Cambricon posts first annual profit as China's AI chip boom gathers pace

Cambricon has ended a year with a net profit for the first time in its history, and for China's AI hardware market this is more than a strong earnings report. The company has shown that Beijing's push to replace imported AI accelerators with domestic ones has started to turn into real revenue and market share.

The first profitable year

On March 12, 2026, the company published its results for 2025: net profit came to 2.1 billion yuan, whereas a year earlier Cambricon had recorded a loss of 452 million. Revenue over the same period jumped from 1.2 billion to 6.5 billion yuan. For a manufacturer long seen as one of the main local contenders for Nvidia's niche in China, this is a turning point. It is no longer about hopes for the future but about proven business economics.

• Net profit: 2.1 billion yuan
• A year earlier: a loss of 452 million yuan
• Revenue grew from 1.2 billion to 6.5 billion yuan
• This is the first profitable year in the company's history

Why demand grew

The main reason is a sharp shift in purchasing by Chinese AI companies. Because of restrictions around Nvidia and uncertainty over future deliveries of advanced accelerators, big players such as Alibaba and Tencent are looking ever more closely at local alternatives. At the same time, Beijing is openly nudging the market toward Chinese chips, turning technological self-sufficiency from a political slogan into a practical procurement strategy. For Cambricon this opens a rare window: demand is driven not only by the market but also by state industrial policy.

"This shift reflects growing geopolitical restrictions on imports of

advanced semiconductors and the strategic importance of computing power for large language models."

China's new five-year plan, which names accelerated development of AI chips as a priority, gave the market additional momentum. Against this backdrop, investors and customers have started to judge local manufacturers differently: not as a fallback, but as the foundation of future infrastructure. For the sector itself, this is an important psychological reversal. Not long ago, Chinese accelerators were discussed in terms of "catching up"; now the conversation is about who will take the largest share of domestic demand.

The next stage of growth

The question is no longer whether Cambricon can sell chips, but whether it can scale production in time. According to sources close to the company, in 2026 the manufacturer wants to more than triple its output of AI accelerators and ship about 500,000 units, including up to 300,000 of its most advanced Siyuan 590 and 690 models. Production is planned to rely primarily on SMIC's N+2 process, which is usually described as China's 7 nm-class node.

The ambitions are twofold. On one hand, Cambricon is trying to capture part of the niche freed up by China's limited access to Nvidia products. On the other, it has to win share from Huawei, which is also rapidly strengthening its position in AI infrastructure. Morgan Stanley estimates that China's AI chip market could grow to 67 billion dollars by 2030, with more than three quarters of shipments supplied by local companies. If this forecast comes true even in part, Cambricon's current report looks not like the peak of a cycle but an early signal of a much larger reshuffling of the market.

What this means

Cambricon's first annual profit shows that China's AI ecosystem is starting to be built not only around models and clouds, but also around its own hardware. For the market it is an important marker: restrictions on Nvidia did not freeze demand but redistributed it in favor of local manufacturers. If Cambricon delivers on its production plan and holds its margins, it could become one of the key beneficiaries of the next phase of China's AI boom.