🧔♂️ A friendly human may check it before it goes live. More news here
Cambricon posts first annual profit with $305.1m net income
Cambricon, a Beijing-based maker of AI accelerators, said in a statement that it reported its first-ever annual profit with net income of 2.1 billion yuan (US$305.1 million) and revenue of 6.5 billion yuan (US$944.5 million), reversing a 452 million yuan (US$65.7 million) loss the prior year as Beijing pushed domestic adoption amid limits on Nvidia supplies.
Nvidia has secured limited licences to sell H200 chips in China and Bloomberg Intelligence said uncertainty over future supplies has prompted firms such as Alibaba and Tencent to prioritise locally made accelerators.
People familiar with the matter told Bloomberg News that Cambricon plans to more than triple chip production in 2026 and aims to deliver about 500,000 AI accelerators, including up to 300,000 units of its Siyuan 590 and 690 models, mostly using Semiconductor Manufacturing International Corp.’s (SMIC) N+2 7-nanometer process.
Morgan Stanley estimates China’s AI chip market could reach US$67 billion by 2030 with domestic firms supplying more than three-quarters of demand.
🧠 Food for thought
Implications, context, and why it matters.
Cambricon’s profit surge rests on one big customer
- Profit now leans on a small set of buyers, with one customer, ByteDance (the Chinese owner of TikTok), placing more than half of its orders 1.
- Huawei takes up much of Semiconductor Manufacturing International Corp.’s (SMIC) limited 7-nanometer manufacturing capacity, with estimates as high as 70%, which leaves Cambricon less room to scale output 2.
- A target of 500,000 chips looks hard to reach when yields for its most advanced processors on SMIC’s 7nm-class processes run around 20% 1.
US sanctions are reshaping China’s domestic AI chip market
- U.S. export controls aimed at slowing China’s AI work ended up sheltering local suppliers, which helped some money-losing startups move into the black and become national champions 3.
- Forecasts put Nvidia’s share of China’s AI chip market at about 8% after 66% in 2024, while domestic vendors like Huawei and Cambricon together take nearly 80% 4.
- Software is splitting into camps as Huawei backs its MindSpore ecosystem (an AI development framework) and Cambricon tunes its inference toolchain (software used to run trained AI models) 3. That divide may slow efforts to build one AI stack that can challenge Nvidia’s CUDA worldwide 3.
Recent Cambricon developments
🔗 Source: Bloomberg
Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.