Oct 30, 2025 · 31m · a16z
"Is there an AI bubble?” Gavin Baker and David George
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
At the a16z Runtime conference, Gavin Baker and David George analyze whether artificial intelligence is in a financial bubble, concluding that current investments are backed by strong revenue and real GPU utilization. They also evaluate shifting software business models, semiconductor competition between NVIDIA and Google, and the rapid rise of humanoid robotics.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The host as informed peer 4.0 Guest teaching 5.1 Guest disagreement 1.6 The host pushing back 1.5
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 21:40 Rant on scaling law skepticism
Gavin Baker forcefully rejects commentators using GPT-V to claim scaling laws are dead, calling such references crazy because the model was optimized for inference cost.
Hardest push from the host ▶ 25:05 Pressing on ASIC failure timelineDavid George directly interrupts Gavin Baker's bold claim about custom ASICs failing to demand a specific timeline.
Biggest teaching moment ▶ 3:21 Dark fiber vs GPU utilization analysisGavin Baker educates the host on telecom bubble metrics, showing how 97% unlit dark fiber contrasts sharply with today's fully utilized GPUs.
The host holds their own ▶ 6:06 Hyperscaler cash flow and balance sheet bufferDavid George demonstrates strong financial mastery by detailing the $800 billion liquidity buffer generated by top tech companies that absorbs AI capex.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The host as informed peer | Guest teaching | Guest disagreement | The host pushing back | Why |
|---|---|---|---|---|---|---|
| Highlight Preview: Is There an AI Bubble? | 1 | 1 | 0 | 0 | David George introduces Gavin Baker and sets up the episode's central themes around AI bubble concerns, with light banter regarding intro music. | |
| Macroeconomic Analysis: Is There an AI Bubble? | 6 | 5 | 1 | 0 | David George brings strong macroeconomic data comparing data center spend to the interstate highway system, while Gavin Baker provides a deep historical comparison between 2000 dark fiber and today's utilized GPUs. | |
| Round-Tripping Concerns and Strategic Chip Competition | 4 | 6 | 2 | 1 | Gavin Baker reframes round-tripping fears by detailing Google TPU and DeepMind competitive pressure on Nvidia, while David George concurs with the strategic rationale. | |
| Frontier Model Dynamics and Incumbent Advantages | 5 | 5 | 2 | 4 | David George gently interrupts Gavin Baker to distinguish between model companies acting as infrastructure providers versus application layers, prompting a discussion on sustaining innovations for Big Tech. | |
| AI Lab Economics and Gross Margin Realities | 4 | 6 | 2 | 2 | Gavin Baker explains how scaling laws drive down gross margins for AI frontier labs and criticizes application SaaS companies for fearing lower margins, referencing Microsoft's cloud shift. | |
| SaaS Incumbent Defense vs. AI Native Startups | 6 | 4 | 1 | 1 | David George shares internal venture capital insights where low gross margins signal genuine AI usage, using Figma as a case study while Gavin highlights Cursor's advantage over traditional coding tools. | |
| Consumer AI Interfaces and Reasoning Flywheels | 3 | 6 | 3 | 1 | Gavin Baker details how reasoning flywheels alter consumer model economics and launches into a fiery rant dismissing claims that GPT-V signals the end of scaling laws. | |
| Semiconductor Landscape: NVIDIA vs. Google TPUs | 4 | 7 | 2 | 5 | Gavin Baker outlines the hardware battle between Nvidia and Google TPUs, prompting David George to press him on whether custom hyperscaler ASICs will fail short-term or long-term. | |
| Outcome-Based Business Models and AGI Outlook | 5 | 6 | 2 | 1 | David George grounds the conversation in outcome-based startup models like Decagon, while Gavin Baker explains why search advertising systematically causes overpaying compared to outcome pricing. | |
| The Rapid Evolution of Humanoid Robotics | 2 | 5 | 1 | 0 | Gavin Baker explains why humanoid robotics are winning over domain-specific robots due to video learning data and teleoperation suits, concluding a supportive interview. |
Statements from this episode (23)
US data center spending over three years topped interstate highway budget
“Over the past three years, we have already built out in data center capacity a larger amount of dollars than the entire US interstate highway system, which took 40 years.”
Baker: Artificial intelligence is not currently in a financial bubble
“I do not believe we're in an AI bubble today.”
Baker: Unlike 2000 dark fiber, no dark GPUs exist today
“At the peak of the bubble, 97% of the fiber that had been laid In America was dark. Contrast that with today. There are no dark GPUs.”
Top GPU spenders gained 10-point ROIC increase after raising AI CapEx
“Those companies, since they ramped up CapEx, have seen, call it a 10 point increase in their ROICs. So thus far, the ROI on all the spending has been really positive.”
