From Memory Chips to Modular Data Centers: How China Is Building AI Infrastructure
Growing demand for artificial intelligence is reshaping China's semiconductor industry. Manufacturers are moving beyond individual chip development toward integrated solutions combining high-speed memory, computing hardware and automated logistics.
These developments were highlighted at the 23rd China International Semiconductor Expo, IC China 2026, held in Beijing from November 12 to 14.
Faster memory with lower power consumption
ChangXin Memory Technologies (CXMT) showcased an LPDDR6 memory chip that it described as the world's first mass-produced product using the standard. According to the company, the new chip delivers 60 percent higher bandwidth while consuming 20 percent less power than LPDDR5X.
Such improvements are increasingly important as AI applications demand greater computