Huawei ramps AI chip revenue to $12 billion on internal demand; Chinese fabs strain to keep up as Nvidia's share in region declines.

Huawei expects revenue from its AI processors to reach roughly $12 billion in 2026, up from $7.5 billion last year, representing at least 60% year-over-year growth. The projection, based on orders already received from major Chinese technology firms including Alibaba, ByteDance, and Tencent, would position Huawei as the dominant supplier in a domestic AI chip market that Morgan Stanley estimates could reach $67 billion by 2030. This surge has coincided with Nvidia CEO Jensen Huang confirming that Nvidia's share of the Chinese AI accelerator market has collapsed to zero percent. Just 18 months ago, Nvidia supplied the vast majority of AI training and inference silicon used by Chinese cloud providers, but today Huawei's Ascend 950PR is the primary procurement target for China's largest tech companies, with a training-focused successor named the 950DT scheduled for Q4 this year.

The rapid shift can be largely attributed to the release of DeepSeek's V4 LLM in April, which has been optimized specifically for Huawei's Ascend architecture and its CANN software framework rather than for Nvidia's CUDA ecosystem. Huawei engineers, per reporting from South China Morning Post, collaborated directly with DeepSeek ahead of the model's launch, and the company confirmed that its full Ascend SuperNode product line was adapted for V4 inference on day one. Alibaba Cloud and Tencent Cloud both deployed V4 services within hours of release. The 950PR is currently the only Chinese-made AI processor that supports FP8, a compressed numerical format that allows more operations per second and lowers per-query costs. V4 uses a Mixture-of-Experts architecture with up to 1 trillion total parameters but activates only around 37 billion per inference pass, favoring the 950PR's strengths in inference-efficient hardware. DeepSeek gave Huawei early optimization access but did not extend the same to Nvidia or AMD, and the collaboration has pulled forward procurement timelines across the Chinese cloud industry, with chip prices for the 950PR reportedly rising by about 20% as a result of demand.

Huawei's ability to fill those orders depends on SMIC, China's leading foundry, which manufactures the 950PR on its N+3 process, a 7nm-class node built without EUV lithography. Huawei is targeting production of roughly 750,000 950PR units this year, with full-scale shipments expected in the second half following samples shipped to customers in January, though that figure is expected to fall short of demand. SMIC has been expanding its advanced-node capacity for more than a year, with a goal of a five-fold increase over two years that will lift 7nm and 5nm production to 100,000 wafers per month and half a million by 2030. The combined capacity for 22nm and below could rise from 30,000-50,000 wafer starts per month in 2025 to 50,000-60,000 or higher this year. However, yields remain a thorn in China's side, with SMIC's 7nm-class process delivering substantially fewer good dies per wafer than TSMC's equivalent nodes, and SMIC's cycle time from wafer start to finished and packaged processor is around eight months compared to three months for similar nodes at TSMC.

Huawei's supply chain challenges extend beyond foundry production. In September, Huawei announced it had developed its own HBM chips with up to 1.6 TB/s bandwidth, HiBL 1.0 and HiZQ 2.0, in partnership with CXMT, though how quickly CXMT can ramp competitive HBM production remains an open question. Meanwhile, Nvidia faces regulatory obstacles in the Chinese market. Jensen Huang stated during an interview with the Special Competitive Studies Project's "Memos to the President" podcast that "In China, we have now dropped to zero," criticizing U.S. export policy as having "already largely backfired" and arguing that conceding a market the size of China doesn't make strategic sense. The H200, which Nvidia received U.S. licenses to sell to China earlier this year, has not shipped a single unit despite receiving orders. Contradictory regulatory requirements from Washington and Beijing have created a stalemate at customs: U.S. regulators require H200 chips ordered by Chinese customers be used only inside China, while Beijing has instructed domestic technology companies to limit Nvidia hardware to overseas operations.