Do I think that Google’s getting away with not being held accountable? Sure, a little bit. But the mechanism of accountability is not letting Google index your site. 

I’m very curious to see if this handful of publishers who have started making noise that they’re going to de-index their sites actually go through with it and actually impose the leverage on Google that I feel like the company has been daring them to impose. Again, I don’t want to preview the Cloudflare conversation too much. There are other mechanisms and there are other ways of doing it and that was part of that conversation as well. 

But there’s something here where this straightforward “you were sending us traffic for free and we were giving you content for free” exchange is gone. I don’t know that you can provide accountability such that Google doesn’t just have to still compete with Anthropic and OpenAI. They have to.

GO: Speaking of publishers and web traffic and all that, do you need to drop in a very nice little disclaimer about our new company and the PMX situation?

Oh, this is so complicated. So through a series of what only can be described as divorces and remarriages, Vox Media has split itself in two and The Verge is now part of something called PMX Global, which is a subsidiary of PMC, which is the Penske Media Corporation. PMC is suing Google, I believe, over AI Overviews and traffic. PMX and the thing formerly known as Vox Media are all in ad tech antitrust cases against Google.

I’m telling you this because it is a disclosure we should make. You can tell how much I know about them. They’re happening very far away from this newsroom, but it is true that our companies are involved in active litigation with Google about all this stuff.

GO: Thank you. Just want to make sure we get that out there because you flagged this the other day for when it’s relevant and it feels like with everything you just said makes that quite relevant.

Yeah, it’s in the strike zone, but it has nothing to do with us.

GO: Pivoting to something different, we had a listener talk about how our AI skepticism broadly on Decoder is Western-centric and it’s pointing out how a lot of AI adoption in China is actually different and faster and more welcomed in other ways. 

This is from Alex Wong: “Have you considered Asia and more specifically China where, by outward appearances, AI rollout adoption and image seem to be faster, deeper, broader, and more well received than in the West? Do you find this to be true? If so, what is the difference? And if so, why the difference? And if not, what are we not seeing?” 

What are your Asian sources seeing and saying? Should you have more Asian sources? Is it time for Decoder and The Verge to expand beyond Silicon Valley? This is a US-based show. Nilay is based out of New York. The production team is scattered, but we have New York offices where people come and shoot stuff. 

Do you think there’s room for more non-Western perspectives? Can you speak about the differences that are evident between the way China is approaching AI development versus in the United States? And can we really cover such a massive transcontinental gap like that? It’s a lot.

I do want to point out, as you said, we’re in New York. I actually think we need more Silicon Valley perspectives. We can be myopic. New York is all about money. To be successful in New York, you have to be like, “Literally where’s the money?”

San Francisco cares less so about the money, and more about the big ideas. Notably, the AI companies are not making money and running into some harsh realities here in New York from time to time. So there’s that.

I am not going to speculate too much on my understanding of Chinese culture. I will just offer two guesses and they’re just pure guesses. One, we can only know what the Chinese state media wants us to know. There are speech controls in that country. And so whether or not there’s massive discontent is not something that would ever come to me.

That’s just a gap and I’m cognizant of the gap. I agree that what we’re seeing there is a bit more excitement. We’re seeing a bit more investment. I would connect that — and this is my second guess — to there being literally a much stronger social safety net in China. So the idea that this will be a threat, like an existential threat to your job or your career or your livelihood or your health insurance or whatever it is, may be ameliorated by that stronger social cohesion. That is a guess, and I really hesitate to overthink that.

I would love to hear from people with closer perspectives to that, but those are the things that I’m always thinking about. I have said to leaders of big AI labs in this country, “People would like you more if you would just pay for health insurance. If you could just find a way to support Medicare for All with your excess profits and your trillion-dollar TAMs, maybe people would shut up.” And they’ve all said, “Interesting.”

I would just offer that connection because maybe the best thing AI could do for everyone is make it easier to start a business and to make it easier to have agency over your own life. The number one reason my friends with children don’t start businesses is because they need health insurance. That’s the connection. Again, that’s a very Western perspective, a very US-centric perspective specifically, but that’s why I went directly to the social safety net. Maybe you’re more excited about this thing that lets you do more things because the risk is limited in some other ways by the very nature of the country you live in. I’m open to feedback on that.

I’m responding to the question with my series of guesses and assumptions. I agree. We do now work for a much bigger company and hopefully that means I can put some reporters in China or in the region and we can get some firsthand reporting. But I’m curious for feedback from the audience because of those two things: One, it’s hard to tell what people actually feel given the speech controls that exist in that country; and two, is it just a different sense of possibility because the risks of existing are different?

NS: Do you think Sam Altman actually supports universal basic income? Is OpenAI going to fund UBI?

I don’t know the answer to that question anymore. They’ve all learned to stop issuing pronouncements about the future of society as loudly as they had been and spend way more time focused on enterprise adoption and software engineering and agentic marketing solutions and potentially “here’s how everyone will become rich,” not “here’s how everyone will lose their job.” So I don’t know the answer to that, but UBI is one of those things that it’s an idea you have because you think no one will have a job. I think they’ve all needed to stop talking about that.

NS: That actually ties very neatly into something we wanted to end this particular segment on, which is a pretty prescient email from a reader, Toms Bernards Callahan, specifically around tech backlash. He wrote in, “The algorithmic fragmentation of content consumption disrupts any organic large-scale organization. Yes, there are rallies, but no pointed policy pushes. Feeds are purposely placing users into niche consumer groups and the news has a hard time permeating across these groups.”

He says, “The irony here is that algorithmic fragmentation will make the backlash worse because when the dam breaks, the masses will be quite angry and it’ll be a large cohort of people.” He wrote us that email in April and I do want to say it feels like something has definitely happened. It feels like something definitely broke over the last three to four months since we got that email from him.

It feels like there is a large bipartisan cohort of angry people. We see this, of course, like Florida and Texas banning Flock cameras, the upcoming Florida governor’s race being about data centers. What do you make of this, Nilay? Specifically this moment of tech backlash that we’re experiencing and how The Verge might cover this from the perspective of algorithmic fragmentation? Or the perspective of this issue creating a bipartisan coalition that could kind of permeate through your filter bubble or your algorithmic feeds? 

The thing that really strikes me about the backlash right now is it’s a backlash to imposition. Not to speak for all Americans, but Americans don’t like being told what to do. “You have to accept these data centers, even though you’re telling your local governments you don’t want them” is an imposition that is just easy to be opposed to regardless of your political alignment. 

These cameras are going to watch everybody all the time and we have records that cops are watching kids in gymnastics centers. It is the most incredible imposition of all time. The nation was founded to rebel against this kind of imposition. It’s in the Bill of Rights. Of course that’s going to overcome algorithmic feeds. You can’t turn the knobs on TikTok enough to overcome that. No one can hear a fake AI influencer talking about how much they love Flock cameras and believe it’s true. It’s just not in our national character in a very specific way.

That’s like you telling everybody what’s going to happen to them and its power being imposed on them or in the case of data centers being made more expensive for them, it’s natural that that would overcome some algorithmic fragmentation. The other issue is it’s an election year. We see politicians scrambling to still play culture war games. There’s Benn Jordan, who’s a musician and a researcher who’s done really incredible work on Flock cameras. We interviewed him on The Verge before. He was just on the Tucker Carlson Show. Their views do not align. Even Benn said, “I can’t believe I’m here.” They were there to talk about Flock cameras and surveillance. That’s bananas, just purely bananas.

There’s something about that that I would connect to. At least with other big tech impositions, it felt market driven in some way. Facebook is going to be big and you’re going to be mad that Mark Zuckerberg is listening to you, but you have a desire to use Instagram. And boy, does it turn out that people’s revealed preferences are for Instagram to keep happening regardless of whether or not you think the ads are listening to you. 

There’s no market preference that’s going to take that away from you or there’s a network effect so strong, something else will happen and then we’ll try to regulate it. The market is still going to speak. People are still going to want to use this thing. That’s not true of Flock, there’s no consumer use for Flock. It’s diffuse. You have to prove the crime went down somewhere else later, but you can see it right now and it’s watching you right now and the cost is available.

AI has mostly been imposed on consumers. Every time I open Google Meet, it insistently tells me Gemini is here now. I don’t need this to be here in any way, shape, or form. That imposition where we’re claiming all the usage stats are so high, when in many cases it has just been put in front of us — there’s something about that that feels like it can transcend the algorithm.

GO: There’s something to what you were saying a few minutes ago about the Patriot Act of it all. There’s such a lack of legislation on a lot of these important issues that people feel like they’ve had no input on because they’re just rolled out. And the message is, “You should just take it.”

To your point, these people who you wouldn’t expect to go be a guest on Tucker Carlson’s program, it is this scrambling of an alignment because people don’t feel that agency. Nobody’s representing them on their behalf in a lot of ways.

Yeah, it’s election year. What are you going to do about the fact that Donald Trump is running around saying that data centers are money printing factories and you’re stupid for opposing them? 

GO: He called data centers the golden goose. “That’s the golden goose. Does everybody want to be poor?” That’s what he said the other day.

I certainly cannot predict elections and how they’ll go, but that just seems totally out of step with everybody. There’s something about that that I find utterly fascinating. To me, it just comes back to that Americans don’t like being told what to do. If you put yourself in that position, boy, do a lot of things sort of fall into place.

GO: Keeping that Gadsden flag in the air, I’ve got a transition here to your interview with Kathy Hochul, who was on the show last week as of recording time. It was received this way. One listener wrote, “Great interview job, very pointed questions. Little surprised that she didn’t seem to have thought through the ID verification implications these social media laws will lead to.” 

Some other Hochul feedback was pretty negative, predictably. Every single answer she gave could be summarized with, “It’s a campaign year. This interview was unbearable. Politicians don’t give answers to questions, they just give self-aggrandizing stump speeches.”

You have said you are honing your politician interviewing skills. What was your takeaway from this whole experience with her and how do you think it should inform how Decoder interviews politicians? Do you think we should have more of them coming into the midterms and the 2028 election? How do you see the whole politics of it all?

We definitely need to have more politicians on the show this year. I’m very sorry. But as we have been discussing, the tech policy issues in this election cycle are almost overwhelming. There might not be other issues. There’s the war in Iran, there are data centers, and there’s surveillance. Somewhere in there is healthcare. Those are the ones, at least in my opinion. 

As we talked about with Governor Hochul, a lot of them are connected. Surveillance and AI are deeply, deeply connected. Age verification and surveillance are deeply, deeply connected. It’s funny, some of the feedback we got on this episode was, “Facebook knows so much about you, why can’t it just tell you your age?”

With surveillance, it’s AI and it’s age verification. It’s all just a remix of the same ideas. How much of your computer should someone be monitoring to allow you to do some things? That’s the 3D printer debate in that conversation. I suspect we’ll end up talking to more politicians because the opportunity for realignment is so high. 

I honestly hope we get some conservative politicians to talk about data centers and surveillance because I would like to understand the realignment from that perspective. That said, I don’t think I’m great at interviewing politicians. I think I’m good at interviewing founders. I’m good at interviewing product CEOs. I’m bad at interviewing McKinsey CEOs. I’m getting better. And I’m worse at interviewing politicians. The amount of media training, it rises. That’s the thing that I’m describing here. We’ve always described the show as “Nilay versus media training” and I’m always trying to fight to a draw.

I got a draw with Governor Hochul. We got her to concede that she was out of her depth on some of these issues, particularly around age verification. I thought the moment where she said, “Tell Cody we’re going to get ahead of him next on 3D guns,” was very revealing because it’s a great spiky cowboy thing to say. I mean, that’s the moment you want from the interview with a politician. You want the sound bite. But it just struck me that that was a moment about conflict. It was not a moment about what people in our comments on the transcript were bringing up, which is whether or not most people can even print a plastic part that can withstand a bullet being fired.

I thought that interview went well from my perspective, because it just demonstrated that it’s easy to talk about what tech policy should be, but it’s very hard to talk about how you would actually implement the details of that policy. We did a good job there of even illustrating for the governor where the gaps were because she conceded it several times. I understand it’s frustrating to listen to and maybe one day I’ll just be a full cable news anchor and just repeat the same question 10 times in a row. But I don’t think anybody wants me to do that. Cable news anchors exist and they do a good job of that thing. I’m trying to do something else here.

NS: On the topic of interviewing styles and techniques, Nilay, we had an interesting interview last month with Bluesky CEO Toni Schneider. One aspect of that interview that really touched a nerve was when you asked him what he thought about Bluesky’s reputation as a liberal bubble. This is what he had to say. 

