China's AI capabilities are rapidly advancing, with Zhipu going so far as to start targeting the full-stack AI infrastructure. Meanwhile, the overarching US policy is largely in disarray, with the second US AI security chief now having resigned in the past three months or so. It is, therefore, high time indeed that the US and China meet to lay down some ground rules for this future-critical race, as is now scheduled to take place in September.
The US and China are now scheduled to meet in September to hammer out ground rules for the ongoing AI race at a time when China's AI labs such as Zhipu and Moonshot are showing breakthrough capabilities
According to Reuters, the US and China will hold direct talks in September to figure out a way to mitigate the risks and threat vectors emanating from each other's frontier AI models, with the US side slated to be led by Treasury Secretary Scott Bessent.
Significantly, these talks will likely take place before President Xi's scheduled visit to the US on September 24, and focus on regulating frontier AI models that can be weaponized via cyber attack vectors.
Also, the US is likely to broach the subject of China's open-source models, which are rapidly approaching frontier-level capabilities but with markedly lower inferencing costs. Left unchecked, these open-source models can dramatically reduce the demand for AI models from OpenAI and Anthropic, imperiling ongoing model development and trillions of dollars of planned CapEx spend.
An emerging thorn in this arena is the proclivity of China's AI labs to distill frontier US models, especially given the fact that there is no easy way of preventing this. After all, Chinese engineers often take their model-laden hard drives to US-friendly countries to imbue their models with frontier-level capabilities by using reinforcement learning techniques on cutting-edge NVIDIA GPUs.
One possible solution might be for Chinese AI labs to agree to freeze their progress so as to maintain their current 9-month lag relative to the frontier models in the US. But, this idea is a hard-sell for China.
Meanwhile, Zhipu (Z.ai), the creator of GLM-5.2 model, has now acquired XCore Sigma, which has previously worked with Huawei's Ascend and the designer of AI ASICs, Cambricon. Combined with a 1GW data center that Zhipu has reportedly brought online, the acquisition suggests that the AI lab is now targeting full-stack capabilities, ranging from compute to multi-chip optimizations, compiler suite, and inference engine. Do note that China will likely ship a whopping 2.5 million units of homegrown AI chips in 2026!
About the author: Writing is my one incontrovertible passion. Over the past six years, he has authored over 2,200 distinct articles on financial and tech-related topics, spanning nearly 1 million words. And he has been a member of Wcctech mobile team since 2025. As an alumnus of the University of Toronto, Rotman Commerce Program, I bring nuance, in-depth knowledge, and a unique perspective to every topic that I cover. When I'm not writing, I'm traveling the world, exploring hidden confectionaries and restaurants as an aspiring food connoisseur.
Follow Wccftech on Google to get more of our news coverage in your feeds.
