Unlock the editor’s summary for free
Roula Khalaf, editor of the FT, selects her favorite stories in this weekly newsletter.
The reprisals seemed safe. When the United States clenched its control over advanced artificial intelligence technologies in January, blocking China’s access to advanced IA chips and patented models behind commercial barriers, the answer seemed predictable. China would build their own walls, protect their advances and bend in the secret.
Instead, Porcelain He is doing something unexpected: he is giving away his most advanced AI models.
In recent weeks, Chinese technological groups, including Alibaba, Baidu and Tencent, have been flooding the market with powerful AI models. But in an industry where the secret is the norm, the real shock is its opening: these models are free to download, modify and integrate.
The rhythm of China’s open source thrust has been relentless. From the debut in January of Deepseek R1, China’s response to the OPENII O1 series, has followed a wave of increasingly capable models. Alibaba affirms that his latest AI QWQ-32B Rivals Deepseek’s R1 reasoning model has had a good performance in official reference evidence. Every few weeks, another arrives, pushing the limits of what the AI open source can do.
At first glance, this increase may seem a statement that AI should be open to the world, not just a handful of companies. But in business and geopolitics, generosity rarely has no strategy. The real question is not why China is open to its AI, that is why the world assumed that it would not.
For now, most American technological groups treat AF as an exclusive resource, restricting access to their most powerful models behind Paywalls. Operai, Google Deepmind and Anthrope limit full access to their most advanced AI models, offering them through plans such as paying subscriptions and business offers. Meanwhile, the United States government sees the AI open source as a safety risk, fearing that unregulated models can adjust to cybernetics. American legislators are already pressing to prohibit AI Deepseek software of government devices, citing national security concerns.
But Chinese technological groups are adopting a very different approach. At the open supply, not only avoid the sanctions of the United States, but also decentralize development and take advantage of global talent to refine their models. Even the restrictions on the high -end chips of Nvidia become a less obstacle when the rest of the world can train and improve China’s models in alternative hardware.
AI advances through the iteration. Each new launch is based on the latter, refining the weaknesses, the expansion of the capacities and the improvement of efficiency. To open AI models, Chinese technological groups create an ecosystem where global developers continually improve their models, without assuming all development costs.
The scale of this approach could fundamentally remodel AI’s economic structure. If open source AI becomes as powerful as owner American models, the ability to monetize AI as an exclusive collapse product. Why pay for closed models if there is a free and equally capable alternative?
For Beijing, this strategy could be a powerful weapon in the United States and China’s technological war. The US companies, based on monetization through business license and premium services, could be found in a race towards the bottom, where AI is abundant, but it is elusive.
Of course, this comes with compensation. If AI is available for free, nothing will prevent foreign companies from taking China’s models, refine them and exceed Chinese companies. Over time, companies such as Alibaba, Baidu and Tencent can face the same pressures as US counterparts, which forces them to restrict access to protect intellectual property and generate income.
Beyond market dynamics, Beijing can have your own reasons to rethink this approach. The Chinese government, which prioritizes control over key technologies, can also promote stricted AI regulations to manage erroneous information, maintain supervision and guarantee compliance with state policies.
But for now, open source AI remains the best bet of China, a way to compete without access to the best chips or the advantage of an early advantage.
The time of the open source race is not a coincidence. It is an answer to a closing window. With US chips and technological restrictions established to harden under President Donald Trump and the patented models of AI are rooted, the most effective strategy of China is speed and scale. Flood the market, change the balance before monopolies arise.
If Openai, Google and Microsoft have already won the career of AI as we know it, so the best movement in China would not be to compete, it would be that gain meaningless.