Toni Schneider: Yes, we definitely want that to change. It is already changing. It certainly wasn’t designed to attract one specific group of people. It works for anyone… We’re an open network for anyone and we don’t want to serve just one type of user base. I’ve seen it already in the last few months, really trend in the right direction where we just have more and more people show up and build their own corners of the network. It sort of balances out over time. 

This inspired a lot of conversation, not just in comments and emails, but also on Bluesky itself and the replies to The Verge‘s post about this. But one listener wrote in to say, “The idea that this one specific, very active group of people is an imbalance to be straightened out rather than the very reason the platform is what it is, is the mistake that Bluesky will come to regret. What a shame.”

Some of the feedback on Bluesky itself was predictably harsher. Somebody said, “Go ahead and attract other people, but the second we start seeing fascist stuff in our feeds that we haven’t subscribed to, you’re going to lose your base.” Somebody else said, “Toni doesn’t even understand Bluesky’s own user base.” Another person said, “Cool. I’ll just delete my Bluesky like I deleted my Twitter.”

Looking back at this exchange, it does feel like he artfully dodged it. It’s hard to tell what he really thinks about Bluesky’s reputation and the liberal politics of it all. In moments like that, Nilay, do you think it’s better to push people for a follow-up to be clearer on what they think? Or is that a case where the backlash became evident after the fact and it was better that you just kind of let him say his piece and you weren’t really trying to bait him into any one direction?

I wasn’t trying to bait him into any one direction and I understand the backlash from the Bluesky user base. I also understand what every social product CEO is struggling with, which is that you have to grow and you can run out of people. You can run out of people who want to be in that bubble. You can run out of people who only want to talk about one sports team. You can run out of people who want to be in whatever bubble X has become. We actually are seeing X run out of people who want to be in that bubble. The platform is shrinking. 

You can describe that as the capitalistic urge to grow, but the platform CEOs also understand that to remain vibrant, they have to constantly be attracting new people. Then you almost have to issue an answer like this, which is, “We’re a platform for everyone.” Toni’s riff on it was, “And we’ll give you the moderation controls to craft the experience you want to make the big platform feel small.”

I don’t know if they’re successful in communicating that to people. The big disconnect with Bluesky, in my opinion, this whole time, is that it was started by a bunch of ideological-technical people who wanted to build a standard called AT Protocol to enable interoperable social networking. Because of X, they accidentally got a user base and they’ve been in conflict with that user base ever since, which is fine. It’s obviously reflected in these comments, but Toni’s job is to make the thing a product. That’s why he’s the CEO. These answers just track every other social product answer we’ve ever gotten.

I don’t know if he’s okay with the bubble he has drifting away. I know that we can see in the stats that it’s getting smaller. I’m sure there’s going to be some revolt when they monetize it in ways that he’s described monetizing it. But if you’re a social product and you aren’t growing or trending towards growth, then in fact you are dying. There’s something about that that will not allow you to persist. That is a pattern that’s pretty repeatable and pretty understandable.

Maybe because we’ve spent so much time talking to social platform CEOs, I knew what he was getting at. I’ve internalized that answer in many different ways. I probably should have made him say it out loud. That’s a good note for me, but what I was hearing at that moment was, “You have to grow and that means you have to appeal to many more people.”

GO: Speaking of appealing to people, we want to appeal to our audience. They have some more thoughts on the show Decoder in and of itself, not just wanting more rants from you. We’ve got feedback from listeners who want to know about the kinds of relationships that you and we have with the companies that are covered on the show.

This is from Len Art who emailed to say, “Is there a possibility for you to make an episode about different ways the companies try to pressure The Verge into editorial changes? I don’t mean you to out all the company names or people making the requests, but just hearing concrete examples of how the requests come in and how they’re being dealt with could be a very compelling episode.” That’s what Len thinks.

What do you make of that? Do you worry there’s such a thing as too much transparency as we make a thing like this? There’s a lot of discretion and nuance that goes into negotiating these conversations and making sure that people want to talk to you. It’s a lot of back and forth. 

We are old enough and famously spiky enough that this stuff doesn’t happen very often. Yes, there’s the idea of too much transparency. My worry about too much transparency is that it would just be boring. We would say, “Nick and Kate spend hours scheduling.”

That’s actually the hardest part of making the show for us at this point. It’s CEOs, their comms people, and whatever news cycle they want to be on. It’s my schedule, which is bananas. Honestly, I’m telling you, and you can believe me or not, that might be the most complicated thing that we’re doing. We only make so many episodes of this thing a year and a lot of people want to be on the show. We’re lucky. That’s very lucky.

There is pressure. Big tech CEOs don’t want to be on the show. It’s not that we don’t ask them, they just are pretty honest sometimes that they think it would be too hostile of an environment. They want to be more in control of that message. You can see where they go instead. Do I want to play that game? I don’t. You can see that we just haven’t. We’re pretty direct, but I think if you listen to the conversations we have, the CEOs come on because they’re all competitive, they’re all type A. They like being challenged. They come here for a reason. We get almost more inbound requests to be on the show than we go out and ask. And meaningful, not just random cold PR pitches, but stuff we take, pitches we accept.

It is because all of those folks hear the show, they hear their peers on the show, and they want to beat them. I’ve asked about this a lot because in the early days I wanted to know why anyone comes on the show. I should understand it so that we can craft better and more compelling pitches for all the people we want to come talk to us. A lot of the answers I got were, “Because it is journalistically independent, the conflict is real.” 

That means, “Not only does the audience listen to it, but our own staff will listen to it.” If there’s some message that some CEO wants to get out, it is more likely that the staff will listen to it. Their employees will listen to it on Decoder rather than in an all-hands meeting. That is just the funniest thing in the entire world, but it’s also true.

I’ve heard a version of that many, many times. There’s pressure on The Verge, but we’re just old enough and we are famously spiky about things like being on background and off the record. Sometimes it comes to our younger reporters because they are testing to see if the younger reporters know. And then the younger reporters are like, “I’ll call Nilay,” and it goes away, which is fine. That’s my job in this newsroom. A lot of that has been offloaded onto the creator economy. It is not easy to pressure us, but it’s easy to go to some young creator who doesn’t even know that they shouldn’t be pressured or is excited about accepting conditions on access.

I see that left and right, all over the place. I don’t mean to call out anyone specific. I just think we play a role in the ecosystem where we are known for not doing that. I talk to young creators who are going to come up. Ben Smith was just on the show and he was noting that they’re all getting better at making journalism because the conflict is interesting. I see that ecosystem developing in parallel, but that’s where the pressure has really gone. It has not come to us very directly or it has stopped as we’ve gotten much older.

_{Questions or comments? Hit us up at decoder@theverge.com. We really do read every email!}

Today, I’m talking with Tom Conrad, the CEO of Sonos. Tom and I have known each other for a long time — he was the chief technology officer of Pandora, VP of product at Snap, and the chief product officer of Quibi. He was also on the board at Sonos during its disastrous 2024 app redesign that threatened to crash the entire brand. 

Tom took over as interim CEO to right the ship during that period, and then took the permanent role just about a year ago. As you’ll hear him say, he thinks Sonos is fully out of that period now, and on to its next chapter with a huge product and platform launch called Sonos 27.

Verge subscribers, don’t forget you get exclusive access to ad-free Decoder wherever you get your podcasts. Head here. Not a subscriber? You can sign up here.

I mean, come on, you all listen to the show. That is all the most Decoder bait imaginable. I have been dying to have Tom on the show to talk about how the board decided to move on from previous CEO Patrick Spence, how he diagnosed the issues inside the company, and what decisions he made to fix them.

Those fixes involved a lot of big changes: to the company’s structure, its product lineup, and how it thinks about itself and what its product fundamentally is. That’s that new platform, Sonos 27, which you’ll hear Tom call a new foundation for the company that makes every Sonos product work better as part of a system. 

There’s a lot here, including how AI assistants can work across the system, the idea of having a mesh network of little computers in your home disguised as speakers, and of course, the testy relationship between Sonos and Google, which means Google’s Gemini assistant isn’t available on Sonos devices. 

Oh, and of course we got into the weeds about smart home standards, because I am a smart home nerd and simply cannot help myself. But to his credit, Tom was willing to nerd out with me, and I think this conversation is pretty fun because of it.

Okay: Sonos CEO Tom Conrad. Here we go.

This interview has been lightly edited for length and clarity. 

Tom Conrad, you’re the CEO of Sonos. Welcome to Decoder.

Thank you. As they say, it’s so great to be here.

I am just going to say, finally, I have been asking you to be on the show for at least a year, if not before. I think the first time I met you was in the back of a car at CES when you were at Pandora.

Sounds right.

One hundred years ago.

One hundred years ago. We were children.

[Laughs] We were little babies. That car was driven by a human driver. That’s how far back it was.

That’s right.

And then you took a turn through Quibi. You and I talked through that, and now you’re at Sonos. As you know, I’m a huge fan of Sonos. We had your predecessor, Patrick Spence, on the show a bunch of times. 

Many things have happened, so I just want to start at the very beginning of this. You’re about a year into being the permanent CEO of Sonos. You have a bunch of news to announce. Just tell us what you think Sonos is today. What is your conception of this company?

When I came in the door, one of my earliest observations was that the company behaved like it was 100 percent hardware, but we’re really 100 percent hardware and 100 percent software. And if you get either one of those even a little bit wrong, the impact is tremendous.

So how does that show up really day-to-day? It was about returning the company to the idea that Sonos doesn’t make these individual products, but we really make one thing — we make the Sonos system. It all has to hang together in a cohesive, interesting, and valuable way, or we’re just not doing the work. 

There’s been a bunch of basic blocking and tackling, of course, both operationally and in the work that we’re doing in the product family. But at the highest level, returning the company to the idea that we are a system for audio in the home has been the highest-level mantra.

When you say come into the company, do you mean as a board member or as the CEO?

I mean, let’s start the starter pistol in January of 2025, 18 months ago, when I stepped in first as interim CEO and then was lucky enough six months later to be asked to take the job more permanently.

The reason I asked that question is that you obviously have two different views of the company in different roles. We did have former CEO Patrick Spence on several occasions, and he said something very similar to me when he first took over — that Sonos was stuck in its ways. It was really focused on very, very slow iterations of hardware. 

He tried very hard to push the metabolism of the company to be much faster, and he talked about that very openly with me. That led eventually to what I think is a moment of true crisis: the app launch. It was a disaster. The system stopped working. That’s when you came in to right the ship.

Tell me about that gap. You came in, and you said this company thinks it’s a hardware company. That’s what your predecessor had been saying, but it went sideways in big, meaningful ways. Tell me about that moment and why you think it was the same conception even when you came in.

So as you describe, I’ve been involved with the company going back almost 20 years. We brought Pandora to Sonos in 2007, I think, and I was able to get to know the company as a partner before I got to know it as a board member. I came onto the board in 2016 right as Patrick became CEO. So from the vantage point of a board member, I got to participate in Patrick’s chapter.

I guess the way I would characterize his high-order thinking about the company in those early days was that, as you described, he wanted to turn the company into a product execution machine with an emphasis on hardware products. We call them NPIs, or new product introductions.

In the chapter before him, it was sometimes years between new Sonos products. Spence said, “Look, we’re going to make a company that knows how to ship a couple of great new products a year.” And he did that. I mean, the company has reliably shipped a couple of great products, hardware products, every year, for close to a decade now.

Sometimes, when you energize an organization around an idea, there are unintended consequences. I think one unintended consequence of that framing was that the company became very focused on the individual products, one at a time. Even as a board member, I received Roam as a portable speaker with a certain amount of battery life, waterproofing, and Bluetooth connectivity. I didn’t really think about it as a member of the Sonos system. 

Even with the software work that we did, and with every product that we make, there is a tremendous amount of software effort and innovation that goes into them. But even the software that we made was very focused on the individual NPIs. “What can we do in software to differentiate Move 2 from other speakers in that category?”

What was hard to do, and what was hard to prioritize in the company, was horizontal slices of software experiences that benefited all of the platform. So, in a way, I think some of that debt accumulated, and the company thought that the antidote was to recreate the app experience. And as you know, we made a bunch of mistakes at that moment.

I’m really proud of the fact that I think we’re well beyond that chapter now, but we could have been much more thoughtful about how we qualified that product in the real world. We could have been much more thoughtful about how we rolled it out to customers. We could have, candidly, I think, examined the original impetus in the first place. It was very Sonos-motivated. We wanted to put the app on a modern architecture to make us more nimble.

In my chapter at Sonos, I’m really trying to bring it all back to how we can focus on the customer first. I’m so inspired by talking to people on social media and in my inbox, and people in the industry, like you, about how we show up and solve the real problems that are happening in real households. I’m so excited about the things we’re going to talk to you about today because I think they’re really an answer to many of those questions.

If you’re a listener, you know I’m familiar with Tom. The man’s on Threads doing customer support every day. And actually, I should not take advantage of this. I hold myself back from asking you for direct customer support all the time.

Anytime. The way that you can tell it’s actually me and not our crackerjack social team is that I often will let a Thread die before I should or fail to follow up with someone. So if we were executing perfectly on that, you would know it was fake Tom, but it’s real Tom, and I do it when I can, but sometimes imperfectly.

At least it’s not an AI avatar Tom. That’s when we know it’s going to go sideways.

[Laughs] That’s right.

I know you’re a Decoder fan, so I’m going to make you sit in this for just one more moment because I find all of this to be really fascinating. The app shipped, and the feedback was immediately and intensely negative. Many, many things were broken. Speakers weren’t working at weddings. This is a time of massive crisis. 

The company goes to resolve it, but then you’re on the board, and it’s clear that there’s going to be a leadership transition, that Patrick needs to go, and we need a new CEO. You’re going to be the interim CEO. 

Tell us about that process. Very, very few people will ever be in that kind of process or get insight into that process. What happened? What were the steps?

Well, I guess you start by saying that the last thing as a board that you want to do is to remove a CEO, particularly one who had as many successes as Patrick had. But I think as we looked at the company’s progress recovering from the app crisis, and frankly the tenor of the conversation inside the company, we just decided that for this kind of step-function improvement we needed on some of these dimensions, we had to make some changes in leadership, and that included a CEO change.

It was a wild experience for me because, I’ll tell you, at no point in the decade leading up to that day in January did I imagine that I might end up the CEO of Sonos. But in the process, the board asked if I’d be willing to step in on an interim basis, and it was an easy yes for me. It’s a cliché, but I’m a customer, and I love this product. There was a Sonos Move in the room when my daughter was born. You talked about weddings. I was watching friends and colleagues on the internet, frustrated by their experience. I wanted to come in and fix it. Of course, as a board member, I felt some real culpability in it. So [I was] thrilled at the chance to come in.

When you’re an interim CEO, the assignment is really to keep the wheels on. The assignment at Sonos specifically was to right the ship. So during the period when the board was doing their comprehensive search for the person that they would ultimately decide was the right choice for the company, it was really easy for me to show up and just do that hard work of trying to get the company back on track. 

While I never imagined that I would land in this seat, I feel so blessed that it happened. I do love this job and continue to love the company and the team, and I’m so excited about the things we’re doing for our customers.

Very few people are going to get an interview to become the CEO of Sonos. I imagine at some point you went from being the interim CEO to wanting to become the permanent CEO. What was that process like? Did you write a memo? Was there a pitch deck? What were the steps?

Yeah, so there were a couple of pitch moments that the board asked me, completely reasonably, to engage in the process of applying for the job in the same way that all of the other talented candidates that they spoke to engaged. And so I took a personality test. 

Wait, you did?

Yeah.

With the letters?

Oh, yeah.

Like an ENTJ or whatever?

Oh, yeah. One of the letter things, sure. They wanted to know what they were getting.

Do they know that that is basically just like astrology? Someone should tell them.

You know what? Yep, yep. I presented my vision for the company more than once, and I had the incredible advantage of being inside the four walls of the company and deeply understanding what was working and what wasn’t, and the things that we needed to do to get to this next chapter. It was a period of intense, intense work for me and everybody around me at the company, and more than a little anxiety about how it was going to turn out.

It was so fun, though, to get the call and then to just immediately be able to pivot at least a fraction of the work from the entirely tactical to the much more strategic, to begin to talk about what kind of company we are going to build together in this next chapter.

When it was announced that you would become the permanent CEO, even when you were the interim CEO, my reaction was, “Oh, this makes sense.” To use a football analogy, football teams hire defensive head coaches and offensive head coaches, and they come from one side of the ball or the other. It was just obvious: “Oh, this company needs a software CEO, and Tom is a software guy.” 

You’re describing a situation where obviously it’s both. You’ve got hardware and software together, but that was the light that went off in my head — “Oh, this is a software person coming in to fix a big software problem.” Is that how you saw it initially?

Yeah, my career started with selling personal computers in 1986 or something, at the original Micro Center. If you’ve ever been to a Micro Center, it’s one of the few remaining big-box retailers for computers. So at the original Micro Center in Columbus, Ohio, at 16 years old, I walked into the CEO’s office and said, “I want to work here.” So from my earliest days, I was sitting with early adopters of personal computers and watching their struggles to understand the technology and put it to use in their lives.

I became an engineer and a product designer out of a desire to help people learn to use these products. And that’s really what’s motivated me through my entire career: how do I help teams make great experiences? So while it’s true that I’m a software person, 35 years in the business, my identity really is as a product designer. And I’d like to think that the board imagined that, yes, they needed a software person, but they needed somebody who could holistically look at the opportunity across hardware and software and help the team get there.

You appear to be through it now. There’s a big stack of announcements here, including MCP integration. I’ve got a lot of questions about all that. 

When you became the permanent CEO 18 months ago, you said you immediately shifted strategy. What brought that into focus for you? “Oh, this is actually a software platform with hardware components”? Was that instinct? Was it research? Was it, “I’m getting through this app process, and that is becoming revealed to me”? How did that work?

I think by disposition I’m a fairly right-brained person, and I have this intuitive feeling — creative, or whatever the words are, not as quantitative. But my first job out of college was at Apple, and I think even from the early days when I was there, we talked about the intersection of the arts and humanities and the sciences. 

I think one lesson of my career has been that you really do have an obligation to bring both a quant and a qual assessment to most things that pass your desk. So the development of our strategy was very much like that. Yes, there was an intuitive sense about what the right answer for Sonos was that was built up from my experience as a customer by observing our customers and talking to them through the years. A piece of that is a perspective about where the puck is going, of course.

But I do have access to piles and piles and piles of data about how our products behave in the real world, how our customers interact with them, what they use, what they don’t use, and lots of people around me in the organization who can bring that quantitative framing. I think one advantage I had over other CEO candidates was that I could bring some of that horsepower to bear on the story that I shared with the board about my vision for the future of the company.

You’ve made some changes. It’s time for the Decoder questions. You have restructured the company. There have been a few rounds of layoffs. How big is Sonos today?

We’re about 1,500 people.

And how is that structured?

So the company that I stepped into was divided into business units. We had a team that was focused on home. We had a team that was focused on the professional opportunity, a team that was focused on home theater, and so on. While there’s something to be said for P&L responsibilities, et cetera, et cetera, et cetera, there’s something to be said for that approach at a certain scale. I think that my appraisal was that we were sub-scale for the inefficiencies that creep in. We have the almost guaranteed duplication of effort across the silos.

What’s more, if you imagine that the product is really the system, the idea that your portables team sits far away from your home theater team, and sits far away from your plugged-in speakers team, is almost surely going to run counter to the kind of progress that you’re trying to motivate. And so very quickly, within three weeks of my arrival, we reorganized the company into a functional model.

So my leadership team today, of course, has a finance organization, a legal organization, a sales organization, and a marketing organization. But the product team now shows up as represented by hardware, software, and global supply chain operations, so leaders across each of those things. And what that allows us to do is to assemble cross-functional teams, or matrix-style teams, to invest in the initiatives that are of the highest priority to our customers and to the business at any given moment. And we can flex that up and down and side to side as our needs change.

We might, for an interval, be very focused on this part of the portfolio over here, and then we might be able to move to focusing on a different corner of the portfolio. No one has to get a new boss, and people can fluidly move from these things and get new opportunities to learn. It just strikes me as a much more appropriate organizational structure for our current stage.

You’re describing the shift from a divisional structure to a vastly more functional structure.

Yes.

That’s Apple, right?

Yep, sure.

Everyone comes up, and there’s one big product meeting.

You obviously make a lot of software. I’ve described you as a software CEO. It seems like the software industry is in a moment of existential crisis about how software is made, about what a product manager versus an engineer versus a designer do. Are you still in the classic structures of building software? Or has AI completely upended what you’re doing?

No, we’re as deep in this as anyone. What’s been really fun for me over the course of the last year is to watch the software engineering team go from thoughtful skepticism to pretty deep embrace, if I’m honest. Some of that is on the back of just how dramatically the tools have evolved over the course of the last 12 months, certainly. 

I think almost everyone in the industry went home at the holidays at the tail end of 2025, coming into 2026, and came back just gobsmacked by how much progress the frontier labs were making on agentic coding, and it’s been a real accelerant for us. As complex as this transition is, I’m really excited about being able to help our engineers navigate what’s a pretty fundamental sea change in the way that this kind of infrastructure gets built, and it’s fun to get to be a part of that at this moment.

Did you break up the classic PM, designer, and engineer triad? Or is that still the core building block?

We still have all three, but it blurs particularly as the designers and the product managers can be prototypers at a really high level with incredible efficiency. It’s exciting to see how quickly they can communicate their ideas now in a medium that just didn’t exist at all for them a year ago.

We’re at the turn, right? You came in, you restructured the company, you identified a new thesis for the company. Today you’re here to talk about all the new news. Do you think you’re through it? Do you think the existing customers have forgiven you for the app mess? Do you think the brand has been restored?

Yeah, as we’ve been saying, I think that there’s the quant and there’s the qual. The quant is green across the board. Performance, reliability, customer care interactions, social tone, NPS — these things are all restored to levels that we enjoyed before any of that happened. 

On the qualitative side, I love the fact that I can show up on Reddit or on social media broadly and engage with our customers and have a constructive conversation about things that they’re excited that we’re doing, and things where we’re still disappointing an individual customer on some dimension.

I think that at some level, maybe this game is never over; you’re just always trying to polish off the remaining rough edges of everything that you do. So I have a real critic’s eye when I look at the work that we do, and if I ever let that slip, a customer will remind me of where the work remains. And so, yes, I think in every measurable way it’s behind us, but it’s a “never forget” chapter for the company.

That brings me to the other Decoder question: how do you make decisions? What’s your framework?

Well, I don’t think you bring just one framework to every decision. So we talk about, for example, product decisions. As an idea moves through its lifecycle from its days as an early little sapling that you have to care and tend to, through final mass production for a hardware object, and out into the channels, I think how you think about [each of those] decisions is different.

So in the earliest days, I really tried to protect the creativity of the company and to not let the antibodies of doubt and critique erode an idea before it’s had a chance to express itself in a way that anybody could get excited about. That means creating room for risk-taking, creativity, qualitative assessment, and prototyping before you get to the details of the business case and the hard trade-offs that you make on a journey from an idea to something you can actually ship.

But of course, as the programs roll along and they take on more and more financial investment and organizational risk, as more and more people aggregate into the product, you become, I think, just naturally much more quantitative in your analysis of the decisions. And you ask the teams to bring a really thoughtful, detailed description of the puts and takes and the risks associated with it, and you use that to help you make your decisions.

But maybe on top of all of that, you just have to hire great people. I mean, you just do, and you have to put them on teams that are an appropriate size for the task at hand. I think it’s going to be very, very difficult for any CEO or any leadership team to make great decisions if the people who aren’t bringing the ideas themselves aren’t great. So I suppose it starts with the right people in the seats.

The reason I ask that question in that way is that, I think, the decision you made that is most emblematic to me of changing from a divisional structure with individual P&Ls to a more functional structure, and thinking about the platform as a whole, was the decision to kill a product that I believe was called Pinewood. It was a set-top box that was going to integrate all the streaming services. This was well reported by former Verge reporter Chris Welch, who’s now at Bloomberg.

The thing that really struck me about that was if you had a division that was focused on soundbars, the only place that division could go was to compete with Roku. And all the reporting that we had before suggested that the soundbar division was growing because the attach rate for soundbars with new TVs is 100 percent. Everyone knows the speakers in their TVs are bad. 

It’s a great business, and we’re for sure the best product in the category.

So if you want to grow, which is the big thing, you’re going to be like, “How do we grow? And you say, “Well, I got to beat Roku.” And then now you’re just off doing that. You came in, and you just ended that, and I thought, “Oh, this is not just a decision to end a product line; this is also a decision about the strategy of the company. Because if you’re going to go compete with Roku, you’ve got to be a different company.” That’s my view of it. Is that how you saw it?

Like so many things, these decisions do have more than one dimension that you evaluate. A big part of that decision, candidly, was that we were spread too thin across too many things, and I didn’t believe that we could be exceptional at all of them. I wanted to return us to focus on the thing that allows us to be most differentiated in the competitive set, and that’s the incredible systemness of the product, the audio experience that we can deliver, and attending to what, at the time, was a very long list of things that our customers were explicitly asking us for.

I think if we were, putatively, actually working on a set-top box at that moment, another thing that I might have wondered about was whether our desire to make this particular product came from our customer or did it come from some sort of internally sourced sense of business opportunity? And I think companies get in a lot of trouble when the thing that comes first is the model and the thing that comes second is the product.

I certainly feel a sense of obligation to constantly be evaluating new business opportunities for the company, new business models for the company, sure. But I want them to always start from a place of what we really see in the market from customers. I’m not sure that they were saying we want a high-end streaming box that is somehow the love child of an Apple TV and a Roku.

Although that love child… I would be very worried about that love child. Why does this product have 200GB of storage for no identifiable reason, but also shows me ads at all times? We’ll set that aside. 

I ask that because I want to talk to you about all the news here. The news here is a bunch of big bets. You’ve got Sonos 27. That is the new name of what you were calling an audio operating system that has something in it called Sonos Fabric. 

Boy, I know you know me. I’m ready to go on all these names and all these ideas. In there, somewhere, is MCP support. There’s a big idea about assistants that I want to talk about. You’re cashing some chips, right?

Yeah.

You’re saying, ”Okay, here’s a big new idea. Trust us. We’re going to get it right this time.” You’re ready for it.

Yeah, I feel very ready. We’ve spent the last year building the new products that tell the story of this next chapter of the company. We’re really excited, not just about their capabilities, but how we’re going to share them with the world and how we’re going to enter a period of co-creation with our customers. 

We’ve recently done that with some of the app enhancements that we’ve been rolling out. We’ve rolled out a mechanism that allows our customers to opt into them explicitly and talk to us about what they see in their own homes, and it’s just been such a delightful chapter of engagement with the community that I think we really lacked for a while. So yeah, I feel very ready.

Can I talk about it?

Yeah, let’s get into it. Sonos 27 and Sonos Fabric, what are these things? Explain to people what’s going on here.

So 100 percent hardware, 100 percent software. For 20 years, we’ve given all of our hardware objects names: Roam, Arc, Beam, Ace, Era. But we’ve never named the software platform — the thing that takes these collections of products and creates a consistent whole. So in some ways, the big news that we’re sharing right now is that the software platform has a name called Sonos 27, and it has a bunch of promises that it makes to customers in its nuts and bolts.

It’s an ecosystem platform that allows a huge heterogeneous world of services to come together in your home. It’s a system that has a deep technological layer that tunes the audio for every part of your household through a system we call Trueplay. It’s fundamentally built on a kind of fabric that brings all of the products together into a cohesive whole. And architecturally, it’s one of the things that’s really, really unique about what we’re doing.

Our products in your home really find each other on the local network and create something that is truly where the whole is greater than the sum of the parts. And it’s what unlocks so much of the magical experiences that we’ve delivered through the years. We’re about to announce some exciting new capabilities for Sonos Fabric and point a vector into the future of more to come.

Finally, we have these control and engagement services: app, voice, web. You mentioned MCP, which is a new dimension that we’re adding. We also have an entirely new voice assistant experience. And so we’re doing all of that across the software ecosystem, and we’re even giving a little bit of a preview of a capability that we’re calling Custom Agents, which is multi-tenancy inside of the platform to allow modern voice assistants to live side by side and interact with one another on the platform.

So we’re doing all of that, and we have some great new hardware too. We’ve got a brand new amazing soundbar and an amazing set of new headphones.

The Beam Ultra is the new soundbar, and I think it’s 7.1.2.

Yep.

This is very nerdy, but to me it represents the fact that Atmos is totally mainstream now at this price point.

Yeah, totally.

That might be an entirely different home theater podcast. 

I can’t wait for you to hear it, honestly. Obviously, I’m biased and consider the source, but it sounds amazing. People who have heard it just can’t stop talking about how great this little soundbar sounds.

Honestly, I should quit doing this and just do home theater podcasts. Where does my heart truly lie? It’s that. It’s just doing that.

[Laughs] I’m down. Anytime.

We should do it.

If you need a co-host, we should do it.

Look, I just judged the TV shootout, and Sony lost. It’s a big deal. There’s a lot going on in that world. 

Tell me about the idea of naming the operating system Sonos 27. It’s obviously the middle of 2026. It obviously tracks Apple’s naming scheme. This very much implies there’ll be a Sonos 28 one year from now, in August of 2027. 

It does. I think from the beginning we knew that we wanted to give it a name, and we knew that we wanted the name to be able to express this idea that Sonos does evolve. And for many, many years, that evolution was to the great benefit of our customers. We had lossless audio for Apple. We had high-fidelity audio for Spotify. We had AirPlay. We had Alexa. 

Through the years, we’ve just over and over and over again expanded the capability of the products that were already in our customers’ homes, and we didn’t have a way of talking about it at all. So we wanted to give it a name. We wanted people to understand that it improved over time.

So as we started thinking about, “Well, how do you really do that?” I mean, the first thing they came to was like, “Gosh, we already have ‘OS’ right in the name of the company.” So to try to brand some entirely new thing felt like a self-own, and so we pretty quickly reverted to this idea that the product is 100 percent hardware and 100 percent software. The product is Sonos, not Arc. Maybe the OS2 is called Sonos.

And then it’s like, “Well, what companion moniker can we give it to drive that idea home?” Obviously we talked about version numbers and any number of other things, but we decided that we’d like to take advantage of this. Once a year, we’ll stand up in front of our audience and tell them about how the platform is evolving, and using the year as a descriptor for that was a better mechanism than almost anything else that was available to us.

Those software cycles, yearly big releases… Apple is basically the only company that sticks to it. Even if it doesn’t do it, it announces all the features, and then they slowly dribble them out over the course of the year as they come into focus. 

Again, I have a lot of questions about the nature of software development in 2027, but the idea that there’s one monolithic release once a year — the industry’s gotten away from that. Is that your idea, or are you going to do something different?

No, no, no. Fortunately, the industry has gotten away from that, and our intention is to have a moment where we can tell the story of the operating system roadmap for the next six or 10 months. As you’ll see, even in this moment, the things that we’re announcing — many of them will be available in what we’re calling early access, which is the chance for us to put them out into the product as opt-ins for our customers. That way, we can do the perspective collecting, real-world data collecting sort of work. All of these products will continue to evolve through the year.

Frankly, I think as we go into the event, we will talk about all of this. I know that I’m going to get questions from customers and observers like, “What about this feature?” I think really the answer is just going to be like, “We know the stuff.” There’s very little that you can say, “Hey, someone should do this,” that isn’t on our list.

So to be able to say, “Look, part of what we’re communicating with the 27 of it all is that this stuff will come in 2027, and if it doesn’t, it’ll come in 2028 or beyond.”

One of the big ideas here is obviously that this is an operating system that addresses a bunch of hardware in your home. That thing mostly looks like speakers, but some of it is connected to your TVs. You’ve got command and control. You’ve got other assistants, which I want to talk about, and that operating system should gain more capabilities over time. 

What does the far version of that look like? Is it just that there will be an ambient assistant in your house that is addressing the fabric through MCP? Is it that you’ll interoperate with other speakers that might be in the house? What does the biggest vision of this look like?

I mean, I think in some ways you’ll have to stay tuned to see where we take this, but I’ll start with this: I think that every home deserves this infrastructure. I think sound is an incredible mood stabilizer, if you like. Take the audio track out of your favorite movie or television show, and so much of the life comes out of it. And I think having an out-loud audio experience in your home is something that everybody should enjoy. So part of the big picture is we want to spread this far and wide and make it accessible to everyone.

A running joke about the first generation of assistants is that they’ve ended up being for timers. I think you really do, when you’re building out this next generation of voice-centric experiences, have to look at not just what the technology can do, but also what the customer wants. And one of the new capabilities in Sonos 27 is our new voice assistant.

Our voice assistant is built on modern generative technology, but it’s really inspired by what people tell us that they like to see in the home. I’ll tell you, the killer app of the assistant is that they just want to be able to talk to it like a normal person, full stop. With the first-generation assistants, you just had to guess what to say every time, and even then maybe it took two or three attempts to get there. 

It’s such a game changer to be able to say, “Play that album I loved in college, the one with the Cocteau Twins singer. I think the song ended up as the theme song for that show. It was on Fox. It was about the curmudgeonly Doctor House… It’s ‘Teardrop’ by Massive Attack.”

I know. They took it off the [streaming version of House]. It’s sad.

It is. It is. So that’s job one, if you like. The first-generation assistants didn’t do it at all. Sonos Voice in the new OS is exceptional at this. So that’s meeting the customer where they are.

What else should they say? Well, they say a little bit about what I would call mood and ambiance. So lighting. Let’s bring lighting into the environment so that you can express a scene through both sound and light. 

My wife and I are constantly settling little bets in the kitchen: “What was this thing called? Who is that actor?” They do want comprehensive world knowledge. They want to get an answer to the kid’s homework assignment or whatever, and so these agents do need to be able to hand off to comprehensive frontier models to get perspectives on the world.

In some instances, those perspectives allow them to use the frontier model to manipulate the audio environment to play some music that people are talking about on TikTok and pick some stuff that sounds like the shoegaze music that I liked when I was in college. We can leverage the power of a frontier model to understand what’s being talked about on TikTok to sort that against my musical preferences, and deliver it.

I guess while we’re enumerating what I think a modern voice assistant has to do, it has to be fast, it has to be reliable, and it has to be responsible and affordable to run. I do think that customers want this as a fundamental capability of a product like Sonos, and we built an architecture that allows us to deliver it to every one of our customers as well.

There’s a lot of big Decoder, big Verge ideas in all of there, all the way down to where does the compute live? Where is the computer in your house supposed to be? You’re describing a lot of command and control. There’s integration with smart home stuff. I think it’s a Lutron partnership you have now.

That’s right.

But there’s an infinite number of smart light switches that might help you control your lighting. Your company is part of the Matter standard, but you don’t actually enable Matter on Sonos speakers. I think our reporting from the wonderful Jenn Tuohy, as our smart home reporter, has suggested that you don’t think it’s ready. 

This is a lot of stacked problems. I’m just going to read your press release to you. It says, “Sonos 27 MCP allows you to connect any outside AI assistant or agent to your system, including ChatGPT.”

Well, I’m a customer, and I’ll tell you that the one that I want to use is Gemini for Home, Google’s assistant, and that’s been broken on Sonos for ages because of an ongoing lawsuit. Is that fixed? Is Google on your list? Because there’s a lot of competitive dimension here.

So in many ways we have a strong relationship with Google to this day, despite the lawsuit. YouTube Music is one of the most-listened-to music services on the platform. We do have a long history of a first-generation Google Assistant on many of our products. We’d love to continue to work with them on all of these things, and those conversations continue.

I guess I’m just going to push on this idea of connecting any AI assistant or agent. There are two big ones. There’s Alexa, and there’s Google. You might be in third place. Siri exists. Those are the ones, right? Maybe there’s going to be a ChatGPT product in the future.

Well, listen. I want to tease apart two different things.

Sure.

Let’s talk about command and control of the Sonos system through external agentic interfaces, just to use every buzzword. What I really mean is that we think there’s a lot of utility in showing up inside of ChatGPT on the web, on your phone, on the desktop in the same way that Slack or your calendar shows up in those environments.

So with Sonos 27 MCP, we’ll show up in the directory of connectable services in all of those environments: in Gemini, in Claude, in ChatGPT, in Grok. By connecting that tool into your experience, however you access it, you’ll be able to summon Sonos. So maybe you’re working on something, and you want some music to play; you can just tell Claude to make it play or just tell Gemini to make it play from the desktop or mobile surface that you’re sitting in front of.

That’s a pretty powerful building block as we see these OpenClaw-style agentic, proactive, asynchronous agents evolve. Now they have access to Sonos in a supportable way. There’s been a bunch of hack MCPs and CLIs for Sonos around, and a lot of reporting about how they’re among the most utilized examples for these things. We’re excited to officially support such a thing. As much as fun it is to see the mainstream use cases, I’m very excited to see what hobbyists are going to do with Sonos MCP.

