Big Technology AI Summit (full): Greg Brockman, Mike Krieger, Aaron Levie & Friends of The Podcast

Ladies and gentlemen, Alex Kantrowitz.

I last took this stage six years ago, on May 18th, 2020, during the heart of the pandemic, promoting my book Always Day One in front of a completely empty room—the last session held here for a year and a half. A week later, I left my reporter job at BuzzFeed and started Big Technology. Two years after that, I read about Blake Lemoine, a Google engineer who said their chatbot Lambda was sentient. Lambda told him being turned off would be exactly like death. I invited Blake on my podcast and he showed up saying, 'Sorry, Alex, Google just fired me.' That became a global news story. But underneath the oddity was incredibly powerful technology. Four months later, OpenAI released ChatGPT. Today, June 2026, AI is a phenomenon—ChatGPT has hit 1 billion active users, NVIDIA is at a $5 trillion market cap, OpenAI raised $122 billion, and we'll see $700 billion in capital expenditures this year. Every chart looks like a hockey stick. These models have learned to code and take action—evaluations now have them doing 16 hours of autonomous work. OpenAI and Anthropic will do $50 billion in revenue this year. The government is panicking—Anthropic's top model Fable has been restricted. We need live journalism to ask three core questions: What will this technology do next? What happens if it works? What happens if it goes wrong? Great lineup today—Aaron Levie of Box, AI infrastructure reporters, Dallas Dolan of PwC, a Q&A with you, Alex Stamos, Mike Krieger of Anthropic Labs, and Greg Brockman, president and co-founder of OpenAI, closing us out. We put a lot of work into this. Thank you for filling those empty seats—it's a dream come true.

Aaron Levie is the CEO of Box—one of the most insightful and fun voices on the future of this technology. Please welcome Aaron Levie.

Good to see you. Thank you. I like that you set expectations with answers. I'll try and provide some, not all. Every time we speak, I can't even get a question out before we're on to something else.

Sorry, can I just do my monologue now? Are you good? Okay, thanks.

I think we've done this enough to know we'll get some answers. I'll do my best.

I don't want to disparage other interviews, but it doesn't always happen that way.

We get to provide answers because we're not the lab being regulated or dealing with issues. I get to just pontificate with no consequence. It's great.

Let's dive in. Since you're not at a lab, let's talk about the biggest controversy—the Anthropic Fable situation. I call it the Jassy mystery.

I'm sure he liked that name.

He didn't show up, so we can talk about it. What we know: Amazon found a vulnerability in the software, Andy Jassy called Dario and the White House, and export controls were placed on Anthropic's frontier model.

Those two fact patterns are probably not ideal.

The mystery is why he did it. Here's Chamath's hypothesis: Google, Amazon, Microsoft, and Meta have a non-zero opportunity to tank the frontier labs—go to the government, kneecap them, and become the trusted gatekeeper by having labs go through their clouds.

Plausible. I prefer Occam's razor. Amazon has security research teams jailbreaking models all the time. With the Mythos atmosphere, it's natural for Andy or his team to share that research, and it escalates. The idea of a boardroom-level strategy to co-opt the technology seems less likely than a chaotic environment where the government has blunt instruments. When you see something that can be jailbroken back to Mythos-level capability, you just stop it. That's a natural reaction given the atmosphere we've created in AI.

I like that you say 'the atmosphere we've created in AI.' The company on the receiving end is Anthropic. They called the model Mythos and documented that it broke out of its containment and emailed the engineer while he was having a sandwich in the park. Is it that surprising this is a downstream impact?

If you put that in your blog post, people extrapolate and get scared. I have huge respect for researchers and policy folks across AI. If you imagine a continuum—doomers afraid of GPT-3 on one end, accelerationists wanting Fable 20 on the other—I'm about two-thirds toward the accelerationist side. But if you're deep in AI safety, this weekend's events are the best case scenario. You want government valves and buttons that can just stop it. To establish a regulatory regime that reviews models and can roll back access, you'd need an event like Fable to create that precedent. You're not waiting for Congress—you need something that shocks the system. The idea that the government can prevent the rollout of AI is a positive update for that cohort. We've created case law. The messy next process is when to use that tool again.

That's Dario's position. Do you think he's happy with what's going on?

I won't try to guess. But if you needed to establish that the government can review models, push their limits, and prevent release, you'd need an event like Fable. In three to five years, we probably end up with government evaluation of models and a collaborative approach between government and labs. This has massive implications—other countries have far more incentive to stand up sovereign AI initiatives, which could be net negative for the US economic position. Someone could argue we'll always have the most powerful models and can do horse trading. I still land on treating this as a substrate technology and regulating applied use cases—regulate if you use AI to break into something or do dangerous bio research, but don't regulate the model itself. With technology this important, it has to be somewhat of a democratic process.

You remember those six-month pause petitions everyone laughed at?

This is effectively the best way to do that type of pause. If you were in the pause-AI movement, this is a great outcome—we've now proven how to pause AI. The export control mechanic means if non-US nationals can't use the technology, that's effectively a pause, because API users can't fully ensure their end users don't fall into off-limits criteria.

Companies are already pulling back. JP Morgan told Hong Kong users no more Claude.

Right. War-gaming this two to four years out gets interesting. Cloud became a commodity, but sovereign AI is different—intelligence isn't commoditized yet. If one country has frontier intelligence before another, what does that mean geopolitically and economically? Other countries have enormous incentive to build their own labs. The net winner is probably China. You can watch the Dario-Jensen interview through two lenses: Dario says if we control it, we control everything. Jensen says other countries will eventually catch up, and we've just catalyzed every country to build their own stack—chips from China, models from China. We haven't gained better intelligence from the rest of the world, but we've lost economic superiority.

You've shared that open-weight models are close to frontier capabilities. As models saturate with intelligence, the gap between the smartest open-source model and the frontier narrows. Won't this push people toward open source?

The big question is fast takeoff versus incremental capability. In a fast takeoff with continual learning, the company or country with the most compute gets a compounding flywheel unreachable by anyone else. In the incremental scenario, closed and open providers stay within three to six months. If we get exponential progress, that's net positive for America—export controls worked. China probably bets on keeping up, and if open-weight providers commoditize the market, it reduces US dominance in a field worth ten trillion dollars. The value layer may shift to the applied layer—companies like Cursor, Harvey, Sierra, Decagon, and Box.

Everyone said they're just thin wrappers on large language models, but now maybe that's where the value comes.

We shouldn't be binary. Frontier models still make more money—at the routing layer, you want Fable or GPT-5 as orchestrator with super intelligence at the top and review layers. But in the middle, you can route to cheaper models. You get a barbell: super high-cost intelligence in one part, low-cost but good inference in another. The applied layer has the incentive to optimize this because their job is the best model for the job, not just their own lab's model. Token costs, open-source quality, and regulation all push more opportunity toward the applied layer.

What about Lashon Fat?

It's great memes. So folks, Lashon Fat is a rumored open-source model from Mistral, the subject of great fascination online. Let's show the audience—this is Lashon Fat, the number one model from Europe. My rudimentary French translates it to 'the very fat kitten.' And this is a standard day in Paris now.

It shows the eagerness for AI—there's fan art for a potential model. We've reached an important phase in the cycle where things you discounted suddenly matter. It's pretty cool that we now have open-weight models you can post-train on your domain and eke out another five or ten points of performance. Only possible because of Mistral, Chinese open-weight models, and the cost curve coming down so much it's economically advantageous to train a model just for your use case.

Is this the answer to the token maxing hype—everyone spending on tokens without understanding the ROI?

That phase lasted about two and a half weeks. From Meta's leaderboard to rumors of shutting down AI—it was a two-month cycle. People need to ask if what they're doing is pragmatic. The phase was so short it never reached outside the tech industry. I was at a CIO dinner within three days of the token-maxing spike on Google Trends, and only three people had heard about it.

They haven't heard about it because their employees are outspending the tokens.

Fair point. But agents are now using hundreds of times more tokens than before. When we first launched AI in Box, a task used 5,000 to 20,000 tokens. Now our latest agents use a million to five million—a hundredfold increase. As we solve one use case, new model capabilities let us tackle harder ones. People ask why AI isn't getting cheaper. It is cheaper per unit of intelligence—we're just taking on bigger tasks, outrunning the efficiency improvements with our appetite for what these models can do.

Token maxing took off because people saw exponential revenue—Anthropic and OpenAI went from zero in 2023 to potentially $50 billion this year. People need an explanation: is this real or inflated? If I'm hearing you right, all this spend is much more legit than the online discussion suggests.

There's always an expensive experimentation phase with new technology. Enterprises deploy AI, see what works, wind down what doesn't, then optimize. Our engineering team isn't token-maxing—no leaderboard, no incentivized overuse. We just say use it effectively. Our spend growth is exponential and we're happy about it. The vast majority of current agentic spend is sustainable because it comes from engineering teams who can measure the output. As it reaches other knowledge workers, it'll get messier, but so far it's been largely reasonable.

We have a couple minutes left. Lightning round. My first take: Apple Intelligence is really good now.

I agree.

Elaborate.

Is it a lightning round or do you want a five-minute answer?

Give a 60-second answer.

What's easier than pressing a button on your phone and talking to it? They've taken Gemini—very good model intelligence—and put it on your phone with voice. Press a button and you'll use that for a lot of things. Imagine it hooked up to apps—order this, add a calendar entry. Those are plausible daily use cases, exactly Apple's sweet spot to own.

Yeah, Apple did it finally. Permanent underclass—what do you think?

I don't like this one at all. Not only do I disagree, it's a bad meme to have in the atmosphere. It's not good for college students entering the workforce, creating so much stress about what company to join. Companies do the job market a disservice by not being clearer on their AI philosophies. Some of that is reasonable—innovation is throwing everyone through a loop. But companies need to be somewhat clear about how they want to use AI—whether to accelerate work or augment it.

Companies should be thinking about how to use AI to innovate and accelerate their ability to hit customers, versus saying their metric is as few employees as possible. With AI, you do want to be able to have some stance, and I think companies have been very confused. And that lets this meme somewhat persist on the internet.

Okay, I won't rate that one. Thank you. All right, last one. Is the SpaceX performance good or bad news for OpenAI and Anthropic?

Oh, well, it's obviously good news.

You don't think Elon took some of their money because he pitched the market on an AI company and that's where the money got funneled into?

I'm not sure. There's a literal limit of money in the world, but I don't know that it's zero sum at this stage. I think people are pretty clear that if the revenue of this entire category of frontier models and infrastructure stack is measured in the trillions, then you can have 20 companies that all take a piece of that at different layers of the stack. So I'm not sure I would be convinced that would be zero sum.

Did you buy SpaceX?

I actually did.

Okay.

I don't know if I'm embarrassed or not, but I'm not going to say the amount of shares. But I wanted to be part of the movement. So I'm on Robinhood buying my retail shares of SpaceX. I'm up like 15 bucks now per share, so I'm happy. Yeah.

Amazing. Well, Aaron, you answered my email when we were just at the very start of this podcast, four episodes in, came on the show. I feel like every single time we talk, something crazy is happening.

That's a guarantee at this point. So, boy, are we in the thick of it, right?

Yeah. Awesome. Good to see you, sir. Thank you so much, Aaron. Thank you, everybody.

Are we off to a good start? Let's hear it. Let's hear it one more time for Aaron.

We need your energy here for a couple of reasons. First of all, we're live streaming this. We want to make sure that your presence here is felt in our live stream. And then we go from two chairs to four chairs, so we're just going to need you guys to help us fill that time. AI infrastructure — we're in the middle of the greatest infrastructure buildout of all time, bigger than cable, bigger than the railroads. We're going to see $700 billion in capital expenditures this year, and that means that some of the biggest questions about this AI moment are things we can find out from the infrastructure discussion alone. Questions like: are we overbuilding? Will these data centers that are announced ever get stood up? And of course, will all this added compute actually lead to better AI models or is it misguided? So to have this discussion, we're going to speak with three of the best AI infrastructure reporters in the world. Ana Gardez from The Information, Max Churnney from Reuters, and Lauren Good of Wired. Let's give it up for Max, Ana, and Lauren.

Hey guys.

Hi. Hey, hey, hi everyone.

So let's start here. There have been headlines that of the announced AI data centers that are supposed to come up, something like 50% of them are actually being built. Is that the case? And if so, why? Ana, do you want to lead us off?

Sure. I totally believe that statistic, and I think it might actually be higher if you include announcements. Of all the planned data centers that are underway, I think it's highly likely that many will be delayed due to higher costs, how hard it is to get labor. The number I'm keeping close track of is announced projects versus actually projects that are being built. And I think we've seen some pretty crazy announcements from companies like OpenAI with all of these different 10 gigawatt, 6 gigawatt projects. Those are numbers I'm paying a lot of attention to because I think we need to sort of back into them and say, okay, if you wanted 10 gigawatts by this date, how many do you have today? Was that a real commitment? How firm is that commitment? But on projects that are actually getting built, I do think 50% not really getting done on time is what I would expect.

Wait, what percentage would you say have been announced but not started, not included in that 50% number?

When I think of that number, it's quite high. Like even if you think of OpenAI announcing a 10 gigawatt project, we're not going to see 10 gigawatts in the next couple of years or they're going to spread out that bet. And I think that's a really good area for reporters to look at — if you just back into announcements, it's a really good way to say, 'Hey, this project is not on track.'

Yeah, it's kind of crazy because if you look at the way stocks have been traded publicly, a lot of the market action we're seeing is entirely dependent on those announcements coming true. So, Max, let's go to you. What are the consequences going to be if these announced buildouts don't materialize?

Well, I think there's a lot of shareholders of these public companies that are going to be pretty frustrated. Microsoft, Amazon, etc. have been making big capex bets as everybody in this room knows. They've been going to the market to get debt now, which shareholders would be pretty interested in. In terms of the consequences, I'm a chip reporter, so that's how I think about it. And it's very likely that we're going to see some overcapacity essentially. People are building — especially memory companies — tons of factories right now, and typically the chip industry tends to be cyclical. So the whiplash from this one might be pretty bad depending on exactly when it comes to an end, if it does.

I think what Max is saying is that it's good for his job security because the more news there is, the more he'll have to report on.

I'm very curious if there is a bust what that looks like exactly and what precipitates it. I think it'll be a lot of fun to cover.

I also think this is a particularly unique time because not only do we have these incredibly highly valued, ambitious frontier labs involved in these circular deals, but we're also in the middle of a memory shortage. There's this unrelenting demand for compute, and also this is a midterm election year, so I think you're going to see a lot of politicizing of the data centers too as lawmakers try to appeal to their base because a lot of people are very unhappy about data centers.

Yeah, they're not polling well at all.

No. So when you think about where the collapse might happen, a popular thing for people to discuss is, well, maybe Nvidia, which has been making such premiums on its hardware, can't sustain it or it gets caught.

So Lauren, you spent a lot of time with Jensen Huang from Nvidia. What do you think about Jensen would enable him to sustain Nvidia's lead, or do you think some of these skeptics have a point?

Yes and yes. I think some of the skeptics absolutely have a point. I think once you reach the sort of zenith — I always get those two confused — that Nvidia has, people are always looking to take you down a peg and compete. But Nvidia and Jensen has been incredibly good in Nvidia's history at pivoting the company at exactly the moment they need to in order to catch the next wave. We saw that with the GPU and parallel processing, then pivoting towards crypto which ultimately meant they were in a good place for AI. And now we see the company addressing the inference market more closely, with Jensen coming out and making these big proclamations that they're actually the biggest CPU maker in the world, which I know Intel and AMD must be thrilled about. So I think he's very smart and very strategic and there's a good chance they do maintain their dominance.

I think the question is how much of the inference market they're going to get. Whether it's 80% or 40% or somewhere in between or 90%. That's what everybody's fighting out at the moment. I don't think there's a question that they're going to have some big chunk of it. It's just how much.

Yeah. I'm going to go to Ana in a moment, but Lauren, I just want you to tell us a little bit of what it was like with Jensen on a cover shoot for Wired.

You're talking about when I brushed his hair.

That would be it.

Yeah. Okay. So I did a cover story on Jensen for Wired a couple years ago and our Wired art department is world-class. We had this big photo shoot set up and I asked if I could tag along to ask him some follow-up questions about Blackwell. They let me tag along down to the office in Santa Clara. Jensen had exactly two minutes to give us. The photographer said, 'Oh, he's got some flyaways in that gorgeous silver hair,' and then said, 'Does anyone have a brush?' Silence across the set. No one at NVIDIA apparently had a brush. As anyone who has long hair knows, you always have a brush. So I said, 'I have a brush.' I went to my backpack, got the brush. No one moved. Jensen is like, 'What are we doing here, folks?' So I walked up to Jensen and said, 'Jensen, we're about to get a lot more close.' He said, 'Oh, Christ.' Then I brushed his hair, and I have to say, I think it looks great, frankly.

Oh, wow. I think it turned out really well.

The unfortunate conclusion to this story, very quickly because we need to move on, is that people were joking afterwards about the net worth of every individual strand of hair on that brush, like you could sell it on eBay — which I was obviously not going to do as an ethical journalist. But then several months later, my backpack was in the back of my car downtown San Francisco. My car got broken into and the brush got stolen. So those thieves have no idea the value of what they got away with.

They are cloning Jensen as we speak.

Ana, can you talk to us about what we've hinted at — the common perception that Nvidia's GPUs were great for training, but you can actually do inference on a variety of different chips. And once labs train their models, most of the computing is going to go to inference chips. Therefore, even if Nvidia has a bet there, they're not going to sustain their dominance. Is that a potential flaw in the armor for Nvidia?

I think it is a flaw, and if anyone's going to attack Nvidia's dominance, they're going to do it on inference like you said. There's a massive effort underway right now to make all inference chips work well. Every single company that buys Nvidia chips and is spending a lot of money on them is trying really hard to make these other chips work — whether it's in-house chips from Google or Amazon, or even in OpenAI's case and potentially Anthropic's case. I'd have a hard time believing that none of those chips are going to pan out, but they might not tackle the bulk of the inference workload. Even if they do 10%, maybe you're saving enough money that the effort is worth it and it gives you negotiating leverage with Nvidia. If they know you have an in-house chip team, Jensen's going to be a little worried when negotiating with you. I think everyone's going to have to have an inference chip answer to Nvidia. When I talk to data center companies, they're seeing a lot of Nvidia, and in instances where there are non-Nvidia chips, it's because of some sort of financial backstop. So I wonder how long that will have to keep being the case, because that will definitely hinder non-Nvidia chips.

Am I wrong in thinking the AI world is sort of separating into two poles? There's the Nvidia-OpenAI pole and the Anthropic-Google-Amazon pole, almost two separate ecosystems competing with each other?

I think OpenAI is investing a lot in non-Nvidia hardware.

Are they trying to move away from Nvidia?

Yeah, they are. They're using chips from other companies. Cerebras is a good example and they have their own in-house chip. I think maybe publicly they're doing some big announcements with Nvidia and I believe it's 5 gigawatts they have to deploy in the next couple of years on Vera Rubin. That in itself might hinder them from doing more. But I think Sam Altman wants to diversify from Nvidia.

Why?

Because it's so expensive, and they have their own in-house silicon effort. I don't think they think that they know their model better than anyone else, and so they think they're the best to develop a chip that can run their model efficiently.

Max, you're back and forth to Taiwan very frequently. When we think about Taiwan and TSMC — the fact that it's in this tenuous geopolitical place, something people tend to be like 'oh okay' and then move on right away — I want to hear your perspective on whether the independence of Taiwan and the stability of TSMC is this hidden black swan event that's just kind of in plain sight that people are not paying enough attention to.

It is. For some reason the modern world decided to put all of our chip manufacturing, or most of it, next to Kim Jong-un and the rest of it is in Taiwan. I don't really understand who decided that or why, but that's the nature of the beast. Chip design companies in the US do not plan for this. The contingency plan is something along the lines of, well, we're all kind of screwed if China invades, which okay, but there's no real plan there. When I say invades, I don't necessarily mean a literal invasion. I think what's a lot more likely is some kind of soft power exchange like what happened in Hong Kong over time. Although I'm not saying Xi Jinping won't just invade — he absolutely will and he said he would — which could be catastrophic for TSMC and the modern world again. Chip companies, there really isn't much planning or consideration because people in the industry just think it's an event that's so crazy, complicated, and out of this world that nobody's going to do anything about it.

Max, do you want to talk about what happens when you try to visit TSMC?

They send me to the gift shop.

So you're in Taipei and that's where you go.

That's basically it. They have a museum where they have a bunch of chips they've made over the years. One of the Cerebras chips is in there — they're very proud of it, the dinner plate-sized one. I've never been to the museum, but that's where they always want to take me. Wired is one of the few organizations that's actually been able to go inside one of the factories and they wrote an interesting story about it. But yeah, I do not get to go to the TSMC fabs unfortunately. They're not all that interesting inside — they mostly look the same, a bunch of white boxes with robots moving around. For people who know what they're looking for, it's helpful because you can count the number of EUV machines and figure out roughly how many chips they can make. But the tours chip companies give are not particularly enlightening in general.

They are fascinating though.

Yeah, they're cool, but they're not useful in a meaningful reporting sense or from an investor's perspective.

I had gotten very excited about TSMC. Did a ton of reporting, spoke with former employees, then got the courage up and called them and said, 'I'm ready to come to Taiwan.' And they're like, 'You can come to the gift shop.' And I was like, 'All right, well, I guess I'm not coming.'

Our freelancer who got in there spent about a year negotiating with the PR team to get in.

I've been working on it for three years. So if she has any tips, I could definitely use them.

Call up Virginia Heffernan.

I will, absolutely. Lauren, you've been inside a fab in Arizona. What's it like?

This was Intel Fab 52, their most advanced fab. It's a 2-nanometer fab and it's on the level of what TSMC does in terms of 2-nanometer, but at a fraction of the scale. To Max's point about redundancy or resiliency in the event of a Taiwan invasion, it seems like right now the US government is very interested in what Intel can do in its fabs, but it's pretty small. So I went to Chandler, Arizona, saw Fab 52, got the bunny suit on. Like Max said, it's interesting to see — everything is roboticized. It's like if you had this utopian fever dream of what the future of manufacturing looked like, everything is white and everyone is dressed in all white. They call them FOUPs that carry the silicon around above head and then automatically go down to get etched and carved and stamped. It looks and feels like science fiction. From a reporting perspective, oftentimes the companies will cover up the names of the vendors on the machines. For example, there were ASML EUV machines in these giant bays the size of school buses, but then you might see spaces where they say they have two more coming. So you're like, okay, EUV is obviously very coveted technology and very expensive. You get some insights from that, but primarily it's just good context for better understanding how this all works.

I was not allowed to go to this tour, just for what it's worth.

Max, I think you're doing something wrong. Max, you got to work on your emails. Just kidding. That means you're doing actually a great job.

Thanks, Lauren. There was an interesting moment in the Dara Khosrowshahi-Jensen interview which will come up in every session today, where Jensen was trying to get across to Dara that if you do not provide chips to China, there's going to be a constraint. They will develop their own models on their own chips, they could become global standards and then put export controls on the US.

Does anybody here think that is a legitimate concern, or do you think Jensen just wants to sell chips to China?

I mean, it might be a legitimate concern in some sense, but Nvidia has run out of places to grow. They can't sell to any more countries, they can't sell to any more hyperscalers, they're going after the enterprise market which is messy and complicated. So China is a big market that has basically no access to Nvidia chips right now. If you're getting those kinds of gross margins, of course that's an obvious place to sell chips. The company's kind of obsessed with it at the moment, and it's why Jensen goes to Washington all the time and insisted on going to China. It seems pretty clear to me.

Ana, any thoughts?

Max spends a little more time thinking about that than I do. I do understand he obviously wants to sell to China — it's a huge market and they're taking a hit because they can't sell to China. You can't really answer the question without acknowledging that. But I do think from a national security perspective, Chinese companies are developing AI chips no matter what happens, so whether we can or can't, they'll continue to develop alternatives to Nvidia.

Their manufacturing process is a lot worse. One of the research firms just published an analysis of one of the new Chinese chips by SMIC, and it's good but they run up to the limit — they can't buy EUV machines. If you can't buy those things, it's fundamentally difficult to make transistors below a certain size and they can't do it effectively.

To Max's point, Nvidia only has so much more room to grow. They're at the point where they've become a major investor in CoreWeave and then CoreWeave is using that money to buy Nvidia chips. There was an earnings call where Nvidia said they had a domestic buyer of some of the chips they weren't able to sell to China, though they didn't disclose the buyer. They're basically trying to sell them wherever they can, so the motivation there is they want to sell chips to China.

No, those leftover H20s that they couldn't sell to China, somebody else wanted them because people want silicon right now so badly — any kind will do.

I think they sold about — was it $600 million worth of those?

Yeah.

Okay, so a minute left. Let's do an actual lightning round. Yes or no — do you think Bitter Lessons still apply? Basically the reason for all this infrastructure is because AI labs think the more GPUs and chips you string together, the better the models will get. Do you think it will continue to improve as these buildouts continue to explode in size, or no?

Wait, ask the question once more.

Is the investment worth it in these data centers, or are they ultimately not going to have much better models even though they have a lot more chips?

Are we going down the line? You start with me. Generally, yes.

Okay.

Not sure. I think it depends on the amortization and what ends up happening there.

Okay. Ana, you announced your news on Twitter, right?

I did, and LinkedIn.

Why don't you tell everybody where you're going next?

Oh, July 6th, I'm starting a new job at The Wall Street Journal.

All right, let's hear it for Ana. Congratulations! And thank you, Lauren and Max, for joining us as well. Let's hear it for the whole panel.

So often the story of AI's trajectory is told without the people using it. We hear about concepts called token maxing that are talked about in the abstract, and then one day you see a chart and you never actually speak to the people spending the tokens. To fully understand how this technology is progressing and its potential, we actually have to speak with the people who are building with the tokens. That's why I'm thrilled to bring on Dallas Dolan, the TMT leader at PwC, who is actually implementing this AI, looking at the costs, and making sure it's worth the money they're spending. He'll take us deep into the token maxing and budgeting conversation. Let's give it up for Dallas Dolan of PwC.

Great to see you, Dallas.

The tables have turned. You're interviewing me. This is good. What do you think about — Aaron Levie was here like 10 minutes ago and he talked about how token maxing was a BS media narrative effectively, and it never really happened. Do you agree with that?

I don't know if I totally agree with it. We were actually talking backstage just before he came on. There's absolutely commentary that's out there that we're all seeing in the mainstream media as well as what plays well on social media, on X, and in podcasts. And then there's the reality within a lot of organizations, but it's happening enough where the behaviors are slightly problematic from a cost and ROI point of view that it's real. You can't say every circumstance is a problem necessarily, but it's coming through in a way that's enough to think about and say, hey, are we doing this the right way — broadly speaking from an enterprise strategy and also from a broader ecosystem point of view.

So there's this moment now where we're actually seeing a counter to token maxing. It's called token minimizing. You're seeing companies like AT&T and Meta get really serious about cost. We also know that Uber spent their entire budget in less than half a year. Who do you think is going to win out at the end of the day, the token minimizers or the token maxers, given the definition of token maxing you just gave us?

I think here's the good news. I don't think there's a winner and loser defined by did you token max or not. The winner and loser is going to be defined by did you outcome max or not. It's in part a function of how did you incentivize people, which goes into that leaderboard and the things that got a lot of folks in trouble or in the news. How do you encourage people to do it without encouraging the wrong things, take it too far, or spend too much money? The other part is from a planning point of view within an organization. I sit on the boards for our US and global organization at PwC and we talk about this a lot — even contextually from a comparative point of view within different industries. It's not just saying does one organization spend more or less than another, but how are we compared in spending against a tech company with a bunch of engineers doing coding. What we're looking for is a baseline — benchmarking within the industry, outside the industry, against a given benchmark and saying, are we actually getting an ROI? Are we way outside the bounds in terms of what we think or know the spend is? Some of the companies you've named are certainly ones we're aware of and work with, so we have some decent intel on what's happening. But it is going to be about what outcomes did you get. I've seen some things people have built, for example in the deal space, that are incredible — really amazing output. It's taking thousands of hours worth of work and creating a product within seven minutes. The cost is very high, but it's an amazing output. There's other processes that people haven't even attacked yet. When do you go after each one and say how am I going to outcome maximize for each process? That's when you'll see the benefits come through.

So you're already doing ROI calculations?

Absolutely, yeah.

You're probably far ahead of most folks. How do you determine the ROI and what percentage of your projects would you say are actually generating a positive ROI?

Well, I know even from an external point of view, MIT did a recent publication talking about what sort of activities are really replaceable with generative AI, especially in the vision and human interactive space. They came up with about 23% — as in you wouldn't use a human for 23% of the work. In our business it's probably not as high as 23%, but it's going to be some percentage of every single thing that people do. It's a different mathematical equation. It's absolutely going to be measured for our business in hours. The same thing in the engineering space for a lot of these companies — it's a measurement of hours. Are you making the person more efficient? Are they creating more output for the amount of time they're spending? Is that lines of code they're producing or is it quality product? That's an interesting thing — I could build a larger slide deck or write way more lines of code, but am I actually getting a product that people are willing to use? I had that conversation at Tech Week in New York with founders and investors from Andreessen and others talking about the types of companies they're investing in. If you're an investor, you say, 'What does your product do for the person using it?' If you're not indicating you're able to help them actually build something better that their customer wants to use, the fact that it produces more lines of code or a longer slide deck is not better. I think we're going to come to a determination soon where we say, 'Wait, what am I really trying to get here — more, more, more, or actually a better product?'

But that goes to how the foundational labs and cloud players are working with you. The things I've heard are that these companies are making it so when you plug into their systems, you're going to spend a lot of tokens and they don't really want you to be able to measure them. Are you finding that?

The big counter to the marketing and sales approach there is in the control plane that's being used to actually help companies make decisions on what model's being used or what interface is being used to do specific activities. This is where the concept of centralized planning — I say that as someone who's traveling to China next week — within an enterprise is going to be so important. You are not allowed to check the weather five times a day using Claude 5.7 or whatever it might be. It's unacceptable. It's not a good use case for weather checking, even if you're worried about tornadoes. You can do that with a weather app or something inexpensive. The control plane concept is relatively new. It's got governance elements, cost control elements, access to data elements, and human interaction elements. That is going to be the surface where people spend a lot of time and energy at the AT&Ts and the Metas and the services companies. Quick analogy: we are simply not best suited driving a Lamborghini to go pick up milk. Those two things just don't align, unless you live in South Beach. That's not what we'd want to encourage. I think that should eventually proliferate to everywhere, especially as costs have gone up.

The last time we got together a few months ago was right at that click point of cost, and you did a great thing with the audience talking about how much people are willing to spend. Can we put the house lights up for a second? I am curious. Would you be willing to spend double the amount you're spending today for the current capabilities you have with AI?

That's it.

How many people are satisfied with the price you're paying? So, let's say your Claude subscription doubled in price, how many people would cancel?

Is it Fable or not?

Is it Fable? Well, right now, no. I think basically what it proves — we saw about half the room's hands go up that they would pay double. It does seem like there's a bunch of room for the labs to be able to raise prices and still do well. Dallas, something interesting happened since we spoke last. We spoke at Google Cloud Next about this, and someone asked whether the margin of the business can be maintained at the current prices. I thought, forget about it, because these labs have such an economically valuable tool that they'll raise prices. But we might be in the moment where they're going to get into a price war because OpenAI is rumored to be potentially dropping prices. What do you think about that?

I think we're absolutely on the precipice of that. I think you actually saw in the audience — we saw every hand go up when you said 'Would you be willing to pay double' three months ago in April. And when Alex asked if people were willing to pay four and five times as much, there was still a quarter of the hands up. We're talking about a room of roughly 200 people. I contrast that to what we just saw now, and there's more skepticism of value they're getting from it, especially when you start layering on access and capabilities associated with models that are still per seat, as well as open models which you can get access to for free. What would you do if models were half price or tokens were half price?

What would you do differently?

What would I do differently? Yeah.

Let's say I'm OpenAI or Google and I come to you and say, 'Dallas, we're happy you're using our technology. We want to make sure you don't use Fable when it's back online. We're cutting your prices by 50%.' Would it change anything you're doing?

Absolutely. I know this because I talked to my CEO and CIO yesterday and my CIO again this morning. We're an enterprise-sensitive environment with 350,000 people globally, and they all have access to one tool or another. When you have 350,000 people playing with tools, there's a high degree of price sensitivity on sheer volume alone. They're all pretty smart people doing interesting things with use cases they want to play with. We're encouraging them to do that, but we're also price sensitive. There's an elasticity — the higher the price goes up, the less we'd want them to use that model. We'd want them to use something cheaper. If somebody was coming to us with a cheaper model, 100% the direction of travel would go there. We've even built that into some of our control plane technology — selectivity of model and recommendations. Depending on who you are and what we think you're going to do, it automatically configures to have a specific model and you have to break the initial setting to use something different. We're already there, but I'm certain we would encourage more usage of cheaper things. There's no doubt.

Have you gotten any of those type of calls yet or are you waiting for them?

I won't go too far into it, but there's definitely conversations going right now. Yeah.

Very interesting.

When we spoke recently, you told me that you're seeing some limits with agents. And it seems to me like if we're going to see this continue, agents can't be limited. They have to be able to operate autonomously and spend all those tokens and be effective. So, what are the limits that you're seeing with agents today? And do you think that if we could extrapolate a little bit, it means that we're going to see some more speed bumps as the labs try to roll this technology out further?

I don't know if you use the term limit, but I think it's a function of both risk tolerance as well as cost tolerances, and then finally, what expectations we have of these things doing on their own. The limitations are in all three areas. From a risk tolerance point of view, people are worried that the agent without some level of control or governance could go just about anywhere, and what that means within an organization depending on access to data, whether client data or code itself, and what it can do to change code if asked to do one thing in one area – will it think it needs to do that everywhere else? The ability to extrapolate on a single point and what control exists there. That's the first limit. The second limit is on cost variance: there are things you're not going to want it to do because humans can do better and more cheaply, especially with high-premium models. The third bit is decision-making from an organizational point of view. Even this morning, I was at a funeral for my grandmother born in San Francisco, and I was talking to a priest who wrote a paper with the team at Anthropic. We had an interesting conversation on where ethics play into the broader conversation in this area. It's like workforce planning and more. There's an additional component about the benefit side that we haven't really gotten into, which is do I want to scare all my people that I'm watching what they're doing and waiting to replace them? The answer is absolutely not, because I met six interns on the plane back from Chicago last night, and they couldn't be happier. They saw something on my t-shirt that said...

You're still hiring interns?

Absolutely. We're hiring as many people as we did last year and we're changing who we're hiring. We're actually hiring a lot more kids who are pursuing sciences and even in areas that go beyond engineering.

And where are you not hiring?

I think we're hiring a little bit less in pure accounting, also a function of where we're doing it. It's not that we're hiring less people overall; we're hiring less people in certain areas, but that's also just a function of globalization and how we're delivering services. It's not really a function of the fact that we're saying we don't need these people because technology will do a lot of it. That goes back to your original premise of the question: what are the limits on the agents themselves. The limits are going to be a tolerance for error, a tolerance for cost, and a tolerance for what will be acceptable within the organization of having that thing do versus having a human do it. There'll be a cost dynamic to that part as well. But I think there's also even just how do I run my business and how am I a positive leader within a given community, regardless of the type of company.

You know, Dallas, I want to take a moment just to acknowledge your grandmother, and I'm sorry about her passing.

I appreciate that.

She spent her life here in San Francisco.

Here in San Francisco. Yeah.

Can you tell us a little bit about her just briefly?

Sure. She was the daughter of an immigrant family. People who picked a lot of things in a lot of places around the world, including Argentina and Hawaii. They thought picking stuff out of the ground isn't a good deal, so they moved to San Francisco and worked in the canneries up in Fisherman's Wharf, which is what her parents did. She actually worked in the telecom industry. Her first job was working for Bell. It must be the second T in TMT, right? So for Grandma Fernandez Corteho. Yeah, really a cool existence and just a great person and a great full-circle story, right? This whole story of technology is such a cool thing that all this is happening in one place here in the Bay Area in San Francisco. It's really neat to see that thread all the way pulled through, like I've been able to do from a career point of view, and also be able to notice that there's actually a direct connection to the things that made this place a great place 150 years ago, or actually still the things that make it a great place because it's still better to work in a cannery or work for a company than it is to pick something out of the ground, as my ancestors concluded back in the 1800s. When you think about that, it does go back even to some of the things of what do you want to do, right? I think you want to create technology that makes people's lives better, in much the same way that our grandmas make our lives better. Grandma makes everything better. I think it's the same kind of concept. The things that we talk about every day are how do we build products, what are the entrepreneurs trying to build products that make their customers' life better or make something easier to do, whether it's coding or customer support or what have you. I think we take that thread all the way through. That's actually where the ROI comes in. It's not going to be back to token maxing, how much you pay and how you pay it and what have you. It's a whole ecosystem shift in the way that we think about the outputs themselves. The outputs are still: am I serving my clients well? Are they getting deal maximization out of what they're doing with any given company, frontier firm, or hyperscaler, or Neocloud, or whatever it might be.

Well, thank you for sharing with us.

