Andreessen Horowitz partner says open source AI is ‘China’s toy now’ — and that’s a problem for the US and its allies

“What was not great was the speed at which the open source ecosystem accelerated the race geopolitically,” Medha told an audience at the Fortune Global Forum in Riyadh. “It does not look good for the United States and its allies.”

“The reality is that if you look at the most powerful open source models today – with the exception of the Mistral model from France and a couple of small, niche models from the US – it’s really China’s game now. And I think that’s not a particularly encouraging picture,” he added.

China’s progress in open source is largely due to DeepSeek, the Chinese startup behind the R1 model. That sparked a sharp sell-off in US technology stocks at the start of the year after investors realized the model was built at a fraction of the cost of frontier US models, but outperformed or matched many of them in key benchmarks.

The popularity of the R1 and the company’s other recent developments have demonstrated China’s prowess in AI innovation and intensified concerns in Washington about how open source development could alter the global balance of power.

DeepSeek has continued to be among the world’s most innovative AI companies when it comes to finding new ways to improve AI models and extract high performance from smaller, less expensive models. they The latest hack has been found Having models that process information as visual symbols rather than linguistic symbols can make the model ten times more efficient.

Leading AI companies, including OpenAI, have changed their stance on open source AI following the launch and market reaction to DeepSeek’s R1. In August, the leading AI lab released two open-weight language models called gpt-oss-120b and gpt-oss-20b, Designed to serve as less expensive and accessible alternatives to their frontier models. CEO Sam Altman teased these models in March, two months after he admitted, in the wake of the success of DeepSeek’s open source R1 software, that the company was “On the wrong side of history“When it comes to opening up its models to developers and builders.

Medha said he believes the West will return to open source AI innovations, and credited the Trump administration’s recent “AI Action Plan” as paving the way for American entrepreneurs and researchers. Medha previously campaigned against Regulating artificial intelligence at the state level, In particular the NY RAISE Act, arguing that a patchwork of regulation across the country would hurt the United States at a time when competing nations are racing ahead.

“Researchers who have the skills to push the boundaries forward should spend their time pushing the boundaries of capabilities, not navigating 50 different pieces of legislation,” he said. “I think what you’ll see as a result of all that free mind space for entrepreneurs and scientists is that three, four, five months from now, you’ll start to see a wave of open-weight models coming out of US labs.”