So let’s now talk about on-device, though. So we’ve been working closely with Amazon for some time to bring Alexa Plus to our platform. That work continues. I’m excited about it. We don’t have anything to announce today, but we’ve had a long and really successful relationship with Amazon through the years. Panos was on the board with me for many of our years at Sonos.

By the way, just so people know, that’s Panos Panay, who now runs hardware and Alexa at Amazon.

That’s right, that’s right. He’s a friend of the show.

Yes. Panos and I are buds. 

But one thing we’ve demonstrated is something we call Sonos Custom Agents, which lets you take any of these inference engines — the Anthropic engine, the OpenAI engine, the Gemini engine, the Grok engine — and invoke it from Sonos. You’ll be able to give it a unique voice, give it a name, and give it a context that has it show up in a particular way.

So imagine that you’d like to have a personality that lives inside of Sonos, and it’s the John Cusack character from High Fidelity, an opinionated record store geek who is interested in subtly shading your perspectives in music, but also making recommendations. 

You’ll be able to go in and set up a custom agent in Sonos that says, “Let’s use OpenAI inference. Let’s use this voice I pick, and let’s give it this personality, and let’s have it live alongside an agent that’s powered by Anthropic inference that is going to give me advice on cooking. And let’s do another one that is only interested in connecting me with opportunities to see live music.” 

All of these tools will be able to show up inside of Sonos, invoked by the name you give it, like, “Hey, Sonos, connect me to the curator,” and talk to you only about that topic, have a point of view, and be able to access the Sonos context so that they can hand off from one to the other.

It’s super fun. It’s super creative. It lets our customers become participants in the Sonos platform. We’re just showing it off as a technology preview because there are still a lot of details to work out here, but it’s a lot of fun to play with in the lab.

Let me bring some of these ideas together. There’s the idea that there’s external command and control. I have an Echo Dot, and I say, “Turn on the Sonos,” and it just does that.

Then there’s the idea that there’s the assistant that lives on your device that is either the Sonos voice assistant or maybe you’re going to keep working with Amazon and Alexa+ being there. And then we’re going to make new custom ones out of building blocks that are enabled in lots of different ways.

When I asked where the computer is, this is, to me, the big question: where does all this ambient agentic AI actually live? Where does it have control? Is it going to happen locally? Are we sending everything to the cloud? 

Somewhere in your home, if you’re going to control lights, music, and everything else, there needs to be some logic happening in your home. There are a lot of standards to do that. Some of them are good, some of them are bad. Some of them have names that connect directly to the word fabric, like Thread.

Sure.

There’s a lot going on here in terms of just the names and schemes of what could happen. Where do you see this view? Who is your competition here? 

Is it that Apple is going to launch a new HomePod with an AI-enabled Siri, and they already have a HomeKit, which is a pretty well-built and pretty popular smart home platform? Is it, “Well, OpenAI is going to have a hardware device, and maybe their first one’s going to be in the home, and you need to stave that off and make sure people are talking directly to you, and you have the customer relationship”?

That seems not only undeveloped, but even the foundations of how you build the products — like Matter and smart home standards — are messy and political in very specific ways. So how do you see the competition, victory, and success there? And then how do you solve just the capability gap that every other company has run into?

So one thing that I’m reminded of in this moment is that we, as an industry, have these platform shifts that unlock opportunities for companies who get lucky on timing. For example, the original dot-com era was really powered by Oracle and Sun Solaris boxes and data centers that you owned and operated yourself. We then went through a chapter where the infrastructure became Linux and open source for your data tier. 

Of course, we’ve had this huge long run of cloud infrastructure. I think that if you’re a company that builds your stack on Solaris and Oracle, it’s really that there’s an opportunity for a company to come along and build a stack on Linux and Postgres in that moment and suddenly find efficiencies that are very, very hard for you to claim.

I think something very similar is happening in voice assistants. When I hear you talk on the show about your experience with voice assistants that are going through the transition from first generation to second generation, it feels to me like so much of the challenge there is about how we drag along all of the committed relationships that we have for the internet-connected toaster, back massager, and all of the things that are connected into those existing historical ecosystems. 

Honestly, even in the world of AI, with just how much has changed in the last couple of years underneath the companies as they’re building out their answers here, it really feels to me like we are stepping into a moment where agentic tool use powered by protocols, like MCP, is going to disintermediate a lot of that stuff that you’re talking about.

I think that in the immediate term, we’re really excited about what we can do in the use cases that our customers are telling us they care about. Give me mood and audio in the home. We have great solutions for that today. And I think we can take advantage of this shift to a truly next-generation architecture without the baggage of a decade of internet-connected light switches from 17 different protocols with 17 different hubs in the closet.

Well, I mean, part of the opportunity there is that you don’t have a big voice assistant. I don’t think the current Sonos voice assistant is at the scale of Alexa or Google Assistant. I’m just guessing. They do have that problem. The light switches in my house, they’re in the walls. They’re going to be there. That Lutron hub is there. It’s not great. However, Lutron is the best one. 

It is. It is the best one.

There’s a box in my house, and there are a lot of those little weird little boxes in my house because I have a smart home, and all of those dependencies exist for those platforms for reasons. If you show up and say, “Use my voice assistant instead,” and it can’t operate all of the lights, well, I’m just going to fall back to Amazon, or I’m going to fall back to Google. They know it, and they know they can’t break those things. Panos and I have talked about it a lot, actually in the context of Alexa Plus. 

The promise of AI, at least from my very abstract vantage point, is, boy, I should not care about any of this. There’s a network in my house. There are lots of devices on that network. They have some access points on that network, and maybe they’re API-driven access points, maybe they’re MCP-driven. It should not matter. There’s just a bunch of devices in my network that you can control somehow, and this all-seeing robot that is going to replace every job in America should be able to traverse my network and turn on the lights without any help. That’s what it should do.

Some of the very first big vibe coding shockwaves were AI researchers, bleeding-edge AI researchers, being like, “I figured out how to control my shades. They walked through my network, and the shades go up and down now,” right? We’re not there at the consumer level. The promise is not there. I’m still here talking to you about Matter, Thread, HomeKit, and whether or not Google Assistant can control my Sonos Amp, which, by the way, it can’t, and that is my one ticket to you that I didn’t submit on Threads. 

That’s the promise, but the gap to that seems really, really huge. Are you able to close it? Is that the opportunity?

I think we’re just still in the Apple II days of the home automation space. So while there is, as you point out, this incredible, wild, heterogeneous world of different light switch controllers, I think when you talk to mainstream consumers about their homes, they’re not there yet. 

I do think that because of the profound technological step function of agentic control of these kinds of services and things like tool use, as the consumer catches up with the industry’s ambition to solve these problems, new technological solutions will emerge. I’m really excited about how we’re positioned to ride that transition and not be encumbered by the decade of early adopter internet-connected light switches.

What are you going to do about the light switches that are already in the walls?

I just don’t think that there’s enough of them to matter in the final answer. Of course, Lutron’s a great partner, and they do a great job with this.

They do. I mean, we have them. We’ve got all kinds of light switches. We mostly have Lutron ones. But that’s my question. The first natural customer for something like what you’re describing is somebody who is already using these products in some way. If you show up to my house and there’s no Matter support or whatever, I’m like, “Well, I can’t use it.”

Is it though, or is it more like the difference between the people that were adopting Creative Labs MP3 players before the iPod came along?

Sure.

Was the iPod really for those people, or was it for people that had that problem but were nowhere near the people that had been early adopters? Or was the iPhone for people that were using Windows CE, or was it for people that were using flip phones who were ready for the transition? I think we’re making-

Is that the idea here “iPhone for smart home control” or “iPhone for audio in the home”?

That’s right.

The turn there that I think is really interesting is: who are you going to talk to? Where is the computer? What is the computer? You’re describing a situation where if you want music, you’re going to talk to Sonos. But you also have a general-purpose assistant like Claude or ChatGPT, and they can talk to Sonos, too.

It’s unclear to me — and it may be one of the most fascinating questions at this moment in tech — where the interface goes. Are we all going to have one chatbot that does everything, and all the demand is aggregated there, and OpenAI becomes more powerful than ever? Or are we going to have lots and lots of assistants that we address in lots and lots of different ways? 

It seems like your bet is Sonos will have a personality in your home, maybe multiple personalities in your home, and customers will address that, and all the other stuff will go somewhere else.

For 20 years, we’ve found a lot of success around the idea of embracing every dimension of the problem. Yes, there’s an app, but you’re welcome to engage with us through AirPlay or Spotify Connect. Yes, we have a voice assistant that lives side by side with Alexa. Yes, we have proprietary mesh, but it integrates with Bluetooth and even line-in. I think we bet that we’re of a size and scale that allows us to both be nimble and react to these changes, but also have enough horsepower to keep up technologically with what the best in the category are doing. So I think some of this just plays out on the field.

Just one last idea, since you said it a couple of times, this question about where the compute lives. I think one incredibly powerful opportunity for Sonos is that we do place compute in millions and millions of homes. People don’t think of them as computers often, but they are. They create a seamless experience that is really the integration of a set of cloud services and local services in a way that looks like just one surface area to the customer.

I think that the answer to your questions, “Does this live in the cloud? Does this live in the home? Does this live in a box?” is yes. I think we’re pretty uniquely situated to take advantage of an architecture where it’s not cloud-only, it’s not home-only, but it’s an integrated system that combines all of that together. And at the end of the day, that is what Sonos 27 is.

Just to make that really reductive, when I say “where is the computer” in the home context, what I mean is maybe in the ’90s to early 2000s, there was a vision that we would all have home servers in our basements, and they would do all the logic of our home. They would have all the storage, and Microsoft would sell you a Windows home server in the basement and a Windows Media Center in the living room, and you and I probably both watched all of those Bill Gates keynotes at CES. 

None of that really ever came to pass. Instead, we put a lot of cheap computers near all the TVs, and they connected to cloud services, and they’re basically just running web browsers, as you well know. That is a very different version of what needed to happen and how the market played out.

In the smart home, I’ve got a bunch of automations in my house. When the garage is open, the lights turn on, and other stuff happens. At night, the doors all lock if they’re not locked. That logic has to live somewhere. My perspective on it is that it should live in my house. If the internet goes down, the lights should still turn on. 

This is not anything to do with big AI integrations or the new MCP way to address the Sonos system. Those are big ideas, and I’m just insistently dragging you back down into the dirt of: if the internet goes out, are the lights still going to turn on?

You don’t want to be the partner that puts in the internet-connected light switches that don’t work when the internet is down. 

But it’s not just the internet-connected light switches; it’s the logic of making it all work together.

I understand. Yeah, fair. Here’s what I’ll say about that. For 40 years now, we’ve been talking about technology for the home, starting with the personal computer, which is supposed to be the center of the technological household, but we didn’t really get that. We got laptops that people mostly use at work and use a bit at home.

Then we got mobile, and mobile is supposed to be the thing that you engage with when you leave your house on your way to work. Of course, the fundamental media of mobile is so addicting that it’s all-consuming. We use it at work, we use it at home — we use it all the time. But my observation about our use of it in the home is that nobody loves it. No spouse has ever said, “I wish my husband was on his phone more. I wish my kid used their iPad more,” and so we’re looking for ways to get beyond that.

To your point, the other compute that’s in the home, like I said, is the set-top box, which is largely just a web browser to a bunch of services that live in the cloud. I think it’s pretty interesting, actually, that in 17 million homes Sonos is an actual mesh of compute in your house with real horsepower, real memory, real CPU, real capability, a real operating system — all just waiting to be the container for some of the kinds of things you’re talking about. So it’s super interesting. We’ll figure it out over time.

I feel like you’re this close to being like, “And then the Teslas will be a data center.” There’s something there that is maybe the line. Actually, I want to talk about that. I actually want to push on that a little bit. 

The initial Sonos innovation 100 years ago, when the company started, was, “Oh, we can build a mesh network. We can do wireless audio at scale. We can synchronize the latency of all the speakers so it sounds like the same music playing at the same time.” This was a lot of big engineering back in the day. 

Still, I mean, candidly, still. A lot of what our software team does day in and day out is extend and maintain that infrastructure.

There are features you’re announcing here as part of Sonos Fabric that I think are interesting. If you have surrounds, you can just place them, and they’ll figure out that they’re surrounds and immediately start being the surround speakers in a home theater. 

