Chinese firms like Alibaba, Baidu and DeepSeek are open sourcing AI models to bypass US curbs, decentralize development, and tap global talent for refinement
but BoFA sees a correction coming Wency Chen / South China Morning Post : AMD CEO Lisa Su visits China, touting AI chip compatibility with DeepSeek, Alibaba models Callum Keown / Barron's Online : Baidu Stock Jumps as Chinese Market Leaves US in Its Dust The Economic Times : DeepSeek's disruption triggers AI race in China as Baidu, Tencent, Alibaba ramp up efforts Dylan Butts / CNBC : Baidu, once China's generative AI leader, is battling to regain its position Bluesky: Tony Tassell / @tonytassell : Why China is suddenly flooding the market with open-source AI models - column from June Yoon. “If OpenAI, Google and Microsoft have already won the AI race as we know it, then China's best move would not be to compete — it would be to make winning meaningless.” www.ft.com/content/13df...
Context & Ripple Effects
US restrictions had already appeared to push Chinese AI startups toward efficiency and collaboration, rather than simply constraining their progress, in an earlier account of sanctions-driven innovation. Meanwhile, DeepSeek was being embedded by Chinese automakers, smartphone vendors and other companies, demonstrating an existing route from model release to downstream product adoption across China's device and services market.
This makes open release a competitive distribution strategy as well as a development choice: it can expand the pool of contributors and make model availability less dependent on a single company’s proprietary platform.
First-order effects
- Alibaba, Baidu and DeepSeek can put their models in front of developers beyond their own products, inviting outside refinement and integration rather than retaining all development work in-house.
- The move gives customers and developers more Chinese-origin model options at a moment when Baidu is trying to reassert itself in the domestic generative-AI race.
Second-order effects
- Rival Chinese providers, including Tencent and Ant, face added pressure to match model access, cost and developer support rather than compete solely on closed-model performance.
- US AI leaders must contend with a distribution channel that can broaden adoption without requiring users to buy access from the model maker; differentiation shifts toward products, infrastructure and ecosystem support.
Third-order effects
- If this approach sustains adoption, the contest may increasingly turn on who builds the largest developer and deployment ecosystem, not only who controls the most capable proprietary model.
- Export controls aimed at concentrated hardware and company access may have less leverage over AI progress when model development and refinement are distributed globally, though the ultimate impact still depends on access to compute and real-world deployment.
The trend: Chinese AI firms are using open model distribution to turn constrained hardware access into a broader ecosystem and adoption contest.