Jun 26, 2026 · 3h 48m · big-technology
Big Technology AI Summit (full): Greg Brockman, Mike Krieger, Aaron Levie & Friends of The Podcast
Alex Kantrowitz · 53m spoken Greg Brockman · 30m spoken Mike Krieger · 27m spoken Aaron Levie · 23m spoken Dallas Dolen · 15m spoken Alex Stamos · 14m spoken Ranjan Roy · 11m spoken Lauren Goode · 11m spoken Max Cherney · 4m spoken Anissa Gardizy · 3m spoken Chris of Weihe · 1m spoken Maria Serra Roberts · 36s spoken Stage Crew / Interjection · 3s spoken
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
The Big Technology AI Summit gathers top industry leaders, researchers, and tech journalists—including Greg Brockman, Mike Krieger, and Aaron Levie—to analyze frontier model safety, enterprise agent adoption, infrastructure bottlenecks, and real-world medical breakthroughs.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Alex holds 27.8% of the talking time here. How this is scored →
Alex as informed peer 5.9 Guest teaching 5.1 Guest disagreement 2.2 Alex pushing back 3.8
speaking balance: gold is Alex, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 2:09:31 Alex Stamos denounces federal export bans as a defensive own goal
Stamos forcefully rejects both government export policy and lab PR framing, warning that restricting US model security checks directly harms defenders while Chinese open-weight models advance unimpeded.
Hardest push from Alex ▶ 3:39:38 Alex Kantrowitz confronts Greg Brockman with Satya Nadella's commodity critiqueKantrowitz refuses to let Brockman brush off competitive friction with Microsoft, specifically quoting Nadella's assertion that foundation models will commoditize while Microsoft retains OpenAI intellectual property access.
Biggest teaching moment ▶ 2:07:40 Alex Stamos educates on vulnerability discovery vs exploit generationStamos systematically breaks down the reality of AI vulnerability discovery, explaining how decades of memory-unsafe C and C++ software created a massive bug backlog that any modern model can now identify.
Alex holds their own ▶ 2:38:07 Alex Kantrowitz highlights the platform-versus-product risk for startup partnersKantrowitz directly confronts Mike Krieger with market valuations and viral industry sentiment, asking whether Anthropic's first-party products like Claude Code undermine the startups building on their platform.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Alex as informed peer | Guest teaching | Guest disagreement | Alex pushing back | Why |
|---|---|---|---|---|---|---|
| Welcome and State of AI at the Big Technology AI Summit | 0 | 0 | 0 | 0 | Opening monologue by Alex Kantrowitz introducing the Big Technology AI Summit, recounting the origins of the podcast with Blake Lemoine, and summarizing the macro state of AI capex and recent regulatory drama. Host scores are set to zero per monologue instructions. | |
| Aaron Levie on AI Frontier Models, Regulation, and the Applied Layer | 6 | 5 | 3 | 4 | Kantrowitz probes Aaron Levie with industry hypotheses including Chamath's cloud gatekeeper theory and token maxing hype. Levie responds with Occam's razor explanations regarding Amazon's security research and breaks down task-normalized unit economics. | |
| The AI Infrastructure Buildout and Hardware Supply Chain Panel | 6 | 6 | 1 | 3 | Kantrowitz guides a panel with reporters Anissa Gardizi, Max Cherney, and Lauren Goode covering gigawatt data center delays, Nvidia's inference challenges, and TSMC geopolitical vulnerabilities. The panel offers technical reporting on EUV lithography limits and memory shortages. | |
| Dallas Dolan on Enterprise AI ROI and Autonomous Agent Governance | 5 | 5 | 2 | 3 | Kantrowitz conducts a live audience poll on price elasticity and questions Dallas Dolan on enterprise token budgets and autonomous agent governance. Dolan articulates PwC's transition from raw token consumption to outcome-maximized workflow metrics. | |
| Live Big Technology Podcast and Audience Q&A with Ranjan Roy | 6 | 4 | 3 | 4 | Kantrowitz and Ranjan Roy review Snap's new Spectacles and take audience questions regarding agentic loop drift and biosurveillance. Kantrowitz argues against face computers in favor of iPhone-based intelligence while Roy and the audience push back on AR utility. | |
| Alex Stamos on Cybersecurity Realities, Mythos, and Model Export Controls | 7 | 8 | 5 | 4 | Alex Stamos provides a detailed technical briefing on Mythos, explaining that superhuman bug finding is now widespread across models due to legacy memory-unsafe code. Stamos criticizes the federal export control reaction as an unforced error that penalizes defensive security researchers. | |
| Mike Krieger on Anthropic Labs, Fable Restrictions, and Safety Realities | 6 | 5 | 2 | 5 | Kantrowitz and Goode question Mike Krieger about the immediate government intervention on Fable and ask whether Anthropic's risk framing is material or marketing. Krieger defends Anthropic's safety philosophy by explaining empirical uplift evals between experts and lay users. | |
| Autonomous Engineering Delegation and the Mission of Anthropic Labs | 6 | 6 | 1 | 3 | Krieger details how Anthropic Labs operates by anticipating six-month model capabilities and explains how Fable allowed overnight autonomous execution of full-codebase conversions from Python to TypeScript. The interviewers explore the verification process of delegated tasks. | |
| Balancing Platform Partnerships with In-House Product Innovation and Ethics | 7 | 4 | 2 | 6 | Kantrowitz and Goode press Krieger directly on the tension between providing foundational APIs and launching competing end-user applications like Claude Code against partners like Cursor and Figma. Krieger emphasizes transparency and open ecosystem building blocks. | |
| Anthropic Labs' Future Themes, Space Data Centers, and Token Economics | 6 | 6 | 1 | 3 | Krieger outlines future themes for Anthropic Labs, focusing on environmental self-knowledge for agents and abstracting away multi-step manual data plumbing. He debunks the idea that top token consumers are the most productive employees. | |
| Consumer AI Adoption Hurdles, Lean Startups, and Safety Governance | 6 | 5 | 2 | 4 | Kantrowitz shares data on app store proliferation versus user stagnation, prompting Krieger to analyze consumer data gravity and reflect on how AI tools could have reduced early Instagram engineering headcount from 13 to five. | |
| Greg Brockman on OpenAI's Evolution and the Zero-Interface Personal AGI | 7 | 5 | 2 | 3 | Greg Brockman discusses OpenAI's transition from conversational text models toward a unified personal AGI with minimal UI. Kantrowitz contrasts today's native tool execution with the failure of early 2023 chat plugins. | |
| Ecosystem Collaboration, Medical Reasoning, and Natural Voice Interfaces | 7 | 6 | 3 | 5 | Kantrowitz pushes Brockman on reports of OpenAI developing custom consumer hardware and competing against Apple's system-level Siri. Brockman details technical breakthroughs required for bi-directional, full-duplex conversational voice. | |
| The Longevity of Scaling Laws, Supercomputing Grids, and Compute Scarcity | 7 | 6 | 4 | 6 | Kantrowitz challenges Brockman on massive capital expenditures, impending price wars, and Satya Nadella's statement that foundation models are becoming commodities. Brockman defends scaling law longevity and argues that compute will remain the primary scarce asset. |
Statements from this episode (55)
Kantrowitz: ChatGPT Reached One Billion Active Users
“ChatGPT, according to third parties, has recently hit one billion active users.”
Kantrowitz: Nvidia Has Reached a $5 Trillion Market Cap
“NVIDIA, which is powering all this, is at a five trillion dollar market cap.”
Kantrowitz: OpenAI Recently Raised $122 Billion
“OpenAI recently raised a hundred and twenty two billion. That's three or four times larger than the largest IPO, pre-SpaceX.”
Kantrowitz: Tech Capex Will Hit $700 Billion This Year
“And we're going to see seven hundred billion dollars in capital expenditures this year”
Kantrowitz: OpenAI and Anthropic head toward trillion-dollar IPOs within next year
“We have open AI and Anthropic. Now they're going to do fifty billion dollars in revenue this year. They're both headed towards trillion dollar IPOs within the next year.”
Levie: Governments will evaluate and green-light AI model releases within 3-5 years
“I think practically in the next three to five years, we probably have to end up in an environment where models do get evaluated by the government. There is a sort of collaborative approach between the government and research and the labs. You know, the governm…”
Levie: Box's latest AI agents use up to 5 million tokens
“When we launched our first kind of AI use case within box, our product, The average number of tokens that was being used on a task was like 5010 1020 thousand tokens. Now our latest agents might use a million tokens or five million tokens on executing a task.”
Levie: Current AI agent spending is sustainable due to engineering use cases
“I would say the vast majority of the current agent expend that's happening is sustainable because partly because it's actually coming mostly from engineering and engineering related, related tasks. And this is an audience that is kind of technically capable of…”
Gardizi: OpenAI will not deploy 10 gigawatts in the next few years
“Even if you think of open AI, for example, announcing a 10 gigawatt project you know, we're not going to see 10 gigawatts in the next couple of years, or they're going to sort of spread out that bet.”
Cherney: Semiconductor factory buildouts are very likely to cause painful overcapacity
“It's very likely, I would say that we're going to lead, we're going to see some over capacity essentially. So like people are building, especially memory companies are building tons of factories right now. And typically the chip industry is, again, I'm sure ev…”
Gardizi: OpenAI uses alternative chips and develops in-house silicon to diversify
“They're using chips from other companies. Cerebris is a good example, and they have their own in-house chip. So I think maybe publicly they're doing some big announcements with NVIDIA, and I believe it's five gigawatts they have to deploy in the next couple of…”
Cherney: US chip designers have no contingency plan for Taiwan disruption
“Chip companies, the design companies in the US do not plan for this. The contingency plan, excuse me, is something along the lines of like, well, we're all kind of screwed if China invades, which, OK, but like, there's no real plan there.”