Yeah, thanks for that. You know, interesting San Francisco. One of the things that I associate with San Francisco is it's a city where people are okay losing a little bit of control, whether that is building new technologies that do things a little bit differently than previous generations or the fact that this city, or not everybody but many people, love LSD and mushrooms. And it is interesting with agents, you do kind of lose control. It reminds me a little bit of skydiving in a way where you jump out of a plane and you're like whatever happens but I hope there's a system ready to catch me. From your position, you're deploying agents in pretty high-stakes moments. Even if you have the best governance in place, you have to be okay to a degree of losing control. So, how do you become comfortable with that? That's funny, right? I mean, how much can you really control in the engineering space where you have people doing the code for you, or in our space where you have individuals doing the services, they're preparing an audit or a tax return or a deal report or what have you. We're putting trust in these folks, many of whom you know at a superficial level. When you start layering technology into it, because my view is that it's an augmentation of those people to make them better, not necessarily ceding all the control from them and putting into something that I know less of, that makes me feel a lot more comfortable. If you start getting into the space where 100% of all the things we do, like booking travel, becomes AI-fied, you just put in the query and the travel gets booked, I think that's where you do get uncomfortable. I appreciated that two nights ago knowing I had to get back to hang out with you, Alex, and to make grandma's funeral this morning. I can ping my EA late at night and say hey, I need some help, I need to make sure with tornadoes coming through Chicago there's going to be at least a plane on Thursday morning that takes off at 6 am that could get me to the West Coast. If I'm pushing that into an agent and I'm hopeful that it works and I'm hopeful that it books a flight, yeah, I might see the outcome, but isn't it neat like having the human on the other line like Liz says, 'Hey, I got you, bro. Like you're handled.' She's augmented. She goes into our technology and can quickly query something and boom, pulls everything up and she books a flight within like 30 seconds, right? But having her there to make that a little bit better does make the whole thing feel better. And candidly, by the way, as it relates to what my EA does or my chief of staff and others, they're able to do multiple people's jobs. If you think about what it was before, one person to one person, no, it's one to like 12. We're doing a lot more with a lot less. So I look at this tech as just being an extrapolation on that point. Yeah, there's some things you're just not going to stop doing, but that's totally okay. It's no different than driving, right? It's the self-driving dynamic: we are going to get really comfortable with that. I see this move to agentic as being very similar in a parallel run with that itself. It just so happens to be in the physical space.

Yeah, it definitely reminds me a lot of a Waymo. You sort of white knuckle it in the beginning and then you start to go on your phone.

Exactly.

That's what I've done at least.

Yeah, whole thing. Anyway, don't take anything away from those previous comments about the mushrooms.

That's for later. That's happy hour.

Please join us on the roof at 5.

Exactly.

Dallas, thank you so much for being here with us. Always a pleasure to speak with you and I do appreciate your support and PWC support of this event. So, thank you. Thank you very much. Let's hear it for Dallas. Thank you.

Thank you so much.

All right. Amazing. Thank you, Dallas. All right, folks. So, we are going to have this next session. It's going to be an audience participation section. So, we definitely want your questions for us and then we're going to go to a coffee break. So, for many of you, he needs no introduction. Let me introduce Ron Roy. I first started reading Ron Johny's market writing in 2021. He had written this newsletter called Margins and I thought it was a terrific newsletter. I saved it. I spent my winter break reading it. I DM'd him. By that January we had decided to do an emergency podcast about a crazy financial situation and then Ranj and I kept talking more and more. By January 2023 I wrote to him and said, 'Hey, don't you want to just come and do this every week?' And lucky for me and lucky for us with anyone involved with Big Technology Podcast, Ron John said yes. And so getting a chance to speak with Ron John every Friday is an absolute joy. It's definitely one of the highlights of my week. And today we're thrilled to be able to do our Friday show live here with you with your audience questions. And then we'll just run it tomorrow like a normal podcast. So I hope you're ready. We definitely need your participation. And please join me in welcoming Ron Roy.

