Is China's Open-Source AI Strategy Sustainable Amid Pressure?
Reuters Breakingviews published commentary on March 25 examining the growing tension between openness and profitability confronting China's leading artificial intelligence companies, arguing that mounting investor pressure to generate returns may force a strategic pivot away from the open-source model distribution that has fueled their global rise.
The Case for Going Proprietary
The analysis, authored by Robyn Mak from Hong Kong, noted that making models freely available has helped Alibaba and other Chinese labs drive adoption worldwide, with U.S. startups like the $29 billion coding company Cursor among the beneficiaries. But the piece suggested that the tension between giving away powerful technology and building sustainable businesses is reaching a breaking point, particularly for publicly traded firms under pressure to show returns on billions in AI spending.
The commentary arrives at a moment when the strategic implications of China's open-source push are drawing intense scrutiny in Washington. The U.S.-China Economic and Security Review Commission released a report on March 23 warning that China's open-source dominance has created a "self-reinforcing competitive advantage" that threatens American AI leadership. The commission found that Chinese labs have "narrowed performance gaps with top Western large language models" despite being cut off from the most advanced AI chips.
China's Open-Source Footprint
The scale of Chinese open-source penetration is striking. Alibaba's Qwen model family surpassed one billion cumulative downloads on Hugging Face, overtaking Meta Platforms' Llama to become the world's most downloaded open-source AI system. According to the USCC report, seven of the ten most downloaded models on the platform from November to December 2025 came from Chinese labs. Some estimates cited in the report suggest roughly 80 percent of U.S. AI startups now use Chinese open-source models.
USCC vice chairman Michael Kuiken warned that the advantage could deepen as AI shifts toward physical-world applications. "There's a bit of a deployment gap in the embodied AI space between the U.S. and China. That's something that over time compounds itself," he told Reuters.
The Pressure to Close Up
Yet the Breakingviews piece pointed to forces that could erode China's commitment to openness. Publicly traded companies face shareholder scrutiny over whether giving away cutting-edge models undermines their ability to monetize AI investments. Alibaba has committed around $53 billion to AI and cloud computing, banking on a strategy where free models drive customers to purchase computing power and related services.
Beijing's own instincts could also work against continued openness. The Chinese government maintains strict control over technology through censorship and data security regulations, and as Chinese models approach or exceed Western capabilities, officials may reconsider sharing technology with potential military or cybersecurity applications. China has already restricted exports in sectors where it holds global leads, including rare earth processing technologies and electric vehicle batteries — a pattern that could eventually extend to AI.