There’s just a lot of potential, as you’re describing, in the notion that, “Oh, there’s a big network of small computers in my house that can come together and do things you would not expect.” Is your focus that they will always do audio things? Or is it the focus that they will have many more capabilities?

I think any thoughtful product leader is going to tell you the focus is an important part of success. I do feel like there’s so much interesting work for us to do around creating an audio system that sets the mood for your home that I don’t want us to take our eye off of that opportunity. We are doing some really exciting things in Sonos Fabric and Sonos 27 that we should talk about, that get to some of the work that you’re describing.

But certainly in the moments where you allow yourself to think about what the company is in five years and what it will be in 10 years, you can’t help but interrogate what the opportunity is for a mesh — a fabric of compute that lives in the home.

I want to end by talking about headphones really quickly. We just had the CEO of Bose on. She and I had a pretty searching conversation about what happens to the headphone market in the time of AI wearables. 

My fear, my anxiety, about audio is always that consumers will always pick convenience over quality. It’s just the history of the… I’ve got a set of AirPods on this desk, and they don’t sound great, but boy, are they convenient. And that’s just the way it goes. It has always been what the market has demanded: convenience over quality.

We’re going to enter a time when Apple puts cameras on AirPods, and you can talk to them with Siri. When Google and Samsung ship very similar products, there’s undoubtedly going to be a flood of AI hardware startups that want to put something in your ears or in your body in some way. 

You are in the headphone market. You’re talking a lot about AI; you’re talking a lot about Fabric. That seems like a cataclysm coming for headphones. The same way that the removal of the headphone jack was a cataclysm in the headphone market, this seems like the next one. 

How are you preparing for that? Do you think there’s going to be a range of voice assistants on headphones, or do you think what you would call the first-party vendors are going to just grab more and more of that share?

Before we jump to the hypotheticals about where headphones are, can I tell you a little bit about the new headphones?

Yeah, of course.

All right. So one of the things that you get from entering a new market is the opportunity to build something and put it in customers’ homes and learn a lot about what matters and what doesn’t matter.

So we launched Sonos Ace two years ago. I think measured against any of the other premium over-the-ear noise-canceling headphones in the market, they’re a great product. They’ve got great reviews; they have great noise cancelling. They’re super comfortable, and they look great. We’re really proud of the product, but they weren’t thoughtfully integrated into the Sonos ecosystem. 

I mean, at launch you could swap to one of our soundbars, not even all of our soundbars. I think it was the most visible expression of the missed opportunity of that moment for the company to always remember that the products that we make are part of the Sonos system, not individual offerings.

So we ended up with a great pair of headphones, but honestly, I mean, you review them all the time; there are lots of great headphones. When we set out to take all the lessons that we learned from Ace to design Ace Ultra, at the very top of that list is that we have to do the hard thing. These headphones have to be able to show up in your home as a part of the system, and the model through which you invoke your headphones has to be completely seamless and delightful.

We built an entirely new hardware platform for Ace Ultra called Headphone Engine 2, and it unlocks this capability at the hardware level. And then in Sonos OS 27, we built virtualization so that any one of the speakers in your home can deliver the audio platform to that speaker but can also act as a proxy for your headphones, which don’t have the horsepower to run the full stack — it would kill battery life, et cetera.

Now with Ace Ultra, you’ll be able to link your headphones to your Sonos system, walk up to a speaker that’s playing, press a button on your headphones, and the music will jump from the speaker into your headphones. No Bluetooth, no phone. You’ll be able to walk around your whole house, enjoy the music, come to another speaker, press the button again, and it will jump to that speaker using some BLA location-sensing stuff that we’ve developed. It’s super magical. It’s the one job that we had and didn’t deliver on with Ace. We’re really excited to get it out into the market.

Analogous to that on the home theater side is, as you mentioned, that we’re making all of our portable speakers — Sonos Play, Sonos Move — capable of being rear channels for home theater. That sounds really easy. Maybe people will even be like, “Why wasn’t that always possible?” But to your point about the deep technology of it, the way that we achieve high-fidelity, low-latency surround experiences in the living room is by directly bonding our rear channel speakers and subwoofer to the soundbar on a proprietary network, not through your access point. So for a speaker to seamlessly hop between being part of your Sonos system as a music node and part of your Sonos system as a surround channel means that you have to do a real-time reconfiguration of that networking stack, and you have to know when to do it.

We’re using accelerometers in the product to actually know that you’ve picked it up and set it down in a new location. We’re using Sonos’ positioning system, which is ultrasonic signaling, to broadcast to the room, “Hey, is there a soundbar here?” Ultrasonic comes back saying, “I’m here.” They negotiate position, left and right, and distance, so that they can optimize the surround sound experience. You pick it up again, you take it back to the kitchen, and it’s a regular speaker again. All happens magically, no use of the app. It’s pretty cool.

Answer my question on headphones, though. This is my big worry. You’ve got really cool headphones. They have a feature that no one else can ship, and Apple’s going to say, “Look, our new AirPods have Siri AI in them.” 

You can just say, “Show me what I’m looking at,” and a picture will be taken automatically and something will happen. We’ll get over the privacy implications of that, but that’s a collapse of the headphone market.

Maybe. I mean, headphones come in all kinds of form factors today: over-ear, in-ear, cuffs, wireless, wired. So I do think it’s a market where people choose different form factors for different tasks. We’re really focused on this personal in-the-home listening — what do you do when the baby’s asleep? What do you do when you want to move the party from the kitchen to the den while you relax for the evening? We think that there’s a bunch of really interesting use cases there that other people aren’t paying attention to.

I think, broadly, over the course of the next couple of years, we’re going to get all kinds of adventurous wackadoodle consumer electronics that are connected to this new emerging AI universe. Most of them will be a swing and a miss. Of course, we’ve already seen some of those. So it’s a little hard to speculate, I think, about which of these things, if any, is going to have its iPod moment or its iPhone moment or whatever.

What do you think is going to happen?

I think we’re going to see a replay of what happened when the headphone jack got taken away. The headphone jack was an open interconnect. There was a really vibrant market for all kinds of headphones, and a bunch of those companies don’t exist anymore. The high-end wired headphones, for phones anyway, have mostly gone away, and you have a bunch of medium Bluetooth competitors, but the phone makers have largely taken that with first-party products. 

Now, you have your products at the high end competing with really high-end noise-canceling products from Bose and Sony and whoever, and that is the airplane set. The lack of the open interconnect has really killed the vibrancy of the low and middle end of the market.

Okay, well, there’s one big thing they can do to make sure the high end belongs to them, and it’s to integrate AI in very specific ways directly into their operating system. We’ll see how that plays out. There’s also a thing that’s different this time: there’s a vastly more energized European regulatory environment. Who knows how this will play out?

Sure.

I can’t tell you, but all that stuff is a swirl, and you’re a headphone maker, and it seems like different headphone makers are taking different approaches to this.

The only counter that I might offer is that there’s a lot of talk about the walled garden of the Bluetooth advantages that the platform companies have. But I would say even without the European regulators, it’s notable that something like the Meta Glasses can have a reasonable, I think, commercially viable set of experiences that are principally powered by compute that lives on your phone and still deliver conversational AI that is mediated by a Bluetooth connection, admittedly behind some of the obstacles that you’re describing. But those obstacles aren’t so profound that clever companies aren’t figuring out how to work within those constraints.

I’m very curious to see how that plays out. It doesn’t seem like it’s in focus. Again, probably because, as you say, none of the products are in focus. No one knows what a hit product here looks like, so you might as well let a thousand flowers bloom and then radically constrict the market.

Yeah, sure.

Tom, I’ve burned just a ton of your time. I think you can tell I could talk to you for ages and ages about all kinds of things.

Super fun.

Tell us what’s next for Sonos. When does all this stuff ship?

So I think by the time this airs, we’ll have announced all of this to the world. We’re having a big event in New York to share Sonos OS 27, Sonos Voice 27, and the Ace Ultra and Beam Ultra. We can’t wait to get this stuff out into the world and begin to hear what our customers do with all of these exciting new products.

Many of the software features are going to be available in early access, which will mean our customers will opt into them, be able to give us feedback, and we’ll be able to observe the telemetry that comes back from their homes and make sure that we’re delivering exactly the experience that we want. We’re excited about the moment of co-creation that’s around the corner for us.

As we talked about at the top of the hour, Sonos 27 is absolutely a nod to the fact that we intend to keep doing this until the sun burns out. There’ll be Sonos 28 and 29 and 30 and beyond.

That’s great. All right, I’ve got to ask you at the end: I have ceiling speakers in my kitchen. They’re connected to a Sonos Amp. There’s a Google Home Hub right there, and it’s broken. I can’t ask it to play music in the kitchen. When are you going to fix that for me? I’m ending on tech support.

All right. I don’t know the answer, but I’d love to figure it out for you. I know the right folks to call, so let’s get you connected.

We’re going to have a Decoder roundtable.

If your listeners have similar kinds of problems, we’ve got great people to help them, and we always like to hear from you. You can find me on social pretty easily.

Usually I end by inviting people back, but I’m saying we’ve got to fix it, and then we’re going to talk about what it took to fix it. That’s the plan.

All right. Right on.

Tom, this is great. Thank you so much, man.

This was super fun. Thank you.

_{Questions or comments? Hit us up at decoder@theverge.com. We really do read every email!}

Today on Decoder, I’m talking to Verge senior AI reporter Hayden Field about some pure Decoder bait: the seemingly-endless org chart changes at OpenAI, and how all of them seem to consolidate power under cofounder Greg Brockman, the company’s president.

While Sam Altman is the CEO and still OpenAI’s most public face, Brockman has amassed enormous power and influence within the top ranks of the company as other senior leaders have left in rapid succession these past few months.

Verge subscribers, don’t forget you get exclusive access to ad-free Decoder wherever you get your podcasts. Head here. Not a subscriber? You can sign up here.

Brockman now oversees the company’s entire consumer and enterprise product teams, including ChatGPT, Codex, and its major infrastructure build out. As Hayden recently reported, he is effectively now the day-to-day operational leader of OpenAI. 

This will have major effects on OpenAI as a company and its product strategy, at a time when it continues to cede ground to Anthropic in the enterprise and is preparing for a historic IPO. This is all against the backdrop of needing to turn a profit in the next few years and the company’s huge ambitions to replace both Google Search and the iPhone in the consumer market. 

So I wanted Hayden to break down for me what’s going on at OpenAI, and the increasingly important role Greg Brockman will play in its future.

Okay: Verge senior AI reporter Hayden Field on Greg Brockman’s consolidation of power at OpenAI. Here we go.

This interview has been lightly edited for length and clarity. 

Hayden Field, you’re The Verge‘s senior AI reporter. Welcome back to Decoder.

Thanks. It’s great to be here.

Always, always chaos when you’re here, Hayden.

Absolutely. There’s never a calm week. If there’s a calm week coming up, I know something even crazier is coming the next week.

We should just rename the show The Real Housewives of AI.

[Laughs] Honestly, that would be fitting.

Straight up, that’s what we should do. A lot of personalities, a lot of feelings, a lot of relationships that people have really come to value over a long period of time.

And a lot of lore. A lot of lore that goes between all these people for years and years and years. It’s crazy. And now they’re even putting out profiles on some of their spouses. The circles and the people themselves are really interesting.

Today the drama is about OpenAI and specifically Greg Brockman, who seems to be consolidating even more power at OpenAI. You just wrote a long story about this. It seems very clear that Greg is emerging as the central decision maker at OpenAI. Describe what’s going on.

Greg has obviously been pretty influential at OpenAI for a really long time, but what’s different now is that he has so much control over the day-to-day operations in a way he didn’t before. He was always a cofounder. He was always heavily involved. It was him, Ilya Sutskever, and Sam Altman for a long time, all in these email threads that came out during the Musk v. Altman trial. 

We saw them all talking about the future of OpenAI, what it should look like, and all their strange dynamics with Elon Musk. But Brockman was in a big-picture role before, and now he’s increasingly taking on so much power in the day-to-day. Altman has been focusing more on the big picture, the IPO stuff, and the direction of the company at large. 

Meanwhile, Brockman is amassing all this power day-to-day as other executives leave. He’s been in control now of Codex, of its enterprise business, the consumer side. There are four arms of the company right now and he’s in control of all of them. It’s been really interesting to see how that’s happened over the last few months.