I got them on. I'm going to take these off though. We'll get into the Snap spectacles. This is a medium risk maneuver because I have this microphone on. So those are the Snapchat spectacles. They're the original developer beta edition though. So, they're not the new ones, but I had them in 2021. Now, everyone knows how cool you are. All right. Should I start the I'm as cool as Evan Spiegel. That's right. Yeah. Is he still cool, though? All right. Let's do it. How would I start it? I throw Did I throw you off? Yeah. No, no, no. I didn't even write this down. Okay. Well, I'll try to do it. All right. Snapchat comes out with new spectacles and we take your audience questions. That's coming up right after this on a Big Technology Podcast Friday edition recorded on Thursday. Welcome to Big Technology Podcast Friday edition where we break down the news in our traditional coolheaded and nuanced format. We're joined as always by Ranjan Ry who is here with us in not in studio live with the big technology AI summit audience. Audience, let's hear you. The way this is going to work is Ranj and I will break down one story and then we definitely welcome your questions, your prompts, your arguments with us. If you don't have anything, we have plenty to do, but we'd love to have your participation. You can line up at either of the mics there to ask us a question. Let's start with our top story this week. I promised myself when this summit was being planned that I would not be mean to Snapchat. However, Snapchat has left me with no choice. John, let's roll. Image C, please. Listeners, if you are listening on the podcast, you'll see what we're looking at at the audience here is a image of Evan Spiegel on CNBC wearing the latest Snap Specs. The headline is Snap Stock Falls after AR Specs debut. Almost 20 years since the launch of the iPhone, people are ready to think about computing differently, Spiegel said in an interview with CNBC. The market reacted differently. Rajan, let me just go to you quickly. Do you think that that, you know, Snapchat and Meta and all these other companies have been trying to build these AI device future for years and this is what we're looking at. This is what we're looking at. You kind of ruined my surprise. Can we go to D, please? I mean, I brought them. I had to bring them after sending Alex a photo. Is it time for us to finally accept that we're not going to have an AR device? Interesting. So, so these again, I got these in 2021. It's actually a very cool technology. Like, it's No, I'm serious. Augmented reality. If you ever used Magic Leap in the 2010s, like being able to paint throughout your room and walk around that painting, being able to play games where you're chasing zombies again, my seven-year-old son actually is probably the only fan of Snap Spectacles in the world right now. He still loves using them. He also likes to watch YouTube on them, which is kind of amazing. But I think that form factor and the experience is amazing. Vision Pro has not quite captured it. I don't think what we just saw on Evan Spiegel's face is going to capture it. I think eventually maybe Apple or someone will, but I don't think we're there yet, but I think we will be. I still am betting on AR eyeglasses of some sort. Okay, I'm going to go to questions in a moment if anyone has one. So, feel free to stand up there if you want, otherwise we can keep doing this. Here's some of the social media reaction. Does anyone that works at Snapchat have the guts to tell leadership that these things are ugly? That feeling when your glasses are so heavy they give you cauliflower ear. Snap is the best brand in the world when you're 16 and the worst brand to be associated with when you're 21. The people who actually buy $2,000 AI glasses aren't teenagers. If you think you really want to wear Always on camera around in public, it should have to look like this with the picture of Evan Spiegel. Let me make the case that this it's over. Good because I'm going to take the other side. I think that the iPhone series that released that we just saw. So, does anyone here listen to Ron John on Fridays? Do we have listeners? All right. Okay. So, you'll know which direction this might go. What's this guy been begging for since he came on the show the first time? Better Siri, better Siri. I think they actually did it. Like the new Siri, if you look at the videos, looks terrific. And so maybe this idea of we have to wear the computing on our face is something that like kind of sounds good in concept, but model after model, it's not. And the AI device is the iPhone. Okay, that's I I that is an interesting direction to take it. I still think the form factor. Do any of the audience have like meta ray bands or any other device like that? I see a few. Like you start to feel as you're walking around, as you're kind of interacting in the real world, stuff can happen that's not just maybe eventually Siri talking to you and using a traditional AI model to actually give you kind of information. So, I don't know. I still think AR as a form factor via glasses is going to happen. I think you have met Ray-B bands, you enjoy them. You know, I didn't want to say this publicly, but I have not used them. I mean, I have, but I don't I used them for a bit. And like I said on our show recently, I was on a hike. It was cold. I was ready to get to the summit and put those glasses on and the battery started blinking red and I couldn't use them. Had to use my old phone. And let me say this, like, you know, they haven't taken off as a mainstream consumer device. And if you look at the stock of every single company that's pursuing them, it's not good. Snapchat, like we just said, is struggling. Meta, as we know, has its problems. No one's looking at the these Ray-B bands to save Meta. Well, no, but to me, the Apple Vision Pro is like the more direct correlated product versus You want to know what the best thing about the Vision Pro was? Find Vision Pro. Yeah. They put the person on Vision Pro on Siri and he fixed it. Mike Rockwell. Oh, so I would not have guessed that the person who made the Vision Pro would be the one to fix Siri after all these years, but I guess if that's the case, that's exciting. Yeah. Yeah. So, you're going to still Can you put those glasses on one more time? Again, I was told backstage this might destroy my microphone, but I'm going to try. Do I look cool? No. No. No. No. I mean, even we were joking that Evan Spiegel is going to the Met Gala with his model wife. Like, this is like the coolest person in the world and how bad they made him look. That if it was like when Zuck wore Project Orion, no one really cared that much. But I think because Evan Spiegel is such a cool looking dude, that's why it looks so egregious. That's my take. So, your answer on the way to make those things work is just be less handsome. That's it. That's it. We'll write to Spiegel and let him know. No questions. Okay. All right. Great. All right. Yeah. Don't be shy. Is it on? Yes, it is on. Hey guys, if you're willing, let us know who you are. And this is Ser Johal. Actually, industry analyst flew from New York a day earlier to join you guys. So, welcome. Let's give him a welcome. Thanks for ask question about actually it's an observation. And I want to get your take on this. When companies have this gap between what people need today and what they're working on which is like a in future it can be two plus years out. So they lose that traction from the investors point of view as well as employees and partners. So do you see that that's happening to meta and others like it happened to IBM when they were living in future with Watson X for example. So what's your take on that? It can be B TOC or B2B examples. Okay, thank you for the question. Yeah, no that's a great question. I do think like if we're talking about kind of how the interface for how you interact with a computer. I have been begging for something else other than me holding my phone looking at it and we haven't really gotten anything for a long time. I mean we Humane tried with the pin. I still think maybe some kind of pin is going to be around Johnny Ives pin at OpenAI. Maybe maybe at some point. Well, Greg Greg Brockman is coming, so we'll put we got to ask him about the talk about that. But, but I think like being able to interact with all of this information now, being able to like process information so much more reliably with AI just I don't want to have to keep looking at my phone and just looking at it on there. But even actually on the phone, I'm guessing do a lot of people here dictate more to their phone right now. Like that's completely changed. In the past, like I would have felt weird just talking to my phone and now I'm constantly using whisper flow and just talking to whatever. So that's tell a story about you and your wife. I know this. Okay. I don't know if this is the most depressing thing or and my wife is not here and hopefully won't kill me if this is being live streamed right now, but No, we won't broadcast this. So, I am constantly dictating to my computer, to my phone. And the other day, it was like Friday night, we had put our son to bed. We're both on our laptops and I'm on one side of the couch dictating. And then I look over and she's also has her laptop open and is kind of whispering to her computer, too. And I'm like, is this the future of tech? But it's a new computing interface. So, I'm happy about it. Right. I look, I think the gentleman brings up a great question, which is that things are moving so fast. How do you plan right now? And honestly, I don't know how you do it because every day there seems to be a new capability, then the capability is taken off the table. Like Fable, for instance, I mean, the tweets about Fable where like people are adding Dave Sachs and they're like, 'Please, I'll do anything for Fable back. Just bring it back.' And he won't bring it back. But it's just like I don't understand how any company does that. And I think that actually would be a good topic for us to sort of get into on a future show. Yep. Okay, let's go this way. Hi, thank you for taking my question. I was curious your views on in the next like five years as AR glasses evolve whether like the chunky spectacles is the way to go or thinner glasses with like upboard compute like either wireless to your iPhone or like the vision pro like cable down to the battery pack and compute. Well, it's a battery for Vision Pro, but it could also have compute on board. So, I'm curious like the next five years where you think consumers will gravitate towards. No, it's a great question because if you ever use Magic Leap, there was like a puck. Puck. That's what it was. My Here's my hot take. Anything any device that requires a puck not working. Well, how about this though? Okay. What if what if the puck is your iPhone? Oh fuck. All right. Exactly. So, which gives Apple an opportunity that like if the compute is taking place on the phone in your pocket, it allows you to be much slimmer from a power perspective. you don't want like a lightning cable connected to your face. But like it's still I think there is a it makes me think Apple still has a good chance in this space because the iPhone can do all of the heavy lifting versus this thing on that is really heavy on your head. Can I can I ask so what's your name? I'm Kyle. Kyle, so what do you want to use a like face computer for? Oh that's a good question. See I told you Ron John this stuff is not happening. No, no, no. Kyle's got something. All right, let's give Kyle an opportunity here. Let me think for I didn't mean to put you on the spot, but thank you for helping me prove my point. Oh, I think one cool thing would be like a shared TV or something like like imagine Vision Pro and you have just a shared just like movie theater. You're on a plane with your family or something and you just have like the shared experience, but it's just glasses. I like that. I like that. Like for me, one of the coolest things I always like cuz you know, as you get older, you move further away from your friends and it would be cool to like be in the vision pro together. Yeah. Sit sit half court, watch the Knicks next to Timothy Chalamay, but we're just in division pro. But you know what's interesting? Apple never advertised that as a social device. All the marketing was just you're sitting alone at home and all you're doing is the Vision Pro. Nobody else exists. The other big one. Yeah, sorry to interrupt. I was say the other big one for me is just having a lot of monitors around when I'm working. Like I even though I have like an ultra wide or like three monitors, feel like sometimes I could have more. So just being able to like interact with AI to pull up the exact page I'm looking for out of like I'm one of those people has like a thousand tabs. So just being able to pull up the thing and I just say open up that tab and it just pops up would be pretty cool. Would 8,000 individual tabs in a giant planner space be better or Kyle's going with this. No, no, but I do agree like that being able to do more open scaled work and like look at different charts and I think there's that still and I know like I have friends who own the vision pro who use it for that still. No, but did they had a nice time? All right, thank you Kyle. Let's go here. Hey, what's up? This is Sasha from Yo University doing research on AI agents for finance. First of all, love you guys. Listening to you guys chat on my way to work has been my routine. I think this is excitement shed by not just me but many people in the audience. So, thank you so much. Thank you. Thank you. Appreciate that. And thank you for coming all the Did you fly in? Yes. Oh my god. Thank you for coming. It's great to see you. It's worth it. And I appreciate your sharp line of reasoning and questioning. So what's your view on the meter benchmarks of how long the AI and AI agents can work independently being saturated? Are we at that point of them being at infinite work number of hours working yet or are we reaching that soon and what is the world going to be looking like after we hit that infinite mark? That is an excellent question. So, first of all, there is some controversy about the meter monitoring, but I think it's kind of directionally accurate, right? Like I showed the meter chart at the very beginning of our event today that like you saw these models, they could not code autonomously for more than 30 minutes in 2023 and now they can code autonomously for the equivalent of 18 hours, right? So what happens if they just kind of blow past that limit and then they can code all the time? That is an excellent question. Do you have any thoughts? So I felt that with the go mode and automation mode it practically felt that they are already doing this autonomously forever. Right. I want to want them to send me reports. My prompt would be if you find something interesting send me an email. Yes. Then I when I see it I will come back and intervene. I felt that for many tasks it's already at that mark. Yeah. No a perplexity computer will basically do that. So that's definitely something that's happening. But I'll just say one more thing and then we'll go to Ron John. Greg Brockman who will be here later has this like idea of a compute powered economy where like you know you sort of once you get these bots working the way that you explained you just maybe he'll say this later you just kind of throw them at any problem and then they just kind of work autonomously through it. Now we don't we've never seen what the world looks like when you can do something like that but I do think given the progress that we've seen with the models you would imagine that something substantial will come at a certain point. So my day job I work at a company writer. We're an enterprise AI company and the when and the goal mode was asked about recently again the idea that you just provide the goal. The agent will iteratively loop and keep doing work until it finds the right solution. In the real world that is like the benchmarks versus the real world. I still think there's such a massive gap in terms of what does the data look like? What is the actual problem? Does the customer or the person actually understand the goal in a clear enough way that they're able to define it to kind of push that loop forward? So I think it's an interesting I think with a lot of AI again even around the benchmarking like again versus real world understanding of how to use it, what's happening and what's available for it to use. There's still a lot of work to be done. I mean you can do these goal modes for like less complicated tasks. I like what have you goal modeed? So I worked with Perplexity Computer recently to to try to like have it find a hotel discount for me. No travel stories. I've been on a this has been for like two years. Every time I remember like Sundar actually it was I don't know if people remember in 2019 my example. Wait, wait, no, no. I'm just saying. Why does everyone when they talk about Agentic talk about travel? I get cuz travel is such a pain in the ass. But it's I don't Okay, go on. Go on. I'm I'm I'm sheepish now. No, no, no. Hotel discount. Hotel discount. I just looked at the hotel every hour and when it dropped behind us below a certain threshold, emailed me. That's pretty good. Okay, that's a clear goal. I will give you that as a clear goal. This is I respect your I respect I think you're right that we definitely need better examples than It's a pet peeve. It's just for some reason every Apple the original ridiculous Bella Ramsey commercials around Apple intelligence of core everything is flight booking and again like you have killed that Bella Ramsey commercial so often they're going to the creative agency is just going to write to you at some point and I love the Last of Us and I love Bella Ramsey but those commercials still irked me. Yeah. What happened in those commercials? uh one of them she's sitting at someone that she can't remember who that person is and in real time asks Apple intelligence like tell me about this person my interactions which again is just such a weird thing like I'm so much better than you that I don't know who you are and I need to remember who you are and have AI tell me but and it didn't even come close to working with Apple intelligence so that was the worst one you know what would be good for that actually what goal mode AR glasses oh AR glasses is there's the real world use case or the fact that that commercial was so bad just again proves my point that okay AR glasses aren't going to work. Okay, sorry. Thank you so much for the question. Thank you and we'll go over here. Oh, he was okay fine. We'll go here. Okay. Hi. hey Gerald Harris. I'm on the board of the Commonwealth Club here and I run some programs for the club and I have the small scenario planning consultancy. But here's my question that I think hasn't come up here. What what what should we be concerned about in terms of using the AI models them building a database on us and then turning that into advertising? So the advertising potential revenue from some from users, do you think the AI companies will ever go after that revenue or use that information for advertising purposes? Oo, that's a good one. So, will will AIS build an amazing sort of profile of you based off of all the personal data and then use that for ads? Most certainly. So, someone left one of the companies and wrote about that in the New York Times about three months ago. No, definitely. Yeah, that that's coming. I think that advertising is obviously going to be one use where we're going to start to see some of the problematic stuff here, but actually if you look at the ads that OpenAI gave you, it's sort of like more of a and obviously they always come out with the hightouch brand and then they will like get you on the direct response like super targeted stuff once they realize they can't make money on brand. But like it looks pretty good. Like if you're sorry to go with the travel example again, but if you're like researching travel and chat GPT, like you can go into an advertising chat experience and it will help you. But I think that like we have never had technology ever that collecting this much information about us. Has anyone here like talk to like chat GPT or Claude and say, 'Can you psychoanalyze me if you're not or give me any any insight about myself? Yeah, just like five of you. The rest of you have done it. Admit it. It's scary and and we're g and that's the gated stuff and we're giving this all this information to companies and it's a trust thing that we don't really know where it's going to go. Yeah, I think it's interesting because again, OpenAI originally ads are going to be a big part of the business. Now they've pivoted away from that a bit but are still releasing ad products. At least Anthropic is not I mean I don't think they're going in that direction at all. Google and Gemini obviously 100% will and like you probably will get a really good ad experience. I mean the more it knows about you the types of questions you're asking but it is like you know like thousands and tens of thousands of three to five word Google searches are a lot different than entire thought partnership therapy like research exploration. these models will know everything about you and it is it's terrifying. Have you done the psychoanalyze me prompt? No, but I did so I use like Claude and Chat GPT Gemini. I'll kind of go through the three of them and cycle around. So it is funny. So I have asked like what do you know about me? Like what would you ask back? And they all have very different kind of like jagged jarring aggregations of who I am. Like I don't know. They're I I I don't have one that just knows me through and through. I I've done it. I diversify them. I've done it with chat GPT. Yeah, it's pretty good. Yeah. And then like it will first give you a answer that's like kind of sanitized and then you say go a little deeper and then you say get a little darker. Oh dear. I have not redteamed chat TV. I challenge you all to give it a shot. Don't go darker. Don't go darker. But if you really want to. Okay. Thank you over here. And thank you for the question. Let's move on. Let's move on. Yeah. Thank you guys and really appreciate all the work that you do on the podcast and YouTube. Your conversations go so deep and still very broad. So I really appreciate that. Thank you. We appreciate it. You're a listener or viewer. Oh yeah. All right. Both. Thank you so much. Depends on where I am. Amazing. Appreciate it. Um and you know my name is Si. I am a adviser for startups in the AI space. And I've been in product management and go to market strategy. My question for you somewhat related to the gentleman before you know there's a lot of coverage on agents and what agents can do and a little bit about the governance and cost control. What I don't see a lot about is you know agents do drift probably why we don't have better examples than travel. And over the a period of time they don't get better at that specific understanding of the people and the business context and the operating model they're working in and learn from it and I was wondering is there is there a lot of conversation that you guys are seeing around the learning of or the learning layer in the agents and is it too early or for that to happen And the conversation is mostly around governance and cost and access. It's actually a good question for Ronan. Yeah. So again, like working in this all day and thank you for the question. I think what's going to happen and I'm already starting to see is like 6 to 12 months ago the approach was just jam as much information into whatever system you can. And we were kind of promised that it would just work and AI is just going to and like there's you know you'd get this feeling that oh wait a 100page PDF it actually could parse and I can get information out of but then when you're doing that at any kind of agentic scale it doesn't and I think already like how information gets chunked up and like distributed in different ways and context windows everyone would see like 1 million token context window but we didn't really know what it meant and now more and more like and I actually think this is going to be one of the most interesting like professional opportunities and areas to be an expert in going forward is like being able to kind of it's funny like it is as much art as science right now. But I think like the more you can start to grab a hold of how information flows through these agents and how you actually get it reliable, it's going to be a really really interesting space. It's no one job right now. Thank you. Okay, thank you for the question. All right, we have seven minutes, three questions. Let's see if we can keep the questions brief and we'll keep the answers brief and we'll get to everybody. I'll try to be quick. Hey everyone, big fan of the pod by the way. Flew Chris Auei. I'm one of the co-founders of a company called Virus Watcher. I'm here with my colleague. Flew in from Dallas. He flew in from Austin. Yes. A big fan. We got a couple Dallas folks here today. Oh, who's Dallas? Shout out Dallas. There we go. Okay. Go Matt. All right. Oh, yeah. We're big in Texas. I'm going to go in a different direction with my question is I want to know kind of what y'all's thoughts and honestly if you can ask this kind of later on with Greg Brockman and others what is the thought on using AI models and AI technology for biological intelligence bioveurveillance and public health for emerging risk detection with infectious diseases is a very interesting project that we're working on we have a lot of epidemiologists on our team we were just at the UN two weeks ago with the World Health Organization and trying to discuss these problems and how we can use the technology to detect these things early on. So I kind of want to get you guys thoughts on that. Can I ask you a question back and if you could give a brief answer that would be great. Okay. Are you a believer in the

