Video · 2025-07-31 · 1h 4m · 6 moments

China Open-Source, Compute Arms Race, Reordering Global Trade | BG2 w/ Bill Gurley and Brad Gerstner

✦ AI generated

GUEST Brad Gerstner GUEST Sunny Madra GUEST Bill Gurley

timeline · colored by role

Mechanism◆

Chinese open-source AI labs are compounding progress by distilling and building on each other's models, which could let them surpass the best US proprietary models by late 2025.

Sunny Madra explains that China's cluster of open-weight model makers can remix and distill each other's work instead of training in silos, accelerating both frontier capability and smaller 'turbo' model quality.

transcript

Sunny Madra: Instead of working in silos and instead of having to create giant training clusters separately they can basically take each other's work, build on top of it, almost consider it like a remix of someone's model — K2 sort of a well-known remix of what Deepseek had done — and now we're starting to see that happen really fast.

explains mechanism · 1

Example

A community of AI labs that shares best practices with competitors, as China's open-source ecosystem does, will achieve higher collective output than one where firms compete without sharing ideas.

Bill Gurley uses a farmers-market analogy — communities that share weekly best practices outperform those that only compete — to explain why China's open-source AI ecosystem is co-evolving so fast.

transcript

Bill Gurley: If you ran that exercise over two years... the community where the best practices are shared across all farms is going to have a higher global output for the community than the one where you've just got proprietary ideas driving... competition without the idea sharing... I do think you end up with a higher fitness level for a community that's behaving that way overall.

supports · 1

Claim

AI model providers are pricing inference below cost to grab market share instead of raising prices, even though demand is outstripping supply, which is unsustainable.

Bill Gurley warns that flush venture capital lets AI labs price inference under cost to chase share rather than raise prices to throttle demand, with some top brands reportedly running negative gross margins.

transcript

Bill Gurley: No one's pricing the cost because they're pricing to take market share... the thing that throttles demand is price, but no one here is raising price. Like Anthropic doesn't need to throttle. They just raise price, but they're not willing to do that because they're afraid to lose share. And so everyone's pricing to share. That means they're pricing under cost.

Prediction◆

As the AI model layer becomes commoditized through rapid open-source distillation, the real economic battle shifts to owning the consumer-facing applications, not the underlying intelligence.

Brad Gerstner argues that with seven-plus Chinese labs distilling off each other to drive intelligence up and cost down, the model layer is being commoditized, so value will accrue to whoever owns the consumer application layer, much as Google's dominance came from owning the consumer.

transcript

Brad Gerstner: The battle for the consumer is ultimately where the value occurs, Sunny, not who runs what hardware... What that tells me is the model layer is being increasingly commoditized and that there's not going to be a lot of intrinsic value in that intelligence layer, that operating layer. You're going to have to build applications that guys like Bill Gurley and you and I are using every day.

explains mechanism · 1extends · 1supports · 2

Claim◆

The Trump administration's tariff strategy has succeeded — trading partners are accepting one-sided tariffs and making massive investment commitments, with no retaliation or trade war.

Brad Gerstner argues the EU and Japan deals (15% tariffs paid to the US, 0% tariffs on US goods, hundreds of billions in committed investment) prove predictions of trade-war disaster were wrong, and the president deserves credit.

transcript

Brad Gerstner: 15% on all goods coming from Europe, 0% on US goods going to Europe... Everybody said this was going to lead to retaliation to trade wars was going to be disastrous for the US and all we've seen so far is deals deals deals deals.

supports · 1

Prediction

Trump wants to strike the biggest trade deal ever done with China, potentially bundling rare earths, chips, military cooperation, and defense-spending cuts, not just tariff rebalancing.

Brad Gerstner predicts a major US-China deal before year end, citing Trump's stated willingness to halve defense budgets for the US, China, and Russia as evidence of an unusually flexible negotiating posture that could extend beyond trade into military cooperation.

transcript

Brad Gerstner: The president said earlier this year something that caught all of our attention. He said, 'You know, if I could wave a magic wand, I would cut the defense budget in half for the United States, for China, and for Russia.' We've never heard a US president in history utter anything like that... I think it could include all of the above. Rare earth chips, maybe even military cooperation, certainly a rebalancing of trade.

provides context · 2

Highlight slides

China's Open-Source AI Labs Are Remixing, Not Siloing✦ from: Chinese open-source AI labs are compounding progress by distilling and building on each other's models, which could let them surpass the best US proprietary models by late 2025.Why This Accelerates Progress✦ from: Chinese open-source AI labs are compounding progress by distilling and building on each other's models, which could let them surpass the best US proprietary models by late 2025.Stakes: Catching the Best US Proprietary Models✦ from: Chinese open-source AI labs are compounding progress by distilling and building on each other's models, which could let them surpass the best US proprietary models by late 2025.Model Layer Getting Commoditized✦ from: As the AI model layer becomes commoditized through rapid open-source distillation, the real economic battle shifts to owning the consumer-facing applications, not the underlying intelligence.Value Shifts to the Application Layer✦ from: As the AI model layer becomes commoditized through rapid open-source distillation, the real economic battle shifts to owning the consumer-facing applications, not the underlying intelligence.The Google Parallel✦ from: As the AI model layer becomes commoditized through rapid open-source distillation, the real economic battle shifts to owning the consumer-facing applications, not the underlying intelligence.Tariff Strategy Declared a Success✦ from: The Trump administration's tariff strategy has succeeded — trading partners are accepting one-sided tariffs and making massive investment commitments, with no retaliation or trade war.The Lopsided Tariff Structure✦ from: The Trump administration's tariff strategy has succeeded — trading partners are accepting one-sided tariffs and making massive investment commitments, with no retaliation or trade war.Predictions vs. Reality✦ from: The Trump administration's tariff strategy has succeeded — trading partners are accepting one-sided tariffs and making massive investment commitments, with no retaliation or trade war.

Related episodes

Video · a16z Podcast

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

→

Video · All-In Podcast

The Fight Over Open Source AI, Anthropic's $1.5B Payout, NYC Socialists: Evictions = Violence?

→

Article · Interconnects

Kimi K3: The open-weights escalation

→