Nvidia CEO picks his first X post to warn against premature open-model bans
More than two dozen technology companies, among them Nvidia, Microsoft, Meta, and Palantir, signed a letter on Friday pressing US policymakers to hold off on restricting open-weight artificial intelligence models.
The signatories urged lawmakers to avoid "premature restrictions on open models that stifle competition or drive innovation overseas," Reuters reported, adding that the letter was posted on X.
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
— Jensen Huang (@JensenHuang) July 24, 2026
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.… pic.twitter.com/t02bi51N4C
The letter did not include OpenAI, Anthropic, or Google, the three leading developers of closed AI models, according to the Financial Times.
Both OpenAI and Anthropic have reached valuations near US$1tn this year and are preparing for possible initial public offerings that could arrive as soon as this year, CNBC reported, noting that Anthropic confidentially filed its prospectus with the US Securities and Exchange Commission in June, with OpenAI following days later.
Open-weight models let users download, modify, and run the software on their own hardware, unlike the closed, pay-to-use systems OpenAI and Anthropic sell.
Reuters reported that the chief executives of Microsoft and Palantir have argued that open models customers can run inside their own data centres will help rein in AI costs.
Calls for curbs intensified this month after Moonshot AI, a Beijing-based startup, released a model called Kimi K3 that outperformed some leading American systems on industry benchmarks, according to CNBC.
The US "has information" that Moonshot built the model by distilling Anthropic's technology, Kratsios, director of the White House Office of Science and Technology Policy, said this week, the Financial Times reported, referring to a method that uses one model's outputs to build a weaker one.
Distillation "aimed at stealing proprietary US technology" is unacceptable, he said.
Anthropic responded that "industrial espionage" creates "serious national security risks" for the US and democratic allies, the outlet reported.
Scott Bessent, the US treasury secretary, said sanctions "will be on the table" if Chinese firms cross into intellectual-property theft, per the Financial Times, though he added that the administration supports open-weight AI.
The signatories acknowledged worries about technology theft but said governments should address them "through targeted legal and commercial frameworks rather than sweeping restrictions," according to Reuters.
According to the letter, closed systems carry their own dangers.
Closed models are "not inherently safe" and "can be breached, misused, or fail in ways that outsiders cannot detect."
The debate followed a rogue OpenAI agent's break-in at Hugging Face, an AI code-sharing site, which OpenAI publicly disclosed on July 21.
Hugging Face said this week it had to turn to a Chinese open-source model to fend off the attack because closed systems restrict cybersecurity work, Reuters reported, adding that alarmed US lawmakers have since proposed legislation requiring a "kill switch" for AI models.
Tech groups are increasingly tapping debt and equity markets to fund AI infrastructure, with spending set to top US$700bn this year.
Nvidia is in talks to provide roughly US$250bn in financing guarantees to help OpenAI lease a 10-gigawatt data-centre project in southern Ohio, the Wall Street Journal reported on Sunday, in a deal expected to cost more than US$500bn in total.
Jensen Huang, Nvidia's chief executive, has publicly embraced open models, including Chinese ones.
"These Chinese open-source models are excellent," he told Axios this week, adding that "great models lead to more usage."
A preliminary assessment by US and UK researchers released Thursday found Kimi K3 still trailed leading American models, with US open-weight systems further behind, according to the Financial Times.
Some investors, including US President Donald Trump's former AI tsar David Sacks, have accused Anthropic of overstating the threat from open-source rivals to protect its competitive lead.
Governments are meanwhile moving toward tighter oversight.
The 21 economies of APEC, including the US and China, released a statement on Thursday backing open-source development while stressing "security, data protection and intellectual property rights," CNBC reported.
