Beijing’s commitment to developing its own chips so as not to depend on the US and its copy and paste of Western technologies are paying off. The large artificial intelligence projects promoted by the Chinese State through national companies are now based on processors designed and manufactured in the country.
Washington’s 7 years of restrictions on the export of advanced Nvidia chips have slowed down the Asian power little and greatly stimulated the domestic industry. Its technology companies, such as Huawei, compensate for the lower power of their processors with node-based platforms where quantity replaces quality. Dozens or even hundreds of processors form a cluster that emulates the operation of a single processor. They are less efficient and are manufactured with processes already outdated in the West, but they manage to satisfy the high processing needs of AI.
DeepSeek, one of the largest Chinese AI companies, with a volume of clients that rivals the North American companies OpenAI and Anthropic, has just partnered with Huawei to use its Ascend processors. The firm assures that these chips, properly grouped with techniques developed by its engineers, can compete with those of Nvidia and in 2031 they will match the best manufacturing processes.
Another Chinese giant with top-level AI models, the Alibaba group, announced two weeks ago the development of a superprocessor, the Zhenwu V900, which will work on AI servers integrated into platforms with thousands of units.
Chinese technology companies, with the support of the State, also want to compete internationally with Nvidia, which has lost the most juicy Asian market and is trying to get Washington to lift its restrictions.