Coatue’s Laffont Brothers. AI, Public & VC Mkts, Macro, US Debt, Crypto, IPO's, & more | BG2

Summary · 5 min read instead of 1 h 1 m

Thomas and Philippe Laffont of Coatue spoke with Brad Gerstner and Bill Gurley at Coatue's East meets West conference in Los Angeles, marking its 10th anniversary. They walked through slides from a deck Coatue published that day on public markets, venture and technology. Coatue holds CoreWeave shares and processes about 100 million credit card receipts a day for its data science work.

Philippe Laffont argues that AI is the biggest tech trend in decades and that tech's share of the economy will keep growing. The Laffonts see ChatGPT already cutting Google usage among paying users and a spread of GPUs across more cloud providers than before. Philippe lays out how faster productivity could stabilize US debt, and Thomas sees IPOs and M&A recovering after the 2021 cycle. They finish with advice for founders by growth and profitability, urging slower-growing companies to reinvent. The hosts add that AI-driven headcount discipline is expanding margins at companies such as AppLovin.

Minute by minute

1. 2:31

   AI is the defining tech cycle. Philippe Laffont says AI is probably the biggest tech trend to come, built on earlier waves such as mainframes, networking, PCs and the internet. Tech is about 50% of the market now, and he wonders whether utilities and power equipment makers should be reclassified as TMT. Gerstner adds that tech grew from 5% to 15% of global GDP over Coatue's life and will exceed 15% in ten years.

2. 6:03

   Mag 7 flat while AI pure plays emerge. Thomas Laffont says the Mag 7 is basically flat year over year, while value is accruing to top AI companies such as OpenAI and Anthropic. He sees CoreWeave, a Coatue holding, as one of very few AI pure plays in public markets. Many large listed names have legacy businesses or disruption threats, he says, using Google as an example.

3. 7:52

   Bitcoin is too big to ignore. Philippe Laffont puts world net worth at about 450 to 500 trillion and Bitcoin at about 2 trillion, and asks whether it could reach 4 or 5. He compares this with Microsoft at 3.5 trillion doubling in ten years, which is 7% growth per year. He says investors must stay willing to change their minds after failed venture bets in crypto, and cites Druckenmiller on mental flexibility.

4. 14:13

   Stablecoins could reshape Treasury funding. Gurley says crypto changed because the government moved from antagonistic to supportive and stablecoins found real business use. Philippe Laffont predicts stablecoins will be allowed to pay interest, followed by one-year to 30-year versions. That would let the government sell bonds straight to consumers worldwide instead of through dealers. Thomas Laffont asks whether Bitcoin could become the alternative to an overspent dollar.

5. 16:04

   ChatGPT is eroding Google usage. Coatue joined credit card and email receipt data to track users who started paying 20 dollars a month for ChatGPT. Absent a subscription, Google page views per user grow about 4% a year, Philippe says. For subscribers they fall 8% year over year, or 11% peak to trough, over almost two years. Gurley notes ChatGPT's adoption held up against Meta, Gemini and Grok, and Laffont says it is beginning to show network effects and switching costs.

6. 22:44

   GPU share differs from cloud share. Coatue's slide compares cloud revenue share with NVIDIA GPU allocation. Amazon holds 44% of cloud revenue but 20% of GPUs, Oracle rises from 5% to 19%, and CoreWeave reaches 11%. Gurley offers three explanations for Amazon: it lags in AI, it follows a different hardware strategy, or NVIDIA avoids dominant customers. Philippe expects a dozen hyperscalers, with sovereigns and possibly Anthropic joining, plus custom chips.

7. 28:47

   Productivity could stabilize US debt. Philippe Laffont says markets are expensive but were also expensive in the 1990s, and he does not see tariffs as the main risk. He says debt to GDP is at 100% on its way to 140. If productivity runs 2.5% to 3.5% a year for a decade, he argues, the ratio could stay at 100 or fall to 80. He adds that a 1993 forecast of debt rising from 60 to 80 ended with it falling to 40.

8. 35:10

   IPO and exit signals turn from red. Thomas Laffont calls 2021 an unhealthy cycle, with too much capital going in and historically low IPOs. This year he sees IPOs rebounding and performing better, and says the average new-issue cohort shows rule of 40 growth and profitability. He cites the 2021 IPO cohort, down 40% after one year and 50% five years later. He also points to CoreWeave's disciplined disclosure.

9. 38:55

   Meta's Scale deal shows AI urgency. Gurley describes Meta buying 49% of Scale at a $30 billion valuation, so effectively paying $15 billion. Thomas Laffont says Zuckerberg sees a multi-trillion prize, which makes 15 billion look like roughly 1% of market cap. He adds that the ecosystem moves fast: Anthropic took about a year to reach its first billion in revenue, three months for the next and two for the one after.

10. 42:05

    Large private companies should go public. Thomas Laffont says IPOs give companies financing options, a brand moment and transparency, and he believes all these companies should list. Gerstner argues OpenAI in particular should be public, since trillion-dollar companies should not be open only to private investors. He calls companies above 5 or 10 billion in value quasi public, and Philippe says those that avoid public scrutiny will face regulators.

11. 45:44

    AI is driving margin expansion. Gerstner says the Mag 7 grew over 20% compounded in recent years while opex grew 2%. Thomas Laffont describes Microsoft's headcount in three phases: ZIRP-era hiring, a get-fit era and now an AI era in which headcount may have peaked. AppLovin's revenue doubled while employees fell over 35%, and Gerstner contrasts Google's 187,000 employees with OpenAI's 2,700. Gurley says willingness to cut headcount signals real confidence in AI.

12. 51:46

    Fewer jobs per company may mean more companies. Thomas Laffont raises Jevons paradox, which a speaker brought up that day, and says he is not sure how AI will affect unemployment. He thinks companies may need fewer employees but more companies will form because starting one is easier. His own view is that AI may create more, and more interesting, jobs with more responsibility.

13. 52:59

    Advice for founders by growth and profit. Thomas Laffont's matrix splits companies at 25% growth and by cash flow. Fast-growing profitable companies should prepare for a possible IPO, and burning ones should build a fortress balance sheet. Slower profitable companies should play offense with AI or M&A, and slower burning ones should reinvent, for example by betting on a small, fast-growing new product. Gurley adds that mid-sized companies protect something that may not exist if their multiple falls from five to three to one times revenue.

Published Jun 20, 2025.