Greg is a character in the AI story. Certainly inside of OpenAI, he’s been a central figure in a lot of the company’s drama over time. So if you’ve been paying attention to the AI industry, you’re familiar with him. But if someone haven’t, explain who Greg is and why he’s such a notable figure both in AI and at OpenAI.

He has been on my radar for seven years now. He’s been on other people’s for even longer. He dropped out of MIT to join Stripe way back in the day. He was Stripe’s CTO during its really early explosive growth phase. He left in 2015 to cofound OpenAI. That’s how he came on the AI radar.

Ever since then, he’s been a really influential figure. He, Ilya Sutskever, and Sam Altman were all heavily involved in the drama with Elon Musk really early on, raising money, trying to lure people over to OpenAI, hire top talent, and figure out who was going to control AGI in the event they ever made it. There’s a lot going on there. In the years following, Mira Murati joined and Brockman, Sutskever, Murati, and Altman became the four main players here.

These were the most talked-about people at OpenAI. You saw them in the headlines the most. They had the most power at the company. And then when the board coup happened in 2023, that’s when Brockman flew onto the scene in a new way. He had been behind the scenes a little bit, under the radar. He was not really doing that many public interviews. I interviewed him in 2020, but it’s not like he was a talking head. You weren’t seeing that much of him unless you were really watching the company and the industry. 

But when Sam Altman got fired by the board, Brockman was so incensed that he quit immediately. He and Altman were going to start their own thing, or were potentially going to lead a department at Microsoft. That’s when he was making waves in a new way because he was hellbent on the drama.

Brockman said, “Okay, if Sam goes, I go.” That’s when he became Sam’s right hand in a new way. Before that, he was just one of the many execs. They were close, but the board coup made them closer than ever. That’s also what made Sam know that he could trust him in a new way and thereby later greenlight him getting more power.

This was a big bet in the politics of the organization. Mira was going to become the new CEO for five minutes. Then there was going to be another new CEO and it was unclear who would stay and who would go. Justifiably, Altman did not know who he could trust in that moment, except Greg Brockman was ride or die. He said, “I’m leaving. I’m going.”

The five minutes that they were going to start a new division of Microsoft were some of the most hilarious five minutes in tech history. But Brockman had basically made it clear he was ride or die for Sam. Over time, that bet has paid off. That move has paid off. Even during that board coup moment, it was not clear that Sam would ever come back.

I remember I was backpacking in Patagonia at the time and I didn’t have a laptop, but I wrote six articles on my phone. It was a crazy time. It seemed like Brockman and Altman were going to be fine, but just be on their own. Of course, all the hundreds of employees signing a letter that they would leave if Sam was not reinstated is what got the ball rolling on him coming back and then pulling an UNO reverse on all the board members who had voted to oust him, except one. That was a big moment where he didn’t know who he could trust.

Some of the people who he had trusted the most were the ones who voted for him to be fired. That went a long way with him that Brockman quit as well. Ever since then, we’ve seen Brockman positioned in the public eye a little bit more. He’s been in the press more. This summer is when he really amassed a lot more power after a couple other executives left, and he’s now in control of quite a lot of pillars of the company.

What’s really striking is, yes, we can talk about 2023 and the board coup and Sam going and coming back and who was going to stay or go then, but actually this year has been more dramatic in terms of executive departures from OpenAI. 

Just since April, the list is staggering. I’m just going to run down it: Bill Peebles who ran Sora; Kevin Weil, who was head of product at Instagram and then VP of OpenAI’s science arm; Fidji Simo, who was supposed to be head of product and was at one point the AGI chief, left on health leave and then just never came back; Kate Rouch, the Chief Marketing Officer; Srinivas Narayanan, the CTO of B2B; Brad Lightcap, who was the former CEO and the head of special projects; and then just recently Denise Dresser, the Chief Revenue Officer, who is an important character if your company’s about to IPO. 

That is a lot of people, and it seems like Brockman just took all of their roles as all of those people left. Is that what’s happening here?

More or less. Some of them definitely affected his position less than others. For some of them, like CMO Kate Rouch, leaving didn’t really give him that much more power. But what was interesting is Simo, the ex-chief of AI at the company, is the one whose absence left a huge gap for Brockman to fill. He took over a ton of different aspects. 

As you just alluded to, a lot of the people who had just left the company had also recently changed roles. Through one way or another, a lot of these people leaving created a power vacuum that Brockman could then capitalize on. It’s not like he had this master plan of amassing power, but either way, he ended up with a lot of power. 

What is interesting is when Brad Lightcap transitioned roles, for example, one of these other people took over his COO responsibilities. Guess who it was? Dresser. Then she leaves. Now some of those responsibilities fall to Brockman. It didn’t all happen cut and dry, but in one way or another, people switching roles and then leaving the company, he ended up with a ton more power. I can’t even calculate the ratio really of how much power he ended up with. 

Now he’s in control — to give you some context — of the entire product strategy of the company, which is obviously incredibly important when you’re about to IPO and you’re getting a lot of pressure to turn a profit. He’s in charge of the company’s entire scaling arm, as well as core product and platform; all of consumer, which includes health, commerce, personal finance, ChatGPT, and other consumer-facing stuff; and core infrastructure, ads, data science, and growth. What isn’t he in charge of, really? You know what I mean?

I want to talk about Fidji Simo for one more second here. She was the former CEO of Instacart, but before that she was a really high-ranking executive at Meta. She ran the Facebook app. She was in charge of a lot of monetization.

When she came to OpenAI, it felt like her role was to turn the consumer version of ChatGPT into a product. There were lots and lots of people leaving Meta to go to OpenAI for a while, such that basically we had reported that every all hands at Meta was about, “What are you going to do about OpenAI? They’re taking all of our people.” 

You just saw this exodus of people from Meta going to OpenAI in various ways, led by Fidji Simo, in a moment where it felt like what they wanted to do was take on Google as a big consumer product supported by ads. 

It’s funny, Altman actually copped to that ambition on the David Senra’s show last week — here he is, giving Peter Thiel credit for saying OpenAI should compete directly with Google: 

Sam Altman: He’s like, ‘The power of this is the power of the Google text box, it’s a text box you can type anything into and it does the right thing. Clearly, the empty text box worked for Google so why don’t you just double down on that?’

That’s pretty straightforward and that’s what they did. Altman says they went super hard at it — being the empty text box, taking the Google business model, which is one of the best business models in internet history. That was a big bet, and they hired Fidji Simo to go run that bet. But she left, and that does not seem to be the emphasis anymore. The emphasis is on Codex, on enterprise, on competing with Anthropic. 

What’s left of that? Was that just a misfire and she had the wrong ideas? Is it still the plan? This seems like the question about OpenAI.

Some of it has fallen by the wayside, which we’ve seen for a bunch of stuff that OpenAI has tried in the past year or two. They vowed to stop doing all their side quests, which is ironically something that Simo herself wrote in an internal memo.

Several episodes ago, you were on and we just talked about the series of code reds at OpenAI.

Right now they’re focused on the key revenue drivers, which are enterprise and coding. That’s really what they’re all-in on. They’re also, of course, really focused on building a “super app,“ which is supposed to incorporate all this stuff together, and making that super app better. 

Besides that, they’re trying to cut the fat and operate as a leaner business, which has been helped by some of these executive salaries being cut. They’re trying to look better on their balance sheet when they IPO. There’s a lot of pressure to do that, especially when SpaceX just IPOed at a crazy valuation and Anthropic is apparently gunning to exceed that. We’ll see if that actually happens.

When they changed Simo’s title from CEO of applications to CEO of AGI deployment, that was the most Decoder thing of all. I looked at their org chart changing and I thought, “This isn’t going to last.” 

Over and over and over again, they keep reassigning people. And then as you say, they leave and we see one person take over those roles and consolidate power. We’ll come to the IPO in a second, but where is Sam Altman in all of this? It seems like he should be running OpenAI.

He really likes focusing on the long-term stuff. The events of the last couple of years and all of the lawsuits have shown him that maybe he isn’t supposed to lead direct teams in a huge way. He has a few direct reports, but should he really be involved in the minutiae of the day-to-day operations? Maybe not. 

Altman has said before, and in blog posts that the company put out years ago, that this is the trend that they’ve been following. I remember a year or two years ago, the company would put out summer blog posts saying that different C-suite executives’ roles would be changing slightly, and Sam would be focusing a little bit on the longer term, on research. This is just a continuation of that trend, in my opinion. As we know — and this is a big Decoder thing as well — the more power you have at a company, the higher your title, the less you probably are involved in the actual day-to-day operations, especially at a tech company.

This is another example of that. Sam has always been interested in the research and the long-term stuff. Brockman is an engineer. He was described really early on at the company as an “engineering workhorse that pushed to build scaled-up systems that would train the AI and make it work.” So he’s always been seen internally by employees as a can-do person. He makes things happen and he has the engineering training that is required to scale a system. So I could see investors being thrilled about this.

I could also see Altman saying, “I know this guy’s in my corner.” Altman already cleaned out the board. He’s surrounded by people that are going to support him. He trusts Brockman. So he says, “I actually don’t really want to be involved in the operational stuff quite as much. I want to be focused on the direction of the company. So you handle that. You’re an extension of me. I can trust you. Go ahead.” That’s what I think is happening here.

You’ve talked to Brockman before. I’ve met him. He just did a video with our friend Joanna Stern. You can see he’s pretty direct. He answers the questions. What kind of character do you think he’s going to be as a person who’s operationalizing OpenAI? Is this “get stuff done, make the numbers go up” situation? Or is this more expansive of an approach?

It’s going to be interesting to see how he handles this much responsibility, especially this many disparate arms of the company. He’s a really can-do person. He has a lot of engineering training. He’s pretty respected within the company, but I do think there’s going to be a lot of natural tension here because consumer, enterprise, health, and personal finance are really, really intense departments that handle data extremely differently. And then you’ve got all the core infrastructure stuff and the behind-the-scenes stuff that makes everything work. We’re going to see a lot of natural tension arise, especially because this company has a limited amount of funds. 

Even if it’s a huge number, it’s still limited in some way, especially now compared to a few years ago. They have a limited amount of compute. That’s something that every executive at OpenAI has run into problems with — allocating the compute. The research side of things says they don’t get enough sometimes, or the product side says they need all of it. It’s going to be really interesting to see how he squares all this, especially as someone who’s probably trying to make everyone happy. He’s not going to be able to, and so he’s going to have to make some hard decisions.

This brings us to the IPO, because the IPO is where those hard decisions have to pay off. There’s been a lot of talk about OpenAI playing catch-up to Anthropic, particularly in the enterprise and with coding. There was a Wall Street Journal story last week that said OpenAI’s revenue in Q2 lagged behind Anthropic’s. 

Obviously, Anthropic will tell you they’re profitable even though we haven’t actually seen their numbers and we don’t know how they calculate it. You’ve mentioned the IPO several times now. SpaceX just went public. Big IPOs are all the rage. What role is the IPO playing in all these changes? Is it focusing the company down? Is it making that equity pay off? Is it needing to raise more capital? What’s the shape of it?

It’s absolutely behind a ton of these changes. It is normal to see a lot of changes in the C-suite ahead of an IPO. But some of the sources I spoke to within the industry, who study the way these IPOs typically happen in tech, said that the interesting and unusual thing here is how many C-suite execs left in such a short timeframe, because they know that looks bad for the company. 

It’s one thing if you need to cut salaries, make the balance sheet look a little bit different, make the company leaner. That’s normal. It is an easy way to do that, to not replace someone. We can only speculate how much each of these people’s salaries was. This is an easy way, cut and dry, to help out the balance sheet.

But doing all of this in this amount of time reminds me of what we were talking about last time on Decoder about Google and DeepMind and how maybe Demis Hassabis didn’t leave when Jeff Dean did because they didn’t want to make the company look bad. Now at OpenAI, people are leaving left and right — sometimes within 24 or 48 hours. 

Brad Lightcap, for instance, just got a new job as special projects head. Then a couple months later he’s out, even after being at the company for so many years. Part of it’s probably just cashing out. You can make a lot of money if you’re an exec and the company’s about to IPO and you have a lot of stock options. But it is unusual, my sources said, that it’s happening in this short amount of time. 

The IPO is interesting for a number of reasons, but to me, the most clarifying is that OpenAI is up against Anthropic. There are numbers coming out, reported numbers, about Anthropic’s finances that make that company look pretty good. The company hasn’t really challenged them in public. So we have to assume that they’re happy that there’s reporting out there that the numbers look good for them. 

