Did you know that the United States has held onto its position as a leader in artificial intelligence (A.I.) over China for almost a decade? However, Alexandr Wang, the 28-year-old CEO of Scale AI—a startup valued at $13 billion—warns that this lead is now being threatened by a revolutionary A.I. model released by the Chinese company DeepSeek, just in time for the holiday season. In an interview with CNBC on January 23, Wang voiced his apprehensions about the diminishing gap between the two nations in the realm of A.I.
This week, DeepSeek launched a new A.I. model that demonstrates reasoning abilities rivaling those of prominent U.S. firms such as OpenAI. The swift progress of this Chinese startup has not only astonished experts but has also ignited discussions regarding the effectiveness of A.I. chip export restrictions intended to limit China’s access to advanced GPUs essential for A.I. development.
Wang remarked, “The A.I. race and the A.I. conflict between the U.S. and China are among the most pressing challenges of our time.” In a recent advertisement in The Washington Post, he called upon the Trump administration to protect America’s technological advantage as China’s A.I. capabilities continue to advance rapidly. He advocated for increased investments in computing resources and data infrastructure, alongside a strategic energy plan to bolster the A.I. sector.
Originating from Los Alamos, New Mexico, Wang is not just a self-made billionaire; he is also one of the youngest in the world. Born to parents who were physicists at the prestigious Los Alamos National Laboratory, Wang transitioned into the tech industry following brief roles at Addepar and Quora. His passion for machine learning took him to MIT, where he eventually founded Scale AI, a company focused on delivering precisely labeled data for A.I. training through contractual partnerships.
With a valuation of $13.8 billion and a client roster that includes the U.S. Department of Defense and OpenAI, Scale AI has carved out a significant niche in the A.I. ecosystem. Wang’s close friendship with OpenAI CEO Sam Altman, with whom he shared living quarters during the Covid-19 pandemic, significantly influenced Scale AI’s growth trajectory. The company was nurtured within Y Combinator, the startup accelerator formerly led by Altman.
How does DeepSeek’s latest release compare to leading U.S. models?
Scale AI recently partnered with the Center for A.I. Safety to launch “Humanity’s Last Exam,” which is regarded as the most challenging benchmark test for A.I. systems to date. While existing models have struggled to achieve a success rate exceeding 10% on the test, DeepSeek’s new reasoning model, DeepSeek-R1, has emerged as a top competitor. Wang disclosed to CNBC that its performance is on par with the best models coming out of the U.S.
Despite stringent regulations governing GPU exports from China, DeepSeek claims to have attained impressive results with significantly fewer resources than their American rivals. For example, DeepSeek-V3, which debuted in December, was trained using around 2,000 Nvidia A.I. chips, while Meta’s Llama 3.1 model utilized 16,000 GPUs for its training.
Wang remains cautious, noting that Chinese laboratories might possess more H100 GPUs—an advanced Nvidia model not legally available in China—than is publicly acknowledged. He hinted that DeepSeek may have access to about 50,000 H100s, a detail they are unable to disclose due to U.S. export policies.
To drive the progress of cutting-edge A.I. models, Wang emphasized the necessity for enhanced computational capacity and infrastructure within the U.S. He predicts that the A.I. market could reach a staggering $1 trillion once artificial general intelligence (A.G.I.) is achieved, a milestone he anticipates may occur within the next two to four years. A.G.I. refers to systems capable of performing at the level of highly skilled remote workers—an achievement Wang believes is just around the corner.
