Will the US restrict Chinese open-weights models to protect its domestic AI labs?

Martin Casado · Jul 28, 2025

Open source AI is currently most dangerous as a policy matter because China is better at open source than the US is

China outperforms the US at open source model development, so open sourcing advantages them

Scope: framed as 'right now'

0:00 20VC: a16z's Martin Casado on Anthropic vs OpenAI: Where Value Accrues | Cursor vs Replit vs Lovable: Who Wins and Who Loses | The One Sin in AI Investing | Why Open Source is a National Security Risk with China

Martin Casado · Jul 28, 2025

Open source is currently a national security danger because China is better at it than the US is, not because of anything the US is doing, and the correct US response is to invest more in its own open source rather than pull back

Chinese open source models are proliferating everywhere and the US has no control over Chinese regulation, so the way to combat software produced by a quasi-adversarial nation state is to be incredibly open ourselves and drive proliferation of our own technology

Scope: danger stems from Chinese, not American, open source

29:05 20VC: a16z's Martin Casado on Anthropic vs OpenAI: Where Value Accrues | Cursor vs Replit vs Lovable: Who Wins and Who Loses | The One Sin in AI Investing | Why Open Source is a National Security Risk with China

Arvind Jain · Jul 11, 2026

It is paramount for the US to build its own strong open source models rather than block Chinese ones, and US open source models will have significant tailwinds

The US cannot be seen as a country that doesn't innovate on technology, and motivated parties like NVIDIA are investing heavily in promoting US open source model development

46:30 20VC: Why OpenAI and Anthropic Won't Win the App Layer | Why Teams Will Get Bigger Not Smaller in a World of AI | Why AI Removes Incumbents Advantage of Bundling | China vs America: Who Wins the AI War with Arvind Jain, Co-Founder @ Glean

Harry Stebbings · Jul 11, 2026

Sam Altman's 5% grant to Trump is a quid pro quo, and will be cashed in as regulatory capture — barriers, taxes and bans on Chinese open models to protect OpenAI and Anthropic

There is no other reason to give away 5%; Altman has a direct line to Trump and can argue Chinese models are a national security backdoor

Scope: framed as a prediction of what will happen

45:36 20VC: Why OpenAI and Anthropic Won't Win the App Layer | Why Teams Will Get Bigger Not Smaller in a World of AI | Why AI Removes Incumbents Advantage of Bundling | China vs America: Who Wins the AI War with Arvind Jain, Co-Founder @ Glean

Harry Stebbings · Aug 3, 2026

The US will impose restrictions on Chinese models because the big American labs will lobby extremely effectively, with Sam Altman as the best politician in the world speaking with intent

The money behind the big American labs overlaps with the people around the table at Mar-a-Lago, and Altman and Dario can coalesce the right group to make it happen

20:17 20VC: 70% of Neolabs Will Die | There Will be a $100BN US Open-Source Model | Data is a Trillion $ Market | Governments Cannot Regulate Models: It is Too Late | The Cyber Attacks to Come Will be Insane with Anastasios Angelopoulos @ Arena

Anastasios Angelopoulos · Aug 3, 2026

A US ban on Chinese models is a real trade-off: it would guard against backdoors and let the American open-source ecosystem capture revenue and flourish, but the bigger cost is crippling American businesses forced to build on the tenth-best open-source model while foreign competitors build on the best

American companies would be building on a worse foundation than rivals in countries that haven't banned Chinese models

17:21 A ban would cripple american builders more than it protects

Martin Casado · Jul 28, 2025

Saying open source increases hostile actors' ability to harm us is tautologically true in the same trivial way that the availability of computers increases it, so it isn't a useful argument on its own

Any increase in availability of general-purpose computing capability trivially increases adversary capability

28:47 The availability argument for restricting open source is trivially true and uninformative

Arvind Jain · Jul 11, 2026

Regulatory capture against Chinese open models won't happen, and open source doesn't need to be curbed right now because US open source is far behind anyway

He believes in the US system, and there's nothing to protect against yet given how far behind US open source is

45:48 No restriction nothing to protect yet

Arvind Jain · Jul 11, 2026 · hedged

Even if Altman and Amodei see Chinese open models as a core threat, they cannot fix that threat through regulation

46:11 Regulation cannot neutralize the open model threat