Video · 2025-10-14 · 1h 2m · 6 moments

AI Bubble, Stablecoin Boom, and Runnin' Down a Dream | BG2 w/ Bill Gurley and Brad Gerstner

✦ AI generated

GUEST Brad Gerstner GUEST Jensen Huang GUEST Bill Gurley

timeline · colored by role

Claim◆

Several recent AI-industry financing structures — in-kind compute credits booked as revenue, vendor-financed deals, off-balance-sheet-style risk transfers — match the transaction patterns historically associated with accounting scandals like Enron and WorldCom, and deserve scrutiny.

Gurley says he fed a description of the recent unusual AI financing transactions to ChatGPT, and it independently flagged them as resembling historical accounting-fraud patterns like Enron and WorldCom.

transcript

Bill Gurley: I just described those things to ChatGPT and asked it for its analysis both as an accountant and as a financial investor. And the AI itself, you know, would would find its way toward company names like Enron and WorldCom and those kind of things merely by describing the type of transaction.

extends · 2

Claim

Nvidia's own AI-startup investments aren't concerning given its massive free cash flow, but the further an investment goes out the risk curve — into startup neoclouds and challenger chipmakers with weak balance sheets — the more likely it is to signal disguised demand problems.

Gerstner distinguishes Nvidia's well-capitalized, non-obligatory investments from riskier deals further down the AI supply chain, predicting more 'yellow flags' will emerge among weaker startups and chipmakers.

transcript

Brad Gerstner: But I do think that as you go further and further out the risk curve, right? Further and further to these startup neo clouds or further and further to startup chips, you know, etc., where people, to your point, are a little bit more desperate for capital, don't have the balance sheets, don't have the market leadership position, I would not be surprised at all in this moment to see more of those yellow flags emerge.

extends · 1

Prediction

There is essentially zero chance of an AI compute glut over the next two to three years, because demand from converting general-purpose computing to accelerated computing hasn't been exhausted, and buildout is going to the hyperscalers with the strongest balance sheets.

Recounting his interview with Jensen Huang, Gerstner relays that Huang sees essentially no chance of a near-term compute glut until general-purpose computing has fully converted to accelerated/AI computing.

transcript

Jensen Huang: Until we fully convert all general purpose computing to accelerated computing and AI. Until we do that. Yes. I think the chances are extremely low.

rebuts · 2supports · 3

Claim◆

A patchwork of 50 different state-level AI regulations, while competitors abroad face none, will inevitably create legal confusion and slow down US AI companies in the global race, so AI regulation should be federally preempted.

Gurley argues that laws like Colorado's AI Act and California's chatbot-safety law create a fragmented, high-friction regulatory patchwork that will hamper US AI companies relative to foreign competitors, and calls for federal preemption or a moratorium on state AI laws.

transcript

Bill Gurley: If we implement 50 different state rules that these companies have to jump through, and companies that are that are competitors that are competing in the broader world don't have any of them, there is zero chance that's not going to create mud and slow down the US players. There's just zero chance.

extends · 1rebuts · 1supports · 1

Claim

Stablecoin-based payment rails through Coinbase and Circle already offer better functionality than the traditional US banking system — paying ~4% daily on any balance and enabling instant, near-free settlement without moving money between accounts.

Gerstner describes how Coinbase/Circle's stablecoin product lets users earn roughly 4% daily on any balance while transacting instantly, arguing the underlying rails are already working better than legacy banking.

transcript

Brad Gerstner: Here, whether it's 10 bucks or or a million bucks, you know, you put it in stablecoin with Coinbase and you start earning 4% daily. And on top of that, and this gets back to the PIX thing, you can transact immediately out of that account.

extends · 3gives example · 2provides context · 1supports · 1

Claim◆

People are far more likely to regret the career risks and chances they didn't take than the ones they did, which is why individuals should be willing to take a chance on work they love.

Citing Daniel Pink's 'The Power of Regret,' Gurley explains the core thesis behind his book: research consistently shows people regret inaction — chances not taken — far more than actions taken, which is why his book urges readers to pursue what they love.

transcript

Bill Gurley: One of the most robust findings in the academic research and on my own is that over time we are much more likely to regret the chances we didn't take than the chances we did. He says again, the surface domain, whether the risk involved are education or work or love lives, doesn't matter much. What haunts us is the inaction itself.

gives example · 1

Highlight slides

AI Financing Echoes Enron-Era Accounting Patterns✦ from: Several recent AI-industry financing structures — in-kind compute credits booked as revenue, vendor-financed deals, off-balance-sheet-style risk transfers — match the transaction patterns historically associated with accounting scandals like Enron and WorldCom, and deserve scrutiny.ChatGPT Independently Spotted the Parallels✦ from: Several recent AI-industry financing structures — in-kind compute credits booked as revenue, vendor-financed deals, off-balance-sheet-style risk transfers — match the transaction patterns historically associated with accounting scandals like Enron and WorldCom, and deserve scrutiny.50 State AI Rules Will Slow US Competitors✦ from: A patchwork of 50 different state-level AI regulations, while competitors abroad face none, will inevitably create legal confusion and slow down US AI companies in the global race, so AI regulation should be federally preempted.The Fragmentation Problem✦ from: A patchwork of 50 different state-level AI regulations, while competitors abroad face none, will inevitably create legal confusion and slow down US AI companies in the global race, so AI regulation should be federally preempted.Regret of Inaction Over Action✦ from: People are far more likely to regret the career risks and chances they didn't take than the ones they did, which is why individuals should be willing to take a chance on work they love.

Related episodes

Video · a16z Podcast

Several recent AI-industry financing structures — in-kind compute credits booked as revenue, vendor-financed deals, off-balance-sheet-style risk transfers — match the transaction patterns historically associated with accounting scandals like Enron and WorldCom, and deserve scrutiny.

Nvidia's own AI-startup investments aren't concerning given its massive free cash flow, but the further an investment goes out the risk curve — into startup neoclouds and challenger chipmakers with weak balance sheets — the more likely it is to signal disguised demand problems.

There is essentially zero chance of an AI compute glut over the next two to three years, because demand from converting general-purpose computing to accelerated computing hasn't been exhausted, and buildout is going to the hyperscalers with the strongest balance sheets.

A patchwork of 50 different state-level AI regulations, while competitors abroad face none, will inevitably create legal confusion and slow down US AI companies in the global race, so AI regulation should be federally preempted.

Stablecoin-based payment rails through Coinbase and Circle already offer better functionality than the traditional US banking system — paying ~4% daily on any balance and enabling instant, near-free settlement without moving money between accounts.

People are far more likely to regret the career risks and chances they didn't take than the ones they did, which is why individuals should be willing to take a chance on work they love.

Ben Horowitz on the Global Race for Tech, Power, and Influence

Kimi K3: The open-weights escalation

6 months to live for open models