Chinese billionaire Chen Tianqiao’s attempt to build a cross-border artificial intelligence (AI) company has been largely reshaped by increasing regulatory scrutiny and geopolitical tensions between China and the United States. Chen founded MiroMind, a California-based AI start-up, in 2025 with research teams spanning Singapore, Beijing, and Shanghai, aiming to develop foundational AI models for scientific and industrial applications. However, within months, the company altered its operations in response to regulatory pressures.
The shift began after Beijing initiated a review in January 2026 of Meta’s planned acquisition of Manus, a Chinese AI firm that had relocated to Singapore, over concerns about potential violations of China’s export controls. The scrutiny of Manus—which involved a transaction valued at more than $2 billion—was later dropped, but it prompted Chen to shut down MiroMind’s research centers in Beijing and Shanghai and to sever Chinese researchers’ access to the company’s overseas code and data. Chen described the move as a necessary “firewall” given the increasingly sensitive regulatory environment.
Chen’s effort to combine Chinese AI talent with U.S. capital and markets highlights the challenges faced by tech companies caught between rising tensions and tightening national security regulations. According to Chen Qiheng, a non-resident fellow at the Asia Society Policy Institute, companies can no longer take advantage of both countries’ resources without accepting the responsibilities and restrictions imposed by each.
MiroMind initially attracted significant talent, including Dai Jifeng, a prominent AI scientist and professor at Tsinghua University, who was hired to lead the Beijing research team. The company quickly gained recognition by launching an open-source AI agent capable of complex tasks such as predicting Federal Reserve rate decisions and performing well on information retrieval benchmarks like BrowseComp. However, after the Manus investigation, Dai remained in China and went on to establish his own AI start-up along with several former members of MiroMind’s Beijing team.
The departure gave rise to a public dispute between Chen and Dai. Dai accused Chen of trying to relocate technology and personnel from China to the U.S., while Chen alleged that Dai violated agreements by using MiroMind’s code and employees to create a competing venture. Chen expressed his disappointment over the episode and stated he would no longer hire large language model researchers based in China.
Following the setback, Chen intensified recruiting efforts in the United States, targeting experienced AI researchers from leading companies such as Meta and xAI. Despite offering compensation packages reportedly reaching $100 million annually, MiroMind has faced difficulties attracting talent, partly due to its lower profile compared to established Silicon Valley firms.
Chen, who made his fortune founding the online gaming company Shanda in the early 2000s before transitioning into AI and neuroscience investment, has pledged significant financial resources to his AI ventures. He said he has invested around $2 billion of his own capital so far and is prepared to commit an additional $2 billion if external funding remains unavailable.
Recently, Chen shifted his focus from MiroMind to Apolo, a new AI agent designed to tackle demanding research problems in areas such as drug discovery and gene therapy delivery. Launched in June 2026, Chen aims for Apolo to generate $1 billion in contracted revenue within a year and become a market leader in scientific AI applications by the end of 2027.