Scale AI CEO says China has quickly caught the U.S. with DeepSeek

The United States may have led China in the artificial intelligence race for the past decade, according to Alexandr Wang, CEO of Scale AI, but on Christmas Day, everything changed.

Wang, whose company provides training data to major AI players including OpenAI, Google and Metasaid Thursday at the World Economic Forum in Davos, Switzerland, that DeepSeek, China’s leading AI lab, released a “earthquake model” on Christmas Day, then followed it up with a powerful AI model centered on the rationale, DeepSeek-R1. , which competes with OpenAI’s recently released o1 model.

“What we found is that DeepSeek … is the best performer, or almost on par with the best American models,” Wang said.

In an interview with CNBC, Wang described the artificial intelligence race between the United States and China as an “AI war,” adding that he believes China has significantly more. Nvidia GPU H100 – AI chips that are widely used to build powerful main AI models – than people can think, especially considering the US export controls.

Wang also said that he believes that the AI ​​sector will reach a trillion dollars, according to estimates that the generative AI market is ready to top $1 trillion in revenue over a decade.

“The United States needs a huge amount of computational capacity, a huge amount of infrastructure,” Wang said, adding later: “We need to unleash the energy of the United States to enable this AI boom.”

Earlier this week, President Donald Trump announced a joint venture with OpenAI, Oracle and SoftBank to invest billions of dollars in US AI infrastructure. The project, Stargate, was unveiled at the White House by Trump, SoftBank CEO Masayoshi Son, Oracle co-founder Larry Ellison and OpenAI CEO Sam Altman. Key initial technology partners will include Microsoft, Nvidia and Oracle, as well as the semiconductor company arm. They said they would invest $100 billion to start and up to $500 billion over the next four years.

In Thursday’s interview, Wang said he believes it will take two to four years to reach artificial general intelligence, or AGI, a widely cited but loosely defined benchmark used in the AI ​​sector. to denote a branch of AI that pursues technology that equals or exceeds human intelligence in a wide range of tasks. AGI is a hotly debated topic, with some leaders saying we’re close to it and some saying it’s not possible. Wang said his own definition of AGI is “powerful AI systems that are able to use a computer like you or could … and essentially be a remote worker in the most capable way.”

Anthropic, the AmazonThe AI-backed startup, founded by former OpenAI research leaders, has stepped up its technology development over the past year, and in October, the startup. he said whose AI agents have been able to use computers like humans can complete complex tasks. Anthropic’s Computer Usability allows its technology to interpret what is on the computer screen, select buttons, enter text, navigate websites and perform tasks through any software and browsing in Real-time Internet, the startup said.

The tool can “use computers practically the same way we do,” Jared Kaplan, Anthropic’s chief scientific officer, told CNBC in an interview at the time. He said that he can do the tasks with “tens or even hundreds of steps”.

OpenAI it is said that the plans to introduce a similar feature soon.

When asked which US artificial intelligence startups are leading the AI ​​race right now, Wang said the models each have their strengths — for example, OpenAI’s models are great for reasoning, while that Anthropic are great for coding.

“The space has become more competitive, not less competitive,” he said.

Correction: This article has been updated to correct the name of DeepSeek’s AI model, DeepSeek-R1.