Chinese semiconductor groups are preparing a concerted manufacturing expansion that would lift the country’s output of artificial-intelligence processors to roughly three times today’s level by 2026, the Financial Times reported, citing people familiar with the plans. The push follows U.S. export restrictions that bar Nvidia’s most powerful GPUs from the Chinese market and has become a priority for Beijing’s drive toward technology self-sufficiency. At the centre of the effort are three new fabrication plants designed to supply Huawei’s Ascend AI chips. The first facility is expected to start production as soon as the end of this year, with two additional lines scheduled to open in 2026. Combined, the plants could eclipse the current output of comparable lines run by Semiconductor Manufacturing International Corp. (SMIC), China’s largest contract chipmaker. SMIC itself intends to double capacity for 7-nanometre chips next year, which would free up more production slots for smaller designers such as Cambricon, Biren and MetaX. Many of these chips are being tailored to DeepSeek’s FP8 computing standard; Huawei’s Ascend 910D and Cambricon’s 690 are among the first processors expected to support the format. The rapid capacity build-out underscores escalating U.S.–China competition over AI hardware. While Huawei and SMIC declined to comment on ownership of the incoming fabs, analysts say the scale of the planned investment indicates sustained state backing as Beijing races to eliminate supply-chain chokepoints and rival Western chip ecosystems centred on Nvidia.