🎙 Listen to a summary of this story
China’s biggest semiconductor foundries are emerging as major beneficiaries of the artificial intelligence boom, with surging demand for computing hardware lifting revenue, factory utilisation and profits even as U.S.-led technology restrictions continue to constrain the country’s access to the world’s most advanced chipmaking equipment.
Semiconductor Manufacturing International Corp., or SMIC, reported second-quarter profit of $479.2 million, more than three times its level a year earlier and nearly double analysts’ average estimate. Revenue rose 36% to more than $3 billion, exceeding market expectations and crossing that threshold for the first time.
The results highlight how the AI investment cycle is spreading beyond the companies designing the most advanced processors. Foundries that manufacture chips for other companies are benefiting from demand for a much broader range of semiconductors used in servers, data centres, networking equipment and power-management systems.
AI boom spreads across the chip industry
SMIC said strong demand was straining available manufacturing capacity, prompting the company to optimise existing production lines and accelerate the introduction of new capacity. Factory utilisation reached 93.7% during the quarter, while the company shipped about 2.9 million 8-inch-equivalent wafers, up 14% from the previous quarter. Average wafer prices increased 5.7%.
China accounted for about 90% of SMIC’s quarterly revenue, demonstrating the importance of domestic customers as Beijing encourages technology companies to replace foreign components with locally produced alternatives. SMIC has also begun raising some wafer prices as demand outstrips available capacity.
Not all of the increase comes directly from the sophisticated processors used to train large language models. AI data centres require an extensive supporting ecosystem of power-management chips, memory, communications components and other semiconductors, many of which can be manufactured using mature or specialty processes.
That distinction is important for Chinese foundries. SMIC can manufacture 7-nanometre-class logic chips, but it remains behind Taiwan Semiconductor Manufacturing Co. and other international competitors at the technological frontier. Yet the AI boom is creating shortages across less advanced categories as leading global manufacturers devote more resources to high-end products.
SMIC executives said earlier this year that overseas customers were moving some orders to Chinese manufacturers because capacity elsewhere was becoming scarce. The company expects third-quarter revenue to increase another 2% to 4% from the second quarter, while gross margins are forecast at between 26% and 28%.
Hua Hong expands its role
Hua Hong Semiconductor, China’s second-largest contract chipmaker, is benefiting from many of the same trends. Its business is particularly exposed to specialty manufacturing processes used for microcontrollers, flash memory, power-management chips and other components increasingly required by AI servers and associated infrastructure.
The company entered the second quarter after reporting a sharp improvement in profitability during the first three months of 2026. First-quarter net profit rose 458% from a low year-earlier base to $20.9 million, while revenue increased 22.2% to $660.9 million. Capacity utilisation reached 99.7%, according to company figures cited by analysts.
Hua Hong had forecast second-quarter revenue of between $690 million and $700 million as it continued expanding 12-inch wafer production. Analysts have pointed to demand for power-management integrated circuits used in AI servers, along with tight supplies of some memory products, as important contributors to its improving product mix.
The company is also becoming more important to Beijing’s ambitions at more advanced manufacturing nodes. Hua Hong Group’s Huali Microelectronics operation has been preparing a 7-nanometre production process in Shanghai, Reuters reported in March, potentially making the group the second Chinese manufacturer after SMIC capable of producing chips at that level.
The development would broaden China’s domestic manufacturing options for AI chip designers, although Chinese foundries still face significant technological disadvantages compared with global leaders. Producing chips at smaller geometries with competitive yields and costs requires highly sophisticated lithography, materials, software and manufacturing equipment.
Self-sufficiency meets export controls
The financial gains therefore carry significance beyond quarterly earnings. Semiconductors have become central to the technology rivalry between Washington and Beijing, with the United States and its allies restricting China’s access to advanced processors and equipment that could accelerate development of powerful AI systems.
Those restrictions have complicated expansion plans but have also intensified China’s effort to build a domestic semiconductor supply chain. Beijing has directed substantial state and private investment toward chip design, fabrication equipment, materials and manufacturing capacity, while Chinese technology companies have increasingly turned to domestic suppliers.
China has recently made progress in one of the industry’s most difficult areas. A state-backed Chinese company has begun production of domestically developed immersion deep-ultraviolet lithography equipment, Reuters reported in July, with initial machines expected to be delivered to manufacturers including SMIC and Hua Hong this year. The equipment remains behind systems produced by Dutch market leader ASML, and achieving reliable high-volume manufacturing remains a major challenge.
The stakes extend beyond China’s domestic market. Chinese customs data showed integrated-circuit exports reached 179.44 billion units worth $177.28 billion during the first half of 2026, nearly doubling in value from a year earlier. Exports of computers, servers, memory products and other data-processing equipment and components also rose sharply as global spending on AI infrastructure accelerated.
For SMIC and Hua Hong, that combination of domestic localisation and global demand is producing unusually favourable conditions. AI is increasing consumption of cutting-edge processors while simultaneously tightening supplies of the less glamorous chips needed to power, connect and control computing systems.
The companies nevertheless face substantial risks. New factories require billions of dollars of investment, depreciation costs are climbing and restrictions on advanced foreign manufacturing equipment could make it difficult to keep pace with overseas competitors. SMIC spent about $3.4 billion on capital expenditure during the first half of the year as it expanded production.
For now, however, high utilisation and rising prices suggest Chinese foundries have more orders than readily available capacity. Rather than simply demonstrating that China can manufacture more of its own chips, the latest results show that the global AI investment boom is giving its domestic semiconductor industry a powerful commercial incentive to do so.
