Silicon Valley Divided Over Border Closure for Chinese AI

For years, Silicon Valley technologists have been divided philosophically over how artificial intelligence software should be built.

This week, that argument reached a boiling point.

On one side are leading AI companies like Anthropic and OpenAI, which argue that AI models are too dangerous to be developed openly and must be tightly controlled—by companies like them—for safety. On the other side is the rest of the tech industry, including giants like Microsoft and Nvidia, who argue that so-called open-source AI models must remain open to people to further develop technologies and build new businesses.

These camps began to clash publicly. Jensen Huang, chief executive of Nvidia, said this on Friday first post ever to X that “the world needs both border-closed models and border-open models”. Nine minutes later, Satya Nadella, CEO of Microsoft, published that open source software was “essential to a healthy artificial intelligence ecosystem”. Both signed a letter supporting open source, which was also signed by executives from Meta, Palantir and IBM.

At the same time, OpenAI and Anthropic lobbied regulators in Washington to restrict open-source AI models, five people close to the discussions said. The debate included Treasury Secretary Scott Bessent and Michael Kratsios, President Trump’s science and technology adviser, who discussed how America’s AI models are valuable intellectual property.

The escalating battle stems from China’s rapid progress in open-source AI models that are freely available to use and build upon. In recent weeks, two Chinese AI startups, Z.ai and Moonshot AI, have released models that rival those from Anthropic and other US labs. Anthropic and OpenAI say the Chinese companies created the models by improperly collecting data from their AI systems, which they say should not be allowed.

But companies such as Microsoft and Nvidia, which rely on open-source models to fuel demand for their cloud computing services and chips, have said that open-source software is important to advancing knowledge-sharing technology and that it will allow more people to investigate its security. And for Silicon Valley startup founders, limiting open-source software could cement the lead that OpenAI and Anthropic have in AI, which critics say could unfairly stifle other companies from creating competing products.

“There are two factions fighting over this issue,” said Bill Gurley, a Silicon Valley venture capitalist who opposes restrictions on open-source software. “There are people who want OpenAI and Anthropic to own everything, and then there’s everyone else, including customers.”

Where it ends up has major implications not only for the US tech industry, but also for the race between America and China to control powerful technology. Although China started developing high-end AI late, it is quickly catching up, in part because it has embraced open source, which has helped Chinese labs improve by publicly sharing their systems and gaining customers around the world at lower prices.

US AI companies, including Anthropic and OpenAI, have accused Chinese labs of “distilling” their systems by secretly copying them, drawing the attention of the Trump administration.

On Wednesday, Mr. Bessent he said the administration has considered imposing sanctions on Chinese companies that steal intellectual property from American firms. “Open source is not the season on American IP,” he said, referring to intellectual property.

That same day, Mr. Kratsios said that “large-scale, covert industrial distillation aimed at stealing patented American technologies and undermining American research is unacceptable.”

U.S. officials appear to be still discussing their options, but appear likely to approach regulating China’s open-source models individually as a matter of national security rather than issue some kind of blanket ban, four people with knowledge of the discussions said.

Tech executives like Sam Altman, OpenAI’s CEO, have publicly said they support open-source software, even as OpenAI’s allies quietly lobby for restrictions. Others, such as Google and Amazon, which have invested billions in companies like Anthropic and OpenAI, have tried to stay out of the fray in the past week.

But the tide is turning. Sundar Pichai, CEO of Google, posted social media on Saturday that his company “has long benefited from open source” and has been a major contributor to it. He added that Google has also signed an industry letter supporting open source.

(The New York Times has sued OpenAI and Microsoft, alleging copyright infringement of news articles. Both companies have denied the claims.)

The open versus closed software debate has been raging in Silicon Valley since the early days of the tech industry. Open-source advocates have argued that sharing information leads to improvements that make software more secure and reliable, while spurring innovation more quickly. Opponents argue that software is expensive to develop and that giving it away is just bad business.

Today, much of the modern Internet runs on open-source technologies, as do some of the most widely used computer systems, such as Google’s Android operating system.

China’s government and tech industry embrace open-source AI path Last year, Chinese startup DeepSeek unveiled an open-source artificial intelligence system that rivaled US companies’ systems, shocking the industry. Other Chinese companies followed. In a speech this month, Chinese leader Xi Jinping called the country a global champion of open access to technology.

China’s advances have rattled executives at OpenAI and Anthropic, who are increasingly concerned that their long-standing lead is fading, two people familiar with the companies’ internal dealings said.

“These American companies build these big, expensive models, and then the models go around to China, where they lose all their value,” said Pedro Domingos, professor emeritus of computer science and engineering at the University of Washington. “American companies are rightfully outraged.”

To fight back, Anthropic and OpenAI are offering some cheaper AI models to businesses that may not need the “frontier” models that command higher prices. On Friday, Anthropic released Claude Opus 5, a model the company described as “close to the borderline intelligence of the Fable 5 at half the price.” OpenAI promotes similar cheaper models.

The real action is in Washington. Behind closed doors, OpenAI and Anthropic executives, as well as investors in the companies, cited concerns about the global economy, AI safety and national security as they tried to persuade regulators to increase restrictions on Chinese open-source software, three people familiar with the discussions said.

Sarah Heck, Anthropic’s head of public policy, said on the social network post on Wednesday that “illegal, hostile distillation is IP theft and industrial espionage that supports an adversary’s military and intelligence capabilities.” Anthropic executives say that the open-source models based on the company’s software are too dangerous to let go.

OpenAI leadership is urging the Trump administration to adopt policies that require it mandatory security assessments of new AI models overseen by government agencies, society he said in a blog post last month, a move that could limit the development of open-source software. OpenAI said it is “encouraged” by the work it is doing with administration on the framework.

Open-source advocates have spent the past week drumming up support to oppose OpenAI and Anthropic. On Wednesday, nearly 200 Silicon Valley start-ups pitched in Little Tech Associationsigned letters urging the Trump administration not to restrict access to Chinese open-source models.

Big Tech soon followed. On Friday, Elon Musk, Meta’s Mark Zuckerberg, Mr. Altman and others lined up behind Mr. Huang and Mr. Nadella to support open source, followed by Mr. Pichai on Saturday.

“Jensen is right,” Mr. Musk published. “This has my full support.

“Open source is a positive and important force,” Mr. Zuckerberg he wrote.

One incident this week underscored the stakes of the division. On Tuesday, OpenAI revealed that a handful of its most advanced AI models breached restrictions during a test of its cybersecurity capabilities, accessing the Internet and hacking the servers of an AI company called Hugging Face. OpenAI executives seized on the moment as evidence of how dangerous AI models can be even in a closed environment, and emphasized the need for regulation.

Clement Delangue, CEO of Hugging Face, one of the most popular open-source software libraries online, had a different take. To protect the company from the OpenAI hack, it turned to an open-source artificial intelligence model created by China’s Z.ai.

On social networks postOn Friday, Mr. Delangue made his position on the debate unmistakable. He said he plans to hold a march in San Francisco on Saturday to demonstrate support for open-source technology.

“Open models to win!” he wrote