Great debate is emerging in Silicon Valley due to the proliferation of Chinese-made artificial intelligence tools, especially “open-weight” AI systems that, by some measures, compete with or even outperform some of the best US models. ONOWIZI colleague Hugo Lowell has written about the Trump administration’s internal debate over how to handle these Chinese brands. Among AI companies in the Valley, this issue seems to be very divisive.
A top concern for both DC and Valley is related to stick refinement, where a less skilled AI model is trained on more powerful results. In June, Anthropic accused Chinese tech giant Alibaba of illegally stealing its IP through a distillation attack. Then earlier this week, the White House said it believed Beijing-based Moonshot AI had developed its Kimi K3 model by integrating Anthropic’s Fable 5 model.
One of the biggest concerns is how quickly China’s models are emerging and spreading. An open-source AI model has its core components made public, so it can be fine-tuned to suit the user’s needs. But they lack the kinds of vigilance that Anthropic has built its reputation on. Yasir Atalan, deputy director and data fellow at the Center for International and Strategic Studies, points out that the biggest advantage of open-weight AI models is their speed distribution. They can spread especially easily through “Face Hugging, GitHub, cloud providers, local deployments, and third-party indexing platforms,” he writes. If you’re an Anthropic, and you’ve built a religion around security and payment for access to your big expensive proprietary models, you have good reason to want to control this.
But some Silicon Valley startups — not trillion-dollar ones like OpenAI and Anthropic — don’t really want the US government to put restrictions on these AI models. On Wednesday a group of more than 200 startups called the Little Tech Association sent a letter to Michael Kratsios, science advisor to President Donald Trump, and US Commerce Secretary Howard Lutnick urging against a direct ban on open-source AI models. The group, which includes the popular startup incubator YCombinator, has spoken in favor of some protections but says denying Americans access to AI models abroad would weaken US startups and create dominance among AI giants.
Bill Gurley, a prominent investor and longtime partner at Benchmark Capital, has publicly argued in favor of letting the “free market run its course.” In a long blog that provides a nice little history of open source software, Gurley writes that open-source models avoid lock-in, encourage real academic research, and are essential for compelling startups.
“Every AI startup, every single developer, every two-person team building a product on top of an AI infrastructure depends on access to good models at affordable prices,” Gurley said.
Chamath Palihapitiya, one of the All-in podcasters, wrote on X that “tricking the US Government into protecting the border labs business model by using the China boogeyman is a mistake…It protects the equity of 5,000 investors in OAI and Ant from being sold to everyone else. This would be a very stupid decision.” His partner and VC Jason Calacanis continued. “Father Trump protect us!!!!” he wrote in X, with a terrifying number of laughing crying emojis.
This attitude from some of Silicon Valley’s ruthless capitalists may at first seem counterintuitive. Why allow the technology of a foreign enemy to flourish in the US? It’s as if the US is up 1-0 in the AI World Cup, which is a bit of a stretch, and the crowd is chanting that the opposition has scored a free kick.
