• Sam Altman, OpenAI CEO, describes China's AI advancements as "remarkable" and warns the U.S. is underestimating its progress in areas like inference capacity and open-source models.
  • China is scaling domestic chipmakers and seeing AI stock rallies, while OpenAI pursues profitability with new revenue streams like ChatGPT ads.
  • Altman questions the long-term effectiveness of U.S. semiconductor export controls, echoing concerns from industry leaders as China pivots to domestic technology.

A Shifting AI Landscape

OpenAI CEO Sam Altman recently called China's AI progress "amazingly fast" and "remarkable," cautioning that the United States underestimates it in critical areas such as inference capacity, research, and open-source models. According to people familiar with the matter, Altman's comments came during internal discussions about global competition, where he highlighted China's rapid scaling of domestic chipmakers and AI stock rallies as signs of a resilient innovation ecosystem.

"What we're seeing in China is a level of momentum that's hard to ignore," Altman noted, pointing to advancements that place some Chinese efforts near the technological frontier. He specifically questioned the effectiveness of U.S. semiconductor export controls, suggesting they may harm American firms' market access more than they curb Chinese innovation. This stance aligns with warnings from Nvidia (NVDA)'s CEO and others in the industry who see China pivoting aggressively toward self-reliance.

The Open-Source Surge and Market Dynamics

China's AI landscape has been energized by the surprise release of DeepSeek R1 in January 2025, a near state-of-the-art open-weight model that sparked what some call a "DeepSeek moment," igniting competition from firms like Z.ai, MiniMax, and Moonshot. These developments mirror ChatGPT's initial boom in the U.S. but with a focus on open-weight models that rival proprietary versions on single GPUs. OpenAI itself has responded by releasing gpt-oss-120b and gpt-oss-20b models partly to counter this growing dominance, according to sources close to the company.

Financially, OpenAI is pushing for profitability through new revenue streams, including ChatGPT ads and subscriptions, aiming to balance rapid growth with sustainable economics. Users still prefer paid U.S. models for superior speed and intelligence despite cheaper Chinese alternatives, though adoption hurdles slow diffusion. Meanwhile, China's domestic chipmakers, such as Huawei with its Ascend 910C matching Nvidia's H100, are gaining traction, boosting self-reliance amid ongoing restrictions.

Implications and Future Outlook

Altman's warnings come amid heightened U.S.-China AI race tensions, with potential incentives for Chinese government support of open models to build international influence. In the short term, experts predict more Chinese open-model labs emerging, with consolidation likely later, while the U.S. maintains edges in model quality and speed via scale from giants like Google (GOOGL) and OpenAI. AI breakthroughs in discovery and problem-solving are expected by 2026, as Altman previously forecasted a "step change" for business innovation.

Long-term, the race is seen as multi-layered rather than having a simple leader, with export controls likely to fail and spur further Chinese innovation. Stakeholders, including businesses leveraging AI for discovery and users navigating model choices, face a rapidly evolving market. Attempts to reach OpenAI for additional comment on Altman's remarks were unsuccessful, but industry insiders say the dialogue reflects a broader recognition of China's accelerating role in shaping global AI trends.

Correction: An earlier version of this article misstated the timing of DeepSeek R1's release; it occurred in January 2025, not late 2024.