Goode: Intel Fab 52 matches TSMC 2nm technology but at fraction of scale
“This was the Intel fab 52, which is their most advanced fab. It's a two nanometer fab, and it is on the level of what TSMC does in terms of two nanometer, but of course, at a fraction of the scale of what TSMC does.”
Cherney: Nvidia has run out of growth avenues outside China
“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 can't sell to any more. Even they're going after the enterprise market, which is messy and complicated and difficult to actually make a …”
Cherney: SMIC cannot effectively manufacture advanced sub-node chips without EUV machines
“Their manufacturing process is, is a lot worse. Like one of the research firms just published an analysis of the sort of one of the new Chinese chips, like homemade by SMIC. And it's like, it's good, but they're, they run up to the limit of, like, they can't b…”
Goode: CoreWeave uses Nvidia investment money to buy Nvidia chips
“They're at the point where they have become a major investor in CoreWeave, and then CoreWeave is using that money to buy NVIDIA chips.”
Dolan: Frontier AI model labs are on the verge of a price war
“I think we're absolutely right on the precipice of that.”
Dolan: All 350,000 global PwC employees have access to AI tools
“We have 350,000 people globally. They all have access to one tool or another.”
Dolan: PwC control plane technology routes enterprise tasks to cheaper models
“So if somebody was coming to us with a cheaper model, a hundred percent, the direction of travel would go, you know, to that. And we've even built that into some of our control plane technology. So there's selectivity of model and there's also recommendations …”
Dolan: PwC maintains hiring volume while shifting from accounting to science graduates
“We're hiring as many people as we did last year and we're changing the who we're hiring. We're actually hiring a lot more kids who are pursuing sciences and, you know, even in areas that they go beyond, you know, engineering.”
Kantrowitz: Apple reassigned Vision Pro lead Mike Rockwell to fix Siri
“They put the person on Vision Pro on Siri and he fixed it. Wait, Mike Rockwell.”
Kantrowitz: AI autonomous coding duration grew from 30 minutes to 18 hours
“That like you saw these models, they could not code autonomously for more than 30 minutes in 20, 23. And now they can code autonomously for the human equivalent of 18 hours, right?”
Kantrowitz: AI companies will build user profiles and use them for ads
“So will AI's build an amazing sort of profile of you based off of all the personal data and then use that for ads? Most certainly.”
Roy: Google will put ads in Gemini; Anthropic will not
“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.”
Ranjan Roy: DeepSeek and Chinese models are now in almost every agent conversation
“The conversation around moving towards deep seek and adding it into your agentic process or adding Chinese models did not exist in any conversation I was in 12 months ago and now is in almost every conversation, at least as an option, because cost has become s…”
Stamos: Frontier AI matched top human bug finders in 2025
“This Rubicon was crossed well last year with the Opus four series with the GPT five series. That is when the models became better or as good or better than the best human bug finders and not just as good as the best Bug finders.”
Stamos: Banning bug-finding in US models degrades code security
“We cannot set the standard that US AI models can't find bugs. That is a terrible, terrible, terrible standard. If you have a, if you are writing software with an LLM, it has to be able to understand what a bug looks like so it does not write those bugs.”
Stamos: US AI labs are not years ahead of Chinese rivals
“This whole conversation is predicated on the idea that like the American labs are years ahead of our adversaries. And that is just not true, right? So while Fable was shut down on Tuesday of this week, GLM 5.2 was shipped right from Zeta AI, which is a Chinese…”
Stamos: Glasswing found 10,000 bugs but only fixed 1,000
“Glasswing is they found like 10,000 bugs, but they've only fixed a thousand. Right. So like you do not want a 10 to one ratio of the bugs you found to fixed.”
Stamos: Running Opus 4.8 will catch 70% of critical bugs
“Use Opus four eight right now, and I guarantee you'll get 70% of the bugs you care about.”
Stamos: NIST research proves models cannot be fully immune to jailbreaks
“No, there's actually a paper from NIST. It's like a kind of a Godel completeness theorem kind of paper that basically says it's impossible to make a model that is completely not jailbreakable.”
Stamos: Banning vulnerability-finding AI will force US firms onto Chinese models
“What would be terrible for the entire American AI and cybersecurity industries is if the Trump administration defines a jailbreak as having any ability to find any vulnerability in code, because then all of us have to go to Chinese models.”