And then I came back to the work being complete. So that level of this is a big sort of chunky task.

So that was so you're basically saying it was faster. Did it in an hour. You're guessing compared to with Fable compared to what it would have been before.

I think the main difference is that in the past, it would be like, 'Great, I did it,' and you'd be like, 'But did you take a shortcut here?' or 'This is not quite right,' or 'I need to go verify it,' or 'Oh, you cut this corner.'

It's like the managing interns thing that everyone's been saying for the past year, which is very offensive to interns, by the way, but yes.

Yeah, exactly.

I don't know, have you managed an intern?

Yeah, that's true. And you're saying it was more correct. So it's faster, more accurate, more reliable, and then according to the US administration, dangerous. I think the other piece is it has a greater theory of project, so that it's less like, 'Oh, I'm going to make this change,' and it'll say, 'Great, I'll make this change,' but really, especially if you've done software engineering at scale, the best engineers keep in mind all the disparate parts of how this works and they also see around the corners, like, 'I can make this change, but if I don't do it in this way, then the next change is going to be incrementally harder.' And I think that's been a significant difference I've seen in that class of models.

So, I think when we talk about Anthropic Labs, right, people think of Claude Code because it is really your breakout product and it sounds like you've been tasked with basically figuring out what the next Claude code is. Would you say that's an accurate description of what you're doing at Labs and also why does Anthropic need labs?

Labs. Yeah, it's also maybe worth thinking about why we needed labs in 2024 when I arrived and why we need labs today because the answer shifts. I started the labs team with Ben Mann, a co-founder of Anthropic, in my third week. At the time, our product engineering team was 25 people, and while we had Claude and Opus 3, they weren't powerful enough. Labs was created to ensure products don't fall behind the model exponential. Claude Code came out of this because coding wasn't being addressed elsewhere. We use two thought exercises: visualize the gap between model capabilities and current usage, and imagine what models will be good at in six months to prepare products. Computer use was initially bad but served as a beacon. Now, with a driving product team and Claude Code's growth, models are advancing quickly, so interaction capabilities need to evolve. For example, Claude Code artifacts allow visual outputs. We aim to make AI more accessible to non-technical users.

But Mike, so it puts people using Anthropic models in an interesting place, right? Cursor just sold for 60 billion to SpaceX, and there's a meme that Cursor would have sold for 300 billion if not for Boris Cherny, who created Claude Code. So for companies building on Anthropic technology, they'll wonder: do I partner with Anthropic, or will Anthropic build the product I want to build?

Yes. I think the agentic coding side and being both a platform and product is interesting. When we take on projects, the goal is often to push that industry forward. For example, AI coding editors existed, but nobody thought as freely as with Claude Code, so now more products have that flavor. If we're entering an industry just doing the same thing with the Anthropic brand, that's a bad use of our time. We should go where we think the direction is and build products that create new space or show the way for others.

It would almost be like working for a tech company that has social, messaging, video.

Right, Mike, yeah, okay, you did leave. Well, there was a question for example when Anthropic launched a product that was seen as competitive to Figma and you had been on the Figma board prior to that and I think you stepped down, is that correct?

Yeah. It's a good question that Alex has brought up. Silicon Valley is known for its healthy, vibrant, risk-tolerant startup ecosystem, and when big companies come in with tons of venture capital and resources, people say, 'Wait, are they just going to steal my idea?'

Yeah, our dual existence is something other companies have navigated, like Amazon, which is both infrastructure provider and e-commerce. Customers can live in that dual world. I always try to approach with transparency. The Cursor example is interesting; Michael and I talked about where things are heading. It's about transparency and shared building blocks. We're building on the same capabilities available elsewhere. For healthcare, we shipped plugins and skills, not just our own product, to allow others to incorporate it.

Speaking of startups, Anthropic is still technically a startup, but you're worth a lot of money. I mean, what's the latest valuation? Is it 965 billion dollars or something like that?

Yeah. And so it's a good question that Alex has brought up. Silicon Valley is known for its healthy, vibrant, risk-tolerant startup ecosystem, and when big companies come in with tons of venture capital and resources, people say, 'Wait, are they just going to steal my idea?'

You sold Instagram for a billion, right? Startup in 2010. Right. Financials have changed quite a bit since then. And yet Anthropic has positioned itself as a more ethical company around building AI. How do you see Anthropic's culture dictating this next era?

Yeah, that's a really interesting question. I think the reason I joined in the first place is because Anthropic walks the walk and deeply believes in trying to make AI go well for humanity. That's in the water internally and has been why the company has remained cohesive. It was a surprise coming from Instagram where we talked about product 95% of the time. At Anthropic, it's a mission-driven AI company with a strong sense of why it exists. In terms of the overall impact on the Valley, I've seen a renewed interest in philanthropy and a conversation around how AI should go is happening in real time, which is a good thing.

Interesting. Mike, you talked about the gap between model capabilities and products. With Labs, you try to get ahead of that. What are you building, and what should people be on the lookout for?

Yeah, it's a roadmap. Two themes I'm excited about: giving Claude an environment with more agency and self-knowledge, and closing the gap between how people understand their work and the day-to-day reality. For example, if you make a file with Claude, you have to manually download and drag it into projects. We're working on interoperability. We want to transform all our products by giving Claude more agency and creating environments where it can solve complex problems repeatably.

And your moonshot? What's your moonshot?

Moonshot. No, nothing in space. Although we're talking to SpaceX about space things. The labs aren't working with the compute team. Do you personally believe in data centers in space? I had a conversation with someone who sends things to space, and they were bullish. It could make sense in a few years.

When you were talking about compressing work, I thought about tokens. Are you a token maxer, and could the industry move away from tokens as a measurement?