Baker: Larry Page said he would risk bankruptcy to win AI race
“Larry Page apparently internally said, I'm happy to go bankrupt rather than lose this race. And I think that is the mentality for sure at Google and perhaps Meta. It's just seen as existential and you have to win.”
Baker: Google's TPU makes it Nvidia's biggest AI chip competitor
“NVIDIA's biggest competitor, it's not AMD, it's not Broadcom you know, it's certainly not Marvell. It's not Intel. It's Google. And more specifically, it is Google, because Google owns the TPU chip.”
Baker: Google surpassed OpenAI and Anthropic in total AI traffic
“I think they've taken 15 or 20 points of traffic share in the last two or three months, and that does not, that's just traffic to Gemini. It does not include search overviews. I suspect on a actual traffic basis, Google is bigger than OpenAI, Anthropic, anyone…”
Baker: Infrastructure is safer than applications early in tech waves
“It's one reason I think the infrastructure layer is often maybe a safe place to be at the beginning of one of these new technology waves.”
Baker: AI may be a sustaining innovation for Big Tech
“I think there's reasonable arguments that AI could be a sustaining innovation because the raw ingredients of kind of data, you know, the capital to buy compute and distribution, which is what you need. All of the big you know, today's biggest tech companies ha…”
Baker: ChatGPT was a 'Pearl Harbor' moment for Google
“No, to me, ChatGPT was Pearl Harbor for Google, and we're going to see how they responded, and they're slowly starting to respond.”
Baker: Frontier AI lab gross margins will stay structurally below SaaS
“And the nature of AI because of scaling laws, Richard Sutton's the better listen they're just more compute intensive. So their gross margins are structurally going to be lower. But that doesn't mean they can't be great businesses. I just, I think it's going to…”
Baker: Software companies cannot succeed in AI without margin pressure
“It is definitionally impossible, given what we just discussed, to succeed in AI without gross margin pressure, and I do not know why they have concerns, because we have an existence proof that a software company can deal well with declining margins in Microsof…”
George: Low gross margins signal real product engagement for AI startups
“We always look at the gross margins, and it's become like a badge of honor for them to actually have low gross margins, because we're like, oh my god, people are actually using your AI stuff. But if you show up, and you're like, I'm an AI company, and it's lik…”
Baker: Cursor's token flywheel will become tough for competitors to catch
“And the reality is, Kirscher, now they have a trillion, trillion tokens. And you know, there will be a point where they have enough coding tokens that it's tough to catch them.”
George: Figma lowered gross margins for AI distribution without investor backlash
“Figma, for example, like, when they went out, they are extremely high gross margin, and they're like, hey, we're gonna, you know, pretty aggressively distribute our AI tools, and our gross margins are gonna go down, and You know, investors asked a few clarifyi…”
Baker: AI browser startups will regret competing against Google Chrome
“I think the AI companies that have launched these AI browsers may come to regret it, because there's something called Chrome that has whatever it is, five billion users, and if you're Google you know, you can just go look at what happened with Google Buzz.”
Baker: Chinese open-source AI helps US firms catch leading frontier labs
“In a strange way, the Chinese open source model ecosystem is a godsend to any American company that's trying to catch those four leading labs.”
Baker: GPT-V is not evidence that AI scaling laws are ending
“GPT-V is a smaller model. It was not designed to be better. It was designed to be more economical for OpenAI and Microsoft to run. Any reference to GPT-V and scaling loss is crazy.”
Baker: Major custom AI ASIC programs will be canceled within three years
“In the next three years, I think you'll see a bunch of high-profile ASIC programs canceled, especially if Google starts selling TPUs externally, which has been all over X.”
Baker: AI business models will shift to outcome-based pricing
“Humans were fundamentally paid based on outcomes, and a lot of AI will be augmenting humans, but probably also replacing some humans, and that will involve being paid paid for outcomes.”
Baker: Google avoided outcome models because advertisers systematically overpay for ads
“Because people overvalue systematically their ability once they've acquired a customer through Google to keep it as an organic customer. So they systematically overpay. And they continue doing that. That's why Google never went to outcomes or marketplace, beca…”
Baker: Global robotics market will be Tesla versus China
“Yeah, very real, and it's gonna be Tesla versus the Chinese in the same way it's Tesla versus the Chinese in, in cars.”
Baker: Debate between humanoid and non-humanoid robots is over
“Is it going to be humanoids or not humanoids? I think that debate is over because humanoids can kind of learn, you know, from watching YouTube videos, and then it's easier for a human being you know, to put on a suit and show the robot how to do it.”