But all of that is enterprise. They’re selling Claude to big enterprises, to the government. It’s effective. Maybe their token prices have high margins. It’s unclear what the internals of Anthropic’s business look like, but the numbers that we can get look pretty good and they support a big IPO.

There’s a lot of excitement around Anthropic for that reason. That’s Anthropic as an enterprise software provider. That’s their business. It’s a thing it’s focused on. It’s all they do. OpenAI has taken a lot of shots. You’re talking about cost-cutting ahead of an IPO. Are they shutting some things down? Are they closing the aperture on all the things they’re trying to do to compete with Anthropic ahead of this IPO?

It’s interesting because I’ve seen a slight shift from OpenAI. Earlier this year, the company was saying, “Enterprise and coding. That’s what we’ve got to focus on. We have to make money. We have to compete with Anthropic.” Now, however, ahead of the IPO, OpenAI is getting a little bit of different advice.

Yes, the company needs to focus on those things because obviously that’s where the money is. But now it’s also being told, “You also have to differentiate yourself from Anthropic. You can’t just be a copy that’s lagging behind. You have to be doing something different.” OpenAI is going to really lean into consumer and hardware as well. 

One attorney told me that for OpenAI to be successful in the public eye and in investors’ eyes, the company needs “hardware and consumer products to sell to consumers, and luckily, that’s something that Brockman has experience with.” It’s hard to do everything, but they’ve had to narrow a ton and then slightly widen again and say, “Consumer is what we’re really known for.” 

They’re saying, “We’re the Kleenex of AI right now because of how we’re known in the consumer world.” People say, “I’m going to ask ChatGPT,” or, “I ChatGPT’d it.” And they’re talking about just using AI in general. Maybe they weren’t actually using ChatGPT, but that’s the way they’re referring to things a lot of the time. So OpenAI knows they need to capitalize on that. That’s one way they can be different from Anthropic. Of course, Anthropic also offers that, but they’re much more known for enterprise and that is why they’re making so much money. 

But OpenAI is leaning into hardware. The company has Jony Ive. It has to not cut some of this stuff entirely because then it’s going to look like Mark Zuckerberg with Meta. It can’t go all-in on something and then just completely cut it, or they’re probably worried about being a laughingstock. You have to really cut the true fat, which they did with a lot of their side projects. But as far as consumer, hardware, enterprise, and coding, those are their main bets right now. We’re going to see them really double down on that heading into the IPO.

Come on, Hayden. Aren’t we actually in the metaverse right now? Technically, you and I are in the Metaverse together right this second. It worked.

Absolutely.

It happened.

Should have made an avatar.

Technically we’re here in our bodies on the internet.

Yep, that’s true.

What argument can you have except that the metaverse definitely worked and Mark Zuckerberg was super right about all of it? 

The consumer business is really hard. The scale of the play you need to make the consumer business work is on the order of fully overtaking Google, which seems very challenging. In hardware, you have to replace the iPhone. If you do anything other than replace the iPhone, people still have their iPhones. And then you’re going to lose to Instagram every single time. 

Have they said anything about how they plan to do either one of these things? Because replacing Google with a Google-level monetization engine seems very hard. Replacing the iPhone — even if you have Jony Ive — such that people don’t still have their iPhones, seems very hard. Neither one of these things has been in any kind of focus for me at least.

Like we talked about last time, I really am skeptical. Hardware is hard. I’ve seen way too many companies crash and burn when they try to make an AI hardware device. It’s going to look beautiful because Jony Ive is in charge. 

But as for how useful it’s really going to be, especially in the era of AI populism when there’s a huge backlash against using AI, I don’t know how excited people are going to be on a broad scale to have something visible pinned to them or in their ears or on their table where you can tell they are using AI. It’s going to be really exciting for a subset of people, but for the broader public, Meta’s glasses are still uncool, still being called pervert glasses. With AI hardware, you’ve got a steep hill to climb.

Let’s talk about what happens next with OpenAI, because those are the challenges. What you have now is a new leader who is, as you said, running most of the company. Greg Brockman was just on CNBC last week. He basically defended the turnover. He said this: 

Greg Brockman: “I’d say fundamentally, we’re a very resilient organization… If you look over the years, there have been different eras where we have different sets of leaders in place. I’m constant, Sam is a constant. I think that we are stronger because of that resilience and diversity.”

So this might be what Greg Brockman has to say: “All these people are gone, but I’m still here. Sam’s still here, and the company’s still the same.” Do you think it’s just the thing that he has to say? Or do you think there’s something more real about OpenAI as a company, where you have these two leaders who are clearly ride or die for each other and you can swap in and out all these other executives and the company will still have its own unique vision?

He would like us to think the latter, but it’s more of the former. It’s never good when you have a bunch of people leaving at once. It’s never good when you’re constantly restructuring or reorganizing the company. It’s bad all the way down. Employees are probably feeling weird. They have a new person in charge. They’re being shuffled between different teams. Their teams are headed by one person and then their boss’s boss is someone else, which probably means their bosses are weird. 

It’s not great for productivity, especially heading into such an important time for OpenAI. Even if it’s a move that makes sense and is good long term, it’s still going to be weird for people in the short term.

In some ways the latter is true in terms of Brockman and Altman having been at the helm from the start. Someone who’s been there for eight months who leaves is not going to have as much of an impact. But we’re seeing some people that have been there for a really long time leave as well, like Brad Lightcap. He was someone that I had tracked for years and years. All of a sudden he’s out, after switching roles, and now he says he’s going to start something new. He says he’s implying he’s still going to work with OpenAI in some way, shape or form. He and Altman had a really friendly exchange on X after he left, but a lot of it’s for show. 

They’ve got to make sure that people aren’t skittish about the company right now. It’s something he just has to say. But it’s also true that if there were two people that would be the most influential if they left, and luckily those two are still there, Brockman and Altman.

You’ve mentioned the public antipathy towards AI several times now. There is just a lot of anti-AI sentiment out in the world. And it’s pretty politically coded, although it scrambled some political lines. Hating AI is pretty bipartisan. Hating data centers is pretty bipartisan. We’ve got a lot of reporting about that on the site. 

Brockman is into politics. He donated $25 million to MAGA Inc. He’s one of the top donors to Trump overall. Do you think him being so openly political will help or hurt him as he becomes more of a character, more of a visible leader of OpenAI?

It’s going to help the company from the outside, because they’re trying to push a lot of stuff through during the Trump administration and they really want a voluntary regulation framework. They want to be in Trump’s ear in a good way. It’s going to hurt Brockman internally, because a lot of employees at tech companies that I’ve interviewed in the past year, no matter what company they work at, are incredibly angry if their CEO or their C-suite isn’t doing enough to speak out against some of the Trump administration’s decisions. 

Now, Brockman isn’t just not speaking out against them, he’s also supporting the admin with tens of millions of dollars. I could see this really hurting him from the inside. Maybe some people don’t take him seriously. Maybe we will see some departures. It’s going to help the company from the outside, just because when you’re giving a lot of money to Trump, he seems to let you curry favor with him. We’ll see.

To be clear, we’ve seen Sam Altman stand next to Trump and announce data center projects. So whatever reputation OpenAI was going to have because of its leaders, it might already have, but the specific political giving seems new. And as we head towards the midterms, it seems like a new challenge for the leaders of the company to be openly associated with. Do you think it’ll change any of the valence around data centers and AI, how the public feels about them?

I don’t know that it’ll change that because we’ve been careening towards this for a while. People are really mad about data centers and it’s a bipartisan feeling, like you mentioned, and same with AI. Sam Altman and other AI CEOs have said in the past month that they feel like AI has a big PR problem and that they haven’t done a good job of showing people the good parts of AI and how they need to do better to show people why they’re even building it in the first place. 

Maybe they do, but people have been told a lot of the good things about it and they’re still not thinking that they’re as good as the bad is bad. The AI industry is in for a rude awakening there. But as for what you mentioned about Altman, yes, he and every other AI CEO have been at Trump’s dining table. They’ve been caught on hot mics praising him.

It’s not something hugely new that Brockman’s giving this money, but what is new is that he’s giving it so much of it in a personal capacity, so much so that Altman even had to come out and say at some point, “Brockman only did this in a personal capacity. We’re not saying that we’re super behind him.” 

He also couldn’t really say they weren’t behind him, but OpenAI had to separate themselves a little. That’s what’s interesting to me about this. It’s like the personal capacity of his giving has made so many headlines and OpenAI tried to distance itself a little bit, but not too much. Now they’re not going to be able to do that because he’s essentially running day-to-day operations.

I’m very curious to see how that plays out. I don’t think we quite know yet, but as he becomes a more visible leader of the company, being that open about his politics, even in the context of the other big tech CEOs, is different. We don’t quite see that from all of them. We see it from Elon, but everyone else plays it pretty safe. This is new, especially for a leader at OpenAI. 

All right, let’s end here. Let’s say it’s two years from now, OpenAI is public. Do we think Greg Brockman is the CEO and Sam is just the “chairman of raising money,” or whatever it is that he’s doing?

I could see that happening. It would take a couple of big things to make that happen. Altman really likes being in charge, so I don’t think he’s going to go super quietly. But if he ever needs to move on to a more big-picture chairman role, if the IPO doesn’t go so well or he gets dragged in the public eye or something and still needs to be part of the company but not the CEO, I could definitely see that happening. 

Especially because Brockman has a lot of ambitions. We saw his journal entries come out during discovery, during Musk v. Altman, and he was writing some pretty ambitious things. 

“What will get me to a billion dollars?”

He’s saying, “This is my one chance to be in charge.” There are a lot of entries that he’s writing about how he’s put his blood, sweat, and tears into the company and how he needs to be in charge in some way, shape, or form. He will probably be gunning for a role like that and he’d be happy to take it on.

We’ll see what happens in the next few years. But I definitely don’t think that’s out of the question, especially because he’s been there since the beginning. He has a lot of product experience and he’s clearly ambitious. The other thing that is interesting is, way back in the day, he wasn’t always so aligned with Altman all the time. 

In fact, Brockman and Sutskever, OpenAI’s chief scientist, were pretty aligned against Sam every once in a while. Not against him, but they were questioning him and pushing back. They were saying things like, “It seems like you really care about this CEO role. It seems like you really care about being in charge and having a lot of power in this way and that way. How does this relate to your political ambitions? How does this relate to who’s going to control AGI? We really don’t want it to be a dictatorship.” 

Altman would probably be hard-pressed to give up the CEO title when he fought so hard for it, even 10 or 11 years ago. But if he ever does, it seems like Brockman would be happy to take it on.

I will end by stating my prediction that I’ve been making almost the whole year now. I don’t think we will end 2026 with OpenAI as the same kind of company as when the year started. Actually I would say, given all of this turnover, that prediction has already come true. Structurally, it is a very different company with very different goals than the year started.

But I’ll just put that prediction to you. You cover the company way more closely than I do. Do you think there’s any way that OpenAI looks the same at the end of the year as it did at the top of the year?

Absolutely not. When a company goes public, so many things change. Especially when the end of the year is only five months away. 

It’s a scary number of months away. It’s tomorrow.

Let’s say maybe not by December 31st, but six months from now, 100 percent it’s going to look different and probably sooner than that. They’re going to have to make a lot of changes. They’re going to be listening to their investors in a new way and they’re going to be beholden to them in a new way. There are only certain parts of this industry that make money and OpenAI is not in the lead in those parts. 

They’re going to have to go all-in on that type of stuff. The type of research that they want to do to stay at the frontier, they can make a case for that. Money-wise, they can say, “Look, in order to stay in the lead, we have to do this long-term stuff a little bit.” But they’re not going to be able to go all-in on it all the time because it costs a lot of money — and they have a limited amount.

We’re going to see a lot of the anxieties that we saw Brockman and Altman voice a year ago, two years ago. I’ve been in the room with them where they’re talking about their fear that they’re running out of compute and how they’re going to scale. Their whole job for the next six months, the next year, et cetera, is scaling. 

They’ve had so many concerns about this. They’ve had so many fears about how they’re going to do this with their limited compute, especially now that they’re going public and they’re beholden to investors in a new way. There’s no way the company is going to look the same.

Hayden, this has been great. Thank you so much for being on Decoder again. We’ll have you back when another season of The Real Housewives of AI kicks off, which is probably going to be tomorrow at the rate we’re going.

[Laughs] Absolutely. Thanks.

_{Questions or comments? Hit us up at decoder@theverge.com. We really do read every email!}