Huawei-led team claims it post-trained DeepSeek's 1.6-trillion-parameter model — 1,000 Ascend 910C chips used in training
A research group that includes Huawei Technologies says it completed full-parameter post-training of DeepSeek's V4-Pro, a 1.6-trillion-parameter model.
This development comes as China continues its aggressive pursuit of AI self-sufficiency following US export controls on advanced chips and AI technology.
It demonstrates China's increasing capability to not only develop but also successfully leverage its domestic AI hardware for advanced model training, directly challenging the Western lead.
The ability for Chinese entities to post-train large, complex AI models using domestic silicon reduces reliance on foreign compute and accelerates the development of a sovereign AI stack.
- · Huawei
- · DeepSeek
- · Chinese AI sector
- · Ascend chip ecosystem
- · US AI chip manufacturers
- · Western AI model developers (in China market)
- · Cloud providers reliant on Western chips
Domestic Chinese AI models will become more sophisticated and prevalent.
Increased investment and pressure on Chinese semiconductor companies to scale production of high-performance AI chips.
Further geopolitical bifurcation of the global AI ecosystem, leading to distinct US-aligned and China-aligned AI stacks.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at Tom's Hardware