May 29, 2026 · 33m · a16z
The New Rule for Picking AI Winners | The a16z Show
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
In this episode of The a16z Show, General Partner David George and host David Clark discuss the explosive growth of frontier AI companies, shifting venture capital exit dynamics, enterprise adoption, token economics, and whether the market is experiencing a speculative bubble.
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 4.7 Guest teaching 4.3 Guest disagreement 1.3 The host pushing back 1.4
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 18:05 Rejecting Low Loss Ratio Pride
David George forcefully dismisses the notion that avoiding losses is commendable in venture capital, labeling a zero loss record a horrible data point and reframe of proper risk-taking.
Hardest push from the host ▶ 17:15 Challenging AI Loss Ratio SustainabilityDavid Clark directly questions the current AI investment boom by contrasting historical sixty percent venture loss ratios with today's single-digit failure rates.
Biggest teaching moment ▶ 1:15 Scale of Frontier Model RevenueDavid George educates the interviewer on the unprecedented revenue velocity of OpenAI and Anthropic compared to legacy cloud and tech giants.
The host holds their own ▶ 11:20 Forbes AI 50 Turnover InsightDavid Clark demonstrates deep sector expertise by using a forty percent turnover stat from the Forbes AI 50 list to challenge assumptions about startup defensibility.
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 |
|---|---|---|---|---|---|---|
| Shifting Priors on Enterprise AI Scale and Revenue | 1 | 5 | 1 | 0 | David George delivers an extended monologue detailing how OpenAI and Anthropic are adding monthly revenue faster than tech giants despite less than five percent enterprise diffusion. The host/interviewer remains largely silent, allowing the guest to lay out financial metrics and market assumptions without challenge. | |
| AI-Native Applications, Workflows, and Startup Culture | 4 | 4 | 1 | 1 | David Clark introduces Chris Dixon's framework on skeuomorphic applications to frame the discussion on native AI workflows. David George elaborates on how modern AI founders operate leanly with agent swarms, keeping the exchange collaborative and informative. | |
| Exploding Exit Thresholds and Market Concentration | 6 | 3 | 0 | 1 | David Clark shares updated firm data showing top one percent exit thresholds rapidly inflating from ten billion dollars to thirty-two billion dollars. David George concurs and reinforces the point by contrasting future AI IPOs with historical VC-backed IPO totals. | |
| Predictability, Defensibility, and Value Capture in AI | 6 | 4 | 2 | 2 | David Clark cites the forty percent annual turnover on the Forbes AI 50 list to question the defensibility of early market leaders. David George acknowledges the rapid technological shifts and outlines the fluid power balance between model providers and applications. | |
| Global AI Competition, Token Economics, and Frontier Demand | 6 | 5 | 2 | 3 | David Clark brings empirical comparisons from China and contrasts historical sixty percent venture loss ratios with current artificially low AI failure rates. David George reframes risk management, arguing that a zero loss ratio indicates insufficient risk-taking. | |
| Industry Valuation Sentiment, Supply Constraints, and AI Bubble Assessment | 5 | 5 | 2 | 2 | David Clark introduces a survey showing eighty percent of VCs view AI valuations as overinflated and inquires about bubble dynamics. David George counters by citing severe hardware and data center supply constraints that prevent an immediate market bubble. | |
| Public Market Impact and Hypergrowth IPOs | 5 | 4 | 1 | 1 | David Clark questions public market capacity to absorb massive hypergrowth AI IPOs. David George explains that public markets are starved for growth outside a few tech giants and will eagerly digest high-growth AI entrants. |
Statements from this episode (21)
David Clark: Top 1% VC exit threshold reached $32 billion
“Between 2020 and 2024, top one percent exit started at ten billion dollars. We updated those numbers in February this year, twenty billion dollars. We just updated them yesterday. It's now at thirty two billion dollars. So we've tenxed over the space of kind o…”
George: Anthropic and OpenAI add more monthly revenue than tech giants
“Anthropic and OpenAI are adding more revenue per month than Meta, Google, or Microsoft.”
George: AI Diffusion Into Real Economy Is Less Than 5%
“Actual diffusion of this technology into the real economy is tiny. It's like less than five percent.”
David George: Anthropic and OpenAI combined run rate could hit $200B
“And I wouldn't be surprised if the combination of those two companies is doing two hundred billion of revenue run rate.”
George: Corporate layoffs reflect trimming fat, not AI efficiency gains
“What's happening with some of the layoff things that we're seeing is kind of like trimming of previous fat. Like I don't think it's actually efficiency gains.”
George: AI model companies are adding more revenue than public software combined
“The model companies are adding more than the entire public software universe in terms of revenue added, you know, combined.”
Clark: Top 1% VC exit threshold could pass $100B by September
“And then if you then think about OpenAI and Anthropic coming in you know, potentially we could be north of a hundred billion dollars by September.”
George: Top AI IPOs will individually exceed all recent VC IPOs combined
“We actually did a similar analysis where we looked at all of the VC backed IPOs that happened over the last six years. And if you sum all of them up, they're A little over a trillion dollars. Like that's probably gonna be smaller than any of the three of the l…”
Clark: 40% of Forbes AI 50 startups turned over year-over-year
“From last year to this year, 40% of the companies that were on that list last year dropped off.”
George: a16z prioritizes AI startups sitting directly in the token path
“First of all, like right now, you have to be in the token path. Like that is the number one thing that we're looking to for our companies.”
George: Fewer than five companies currently offer frontier AI models
“Right now the number is smaller. It's not five. There's a tremendous amount of inelasticity for frontier intelligence right now.”
Clark: Chinese LLMs lag US by six months but are 10x cheaper
“The leading LLMs in, in China are probably six months behind where we are in the U.S. In terms of the capability of their models, but they're 10 X cheaper.”
George: Distilling an AI model costs only 2% of pre-training
“It probably costs in the order of like two percent of the actual training cost, pre-training cost of a model to distill it.”
George: Frontier AI token demand outpaces 10x annual cost drops
“The sort of per token Costs like for like is going down more than 10 X year over year. But the appetite for tokens on the frontier is massively exceeding that in terms of dollars.”
George: Backing the wrong founder in a winning category is VC's worst mistake
“If the space happens to not work out and we have the leader, no harm, no foul. Actually, that's part of our business. That's what we should be doing. The bad box of what I described is the space works out and we pick the wrong one.”
George: AI coding assistant Cursor has reached billions in revenue
“Cursor, as an example, is, you know, billions of dollars of revenue. And they're very small, and it's very early in their life.”
Clark: 80% of respondents in UK VC survey call AI valuations too high
“So it's funny, one of my colleagues was at a conference yesterday that was run by the UK Venture Capital Association. And they surveyed the audience saying, you know, what do you think about AI valuations today? You know, too high, about right, you know, too l…”
David George: AI market is not currently in a bubble
“I feel pretty confident saying that we're not in a bubble right now.”
George: Scaled data center capacity unavailable until late 2028 or early 2029
“You can't get data center capacity at scale until late 28, early 29 right now, and that's just a fact.”
George: AI market will remain supply-constrained rather than bubble
“But I think it's more likely we remain supply constrained for the next three years than then we end up in bubble territory.”
George: All Mag Seven and public software stocks grow sub-30%
“If you exclude the data center supply chain stuff right now, there are very few companies that are growing fast that are available for people to buy in the public markets. You know, the mag seven are all growing sub 30% at this point. You know all the software…”