According to Alexandr Wang, CEO of Scale AI, the U. S. has been ahead of China in the artificial intelligence competition for the past ten years, but a significant shift occurred on Christmas Day. At the World Economic Forum in Davos, Switzerland, Wang stated Thursday that DeepSeek, the premier AI lab in China, unveiled a "groundbreaking model" on Christmas, followed by the introduction of DeepSeek-R1, an advanced reasoning-focused AI model that rivals OpenAI's new o1 model. "DeepSeek. . . is performing at the top level, or is approximately equivalent to the leading American models, " Wang noted. In a CNBC interview, he referred to the competition between the U. S. and China as an "AI war, " insisting that China possesses far more Nvidia H100 GPUs—crucial AI chips for developing top-tier models—than many realize, particularly given the current U. S. export restrictions. Wang is optimistic about the future of the AI industry, predicting it will reach a trillion-dollar valuation, aligning with projections that the generative AI market could exceed $1 trillion in revenue over the next decade. "The United States will require immense computational resources and infrastructure, " he stated, adding later, "We must harness U. S. capabilities to drive this AI revolution. " Earlier this week, President Donald Trump announced a collaborative project with OpenAI, Oracle, and SoftBank to invest billions in U. S. AI infrastructure. The initiative, called Stargate, was revealed at the White House by Trump, SoftBank CEO Masayoshi Son, Oracle co-founder Larry Ellison, and OpenAI CEO Sam Altman.

The initial technology partners will include Microsoft, Nvidia, Oracle, and semiconductor firm Arm, with a commitment to invest $100 billion to start and potentially up to $500 billion over the next four years. During Thursday's interview, Wang estimated that it could take two to four years to achieve artificial general intelligence (AGI), a debated but significant milestone in the AI field that indicates systems with human-like cognitive abilities across various tasks. Opinions vary on the proximity of achieving AGI, but Wang defined it as “robust AI systems that can utilize computers similarly to how humans do. . . essentially functioning as remote workers at a high level. " Anthropic, an AI startup supported by Amazon and founded by former OpenAI researchers, accelerated its technology development over the last year. In October, it announced that its AI agents could manage computer tasks as humans do. Anthropic's Computer Use feature enables its technology to interpret computer screens, click buttons, enter text, navigate websites, and complete tasks across different software, alongside real-time internet browsing. "The tool can utilize computers in much the same manner we do, " said Jared Kaplan, Anthropic's chief science officer, during a CNBC interview. He indicated that the AI could handle tasks involving "tens or even hundreds of steps. " OpenAI is also expected to roll out a similar capability soon. When queried about which U. S. AI startups are currently leading the race, Wang remarked that each has unique strengths—OpenAI is known for its reasoning abilities, while Anthropic excels at coding. "The field is becoming increasingly competitive, " he affirmed. Correction: This article has been updated to accurately reflect the name of DeepSeek's reasoning-focused AI model, DeepSeek-R1.