Cambricon revenue more than doubles on strong demand

Cambricon Technologies, a China-based AI chipmaker, said that January-to-March revenue rose to 2.9 billion yuan (US$421 million) from 1.1 billion yuan (US$162 million) a year earlier.

Net income increased to 1 billion yuan (US$148 million) from 356 million yuan (US$52.1 million).

The company is among Chinese suppliers benefiting from Beijing’s push for semiconductor self-sufficiency as US export curbs have limited China’s access to advanced AI chips from Nvidia and AMD.

Cambricon competes with Huawei in the domestic AI chip market.

Both are on the US entity list, which restricts access to American technology and manufacturing services from Taiwan Semiconductor Manufacturing Co.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

Cambricon’s growth comes with a broader business change

Cambricon’s quarter fits a wider run. It posted its first full-year profit in 2025 since joining the Shanghai Stock Exchange’s STAR Market, a Nasdaq-style market for Chinese technology companies, in 2020 1.
The profit came with more sales reaching the market. Research and development spending fell to 18% of revenue from 91% because revenue rose faster than research and development spending, not because spending was cut 2.
Scale still matters. Nvidia made US$44 billion in a recent quarter, far above Cambricon’s 2.9 billion yuan, or US$402.7 million 3.

Export controls, factory limits and software shape China’s AI chip push

Cambricon is on the US Entity List, a trade blacklist that limits access to American technology and manufacturing services from companies such as Taiwan Semiconductor Manufacturing Co. That has pushed Chinese chipmakers toward domestic capacity at Semiconductor Manufacturing International Corp (SMIC), China’s biggest contract chip manufacturer.
SMIC has struggled to expand advanced production because yields on 7-nanometer-class processes remain low. Some analyses put them below 30%, versus 70% to 80% at Taiwan Semiconductor Manufacturing Co (TSMC) 4.
Chinese chip firms are also trying to cut reliance on Nvidia’s CUDA-centered software stack, the programming tools used to run AI workloads on Nvidia chips. Moore Threads’ MUSIFY aims to translate CUDA code to its own platform 4.
Some Chinese customers run dual-stack setups. They use domestic chips for some AI inference, which is when trained models generate answers or predictions, or for lighter workloads, while turning to Nvidia where available for the most demanding training 4.