O1.ai, the Chinese artificial intelligence startup founded by AI pioneer Kai‑Fu Lee, is preparing to raise funds ahead of a potential initial public offering in 2027. The move places O1.ai among a growing group of AI labs that have either already gone public or are planning public listings.
From frontier models to infrastructure
The company has shifted away from building frontier models toward developing enterprise AI infrastructure. Lee attributed this strategic change to last year’s “DeepSeek moment,” which he says reshaped the cost structure of training AI systems. According to Lee, training large models from scratch is viable only for a handful of companies with effectively unlimited resources, prompting others to adopt different approaches. For O1.ai, that approach has become infrastructure and service offerings aimed at business customers.
International clients and stance on the US market
Lee says the strategy has produced tangible results: international clients now represent about half of O1.ai’s business. At the same time, he ruled out aggressively pursuing other companies in the United States, citing widespread wariness toward Chinese software among US entities.
Why this matters
O1.ai’s shift illustrates a broader trend in the AI industry: as the economics and scale challenges of training frontier models have become clearer, some firms are reorienting toward infrastructure, services, and deployment rather than competing to build base models from scratch. O1.ai’s fundraising and its aim for a 2027 IPO reflect how companies are adjusting their business models to the evolving technical and market landscape.
(Reporting based on Tasneem Nashrulla)