Stamos: US CIOs contract Chinese open-weight models as backups against regulation
“And so now, this week, CIs and CTOs are signing contracts to have open weight models on different hosts, on U.S. Hosts. They're not using Chinese hosts, but they're using Chinese models on as backups through LLM routers because political risk Is now a risk of …”
Krieger: First-week reactions to new AI models should be ignored
“I've learned to not really trust day of or even week of model reactions. You don't really know until you've put it through its paces. And so like, I almost just completely block out the noise in the first couple of days of any new model release, because I don'…”
Krieger: Anthropic dynamic workflows autonomously converted hundreds of thousands of lines of code
“We have a feature called dynamic workflows where you can have it like also break down the task into like a lot of subtasks. And I trusted it to sort of not just do the individual action, but here's like a whole language conversion of millions of hundreds of th…”
Krieger: Anthropic product engineering had only 25 people in 2024
“I started the labs, the original labs team with Ben Mann, who's one of the co-founders of Anthropic in my third week at Anthropic. And it had been something that I've been bubbling under. And at the time, the reason was really different. It was all of our prod…”
Krieger: Anthropic Labs builds for capabilities models will gain in six months
“The two most useful thought exercises we do in labs, one is like visualize the gap between what the models can do today and how most people use it. It can be closed that gap. That's one. And the other one is imagine what the models are bad at now that they're …”
Krieger: Anthropic is in discussions with SpaceX
“We're, you know, we're talking to SpaceX about spacey things.”
Krieger: Top token users do not correlate with top performers
“We found that there's not a lot of correlation between
[10288] Mike Krieger: Like the person who's using the most tokens and like the person that I like, it's an interesting thought. I just made you at your companies like write down your 10 most productive peo…”
Krieger: Outcome-based pricing is very hard for general LLM work
“It gets so much fuzzier on these like tasks that we actually ask Claude these days. Like I had a strategy document. I use Claude to critique my strategy document like. What was the outcome? It's like, well, I don't know. It's like, tell me how the strategy goe…”
Krieger: Instagram could have reached sale with 4 to 6 people using AI
“I think we could have gotten there with like four to six, you know?”
Krieger: Instagram's month-long Android build would take one week with AI
“In about a month for Instagram, we could have done it probably in a week with the models”
Krieger: Anthropic shelved prototypes deemed potentially harmful to society
“There are certainly products that we have either prototyped or conceptualized and been like this product. It sounds so hypey. I hate this. But like this product, if shipped, would be bad for the world or like would nudge people in the wrong direction. Or even …”
Brockman: The ultimate AI product interface is zero interface
“You want almost no interface. You want no product, right? You want this to be like, what's the interface between you and me, right? Just being able to talk to a persistent entity of some form that's able to go and accomplish goals for you.”
Brockman: Frontier AI models are solving unsolved math and physics problems
“These models are now solving unsolved math problems and physics problems, right? Really helping people be able to achieve things they couldn't otherwise.”
Brockman: OpenAI operates almost entirely on Slack rather than email
“Within OpenAI, we basically have the same level of penetration now and usage of Slack. Right. It's like everyone in opening eyes, like an entirely slack based company. We do not use email for the most part.”
Brockman: Doctors use OpenAI's o3 model to diagnose long-unsolved medical cases
“Today we announced I we have in peer-reviewed literature people, doctors who are using O. Three. Remember O. Three? I was like forever ago now, right? It was like one of our earliest reasoning models using that to find diagnoses for People who I had no, no ans…”
Brockman: Future voice AI will process input and output simultaneously
“The obvious thing that you want to accomplish is a model An AI that works much more like you and I do, right? That's able to process input at the same time it's processing output. And all of that is, of course, something that many people in this field are tryi…”
Brockman: AI scaling laws show no wall in sight
“70 years of people, maybe 80 years now of people saying this stuff is never going to work, never going to scale, going to hit the wall, hasn't hit the wall yet. There's still no wall in sight.”
Brockman: Every AI provider will sell out all their compute
“I do think there's a bit of an attractor state where just like from a business model perspective, every provider sells out all their compute. I think that is just like the world that we're heading towards, where there just is not going to be enough compute to …”
Brockman: OpenAI has invested in custom chip program for years
“For example, we've been investing in our own chip program now for multiple years and super exciting progress. Like, you know, we'll we'll we'll have more to announce. I actually pretty soon.”
Brockman: Today's frontier AI intelligence will be much cheaper within a year
“I think the thing you should anticipate is that over a year long time horizon to get to today's level of intelligence that feels very premier, it's going to be much cheaper. But there's going to be a new thing that is going to be so much better.”
Brockman: Nvidia H100 prices are up despite being previous-gen hardware
“You can see that with some of the prices that people are paying for each 100, right? Hoppers are, you know, kind of, you know, not, not obsolete, right? But they're a previous gen chip and nor in any normal situation, we're not totally supply constrained. No o…”