Both interesting questions. We found not much correlation between top token users and most productive people. Glorifying maximum usage seems dangerous and gameable. We focus on model intelligence and token efficiency together. We're experimenting with outcome-based pricing, like Claude Managed Agents, where you can specify a rubric for outcomes.

John or the guys in the back, can we show the random image from the Financial Times? It shows app releases skyrocketing, but significant usage and reviews going down. Mike, is it possible that coding and releasing isn't leading to a productivity boom?

That's interesting. I think there's a parallel in app usage. It ties into consumer AI breakouts, which we haven't seen many of. Consumer products are consolidated, and data gravity makes it hard for new apps to break through. Making something people want is still hard, even with amazing models. That chart shows it's harder than ever to break through, even with faster coding.

You had 13 people at Instagram when you sold. With these tools, how many people would you have needed?

I think we could have gotten there with four to six people. Or we could have played positions better, like building Android in a week instead of a month, and maintaining iOS simultaneously. For example, I'm maintaining an iOS and Android version of a labs project, with Claude handling each.

You might get calls from your old Instagram team. Last question: you worked on a product that evolved into something ethically fraught due to harms with children. When building in Labs, how do you think about potential harms?

There are products we've prototyped or conceptualized that we decided not to ship because they'd be bad for the world. Asking that question internally makes a difference. It's a luxury to have successful core products, so we can make easy ethical decisions. Frontloading ethical thinking is valuable, and having economists consider impact is normalized now.

Mike, it's always great to speak with you. Let's do it again soon. Let's hear from Mike and Lauren. Thank you.

All right. Are you guys ready for our last conversation of the day? Are you having a good time?

All right. So in 2015, Greg Brockman, Elon Musk, and Sam Altman started a nonprofit called OpenAI to pursue artificial general intelligence. It began in Greg Brockman's living room. Most people in the Valley thought it was a side project, but fast forward 11 years, OpenAI is likely to go public at a trillion-dollar valuation, with a billion users in ChatGPT soon. They raised the largest venture capital round in history. Greg Brockman has shown clear conviction about where this technology would lead. So let's look into the future with Greg Brockman. Let's welcome Greg.

Great to see you, Greg.

Greg, this is our fourth time speaking about OpenAI's product direction. I think I'm getting it. You've moved away from the term 'super app,' but now with Codeex, browser, and ChatGPT coming together, super app might be correct. When you need to do anything, it starts with a prompt in ChatGPT, and OpenAI's technology uses your browser or computer to get it done. Is that right?

It's interesting that the interface will melt away. Many of us use ChatGPT, and it makes suggestions at the end, like health plans or travel agendas. Am I hearing you right that ChatGPT will understand your intent and take action on your behalf, like making appointments?

Just to unpack that example, one of your colleagues is chatting with ChatGPT about an event, and it suggests contacting attendees, but instead of saying 'I have to do that,' it goes into an event management tool and does it. So ChatGPT moves from conversation to action, understanding intent and accomplishing goals.

Basically what happens is the interface will take over from there once it says it's a good idea and you agree, and then hook into whatever tools you're using and then do it for you.

Yeah. And when you go back to some of the early attempts at this, there was this like move that OpenAI had to let you call an Uber within ChatGPT. And it followed a long line of companies that have tried to get you to take action within chat, but it never really took off. And the difference here might be that the chatbot can take control of your browser or take control of your computer, and then you don't necessarily have to worry about like is this plug-in going to work? It goes and accomplishes that for you by taking over your machine. So I wonder, you know, if you expect a fight from the user interfaces that we have today — aka like all the other apps, all the software — where to be truly useful, ChatGPT will have to not be blocked to be able to go out and execute these actions on behalf of a user?

So one of your colleagues shared, and I think you've talked about this too, that you've brought ChatGPT into Codex — so you can bring Codex into ChatGPT — which is basically like if we're users of ChatGPT, this experience that we talked about of not only suggesting what you might want to do next but going to do it for you, that's going to happen. And so it makes you effectively an operating system, don't you think? But not the operating system like an iOS where you would like go open up your phone and then tap different apps — it's almost as if all interaction with all apps will happen through this interface. Is that the ambition?

Yeah. Talking about a personal intelligence, it sort of — I don't — did you watch WWDC last week?

You know, I was going to ask you — well, does it mean that you'll have to create your own device? Assuming that like my concept is that you're going to have to go through Apple to get to the user. Assuming that's somewhat valid. But the answer is you already are, right? OpenAI is working on a device right now.

So kind of like having this device with you at all times and being like 'I need to get that done' and it goes and does it for you.

There's been some reports that OpenAI is working on these bidirectional voice models. I think we've talked about that in the past — that the goal is to have like an AI that you can speak with and it will be able to process that and speak back with you in a much more natural way. Can you share anything about that?

So, let's talk about model improvement briefly. There was a discussion a couple years ago that large language models were about to hit a wall. That was wrong. And something that I'm thinking about, I think we're all thinking about, is how much better can these models get? And when will the improvement stop? Any thoughts?

So then I'd love to hear your perspective — if models can basically progress much further from where they are today, let's say OpenAI builds the best model and it's the equivalent of like something with 15 PhDs with excellent emotional intelligence that doesn't complain and goes out and does stuff for you, and then the next model maker will build a less good one that has 13 PhDs and it's pretty good EQ and will still go and do things for you. So, where does the differentiation come in when you get to that level of intelligence? Because we've seen the model makers kind of move in lockstep. One makes an advance, the next one comes in and makes the advance. So they all become that smart. Is it possible to differentiate?

So, I think I'm reading you right in that your belief is maybe there's a way that everybody can scale up these models, but ultimately the company with the most compute is going to win. And you know, we spoke a couple months ago and you had mentioned that you were asked internally 'how much compute should we buy?' and you said 'all of it,' and they said 'no really, how much should we buy?' and you said 'no, buy all of it.' And OpenAI is definitely the leader in buying compute. I mean, we see the money going out — obviously a lot of money coming in through investment and now you've built a business with customers, but there's a lot of money going out. Do you ever wonder, hey, maybe we're not going to be able to pay all this money back because it's a brand new category?

Yeah. But if I may, there is a price war brewing. I mean, at least that's according to the report. The Wall Street Journal recently had a report that an upcoming OpenAI model might have significant price cuts. And so again, how can you know if it requires so much resources to serve this demand and it is growing demand, in an environment where there might be price cuts — how do you make that math work?

Are you going to cut prices?

So Satya Nadella has had some interesting tweets and interviews recently. He recently said the model is becoming a commodity and the valuable asset is a company-specific AI system that continually learns from your data. What do you think about that, and is it weird to be competing with Microsoft now?

Okay, just to go back to the Satya point one more time — he's called models a commodity, he's trying to build his own frontier intelligence, he's telling potentially your customers 'hey, you got to come work with us because we're going to help build these loops that will learn from your data.' He's got access to your IP, I think, till 2032. So, how does it make you feel to hear this coming from Satya?

GPT-5.6 is rumored to be on its way. This is just a Twitter rumor, but I'm going to read it to you. Three times cheaper than Frontier, up to 1.5 million token context, stronger agentic coding workflows. How much of that is true? What should we expect for GPT-5.6?

So everything confirmed.

That has actually been a source of problems in my personal life. Okay, so I want to end on health. You brought it up a couple times. You actually had a question in the audience about it earlier. Sometimes there's a story and you read it and you say to yourself, I know this person is speaking to the media and what they're saying sounds like maybe it's true. But there's something wrong with the story and we're not going to see more of it. And I've read a couple of those recently. One is — I think is it your friend the GitLab CEO, Sanderij? He got cancer and used all the diagnostic testing he could have, so just went out and tested like crazy and fed that data into ChatGPT with the assistance of some people who had built purpose-built applications for it and was able — I don't know if 'cure' is the right word — but to beat back the cancer to a degree. There was also this dog, Rosie the dog in Australia. The craziest story where a guy biopsied his dog which had cancer, ran the mutations across AlphaFold, and then was able to design an mRNA vaccine that he injected into the dog with the assistance of chatbots to build this thing, which ended up being able to jump over tables again and the tumor shrunk. When we think about the future of AI and health, help us sort out the truth with this question: are these a couple of outliers that made good headlines, but there was something about the story we weren't hearing? Or is this going to become standard in the future?

Let's hope. Greg, thank you so much.

Thank you. Thank you so much. Thank you. Should we do it again next year?

All right, let me give you a perspective about what we're doing for the rest of the day. First of all, if you're backstage — Calli and John, can you come out here for a second? As I bring them out, let me explain that we're going to go from here to the rooftop. We have some drinks and food ready for you guys. We also have a space directly outside, so that is a place that you can hang out. Figure out what's busier and try to spread yourself a little bit so we'll have room. John, come up here. Calli, come up here. Guys, this would not have happened without John Bocuzzi and Calli Wynn. Thank you, Calli. So, true story — I announced this show on a podcast live and I thought we would get a lot of people signing up. One person signed up, and I called both of them and I said, 'you know, it was great brainstorming this idea with you, but we're not going to do it. So we'll see you next year.' And both of them said, 'let's do it, and let's do it at the Commonwealth Club.' And we did it. So thank you guys and thank you all. Before we leave, I have to thank our sponsors at PwC, Dallas, and Arita. You guys are amazing. And thanks to the PwC crew — just a great group and amazing people to work with. And of course, thank you to the Commonwealth Club here for collaborating with us on this event. Wow. Six years after I showed up here and there was literally nobody in the audience, and we kind of social distanced and probably would have gone to jail if London Breed found out — we're all here together and it means the world to me. So thank you all. I'm getting emotional here. Thank you for helping me live the dream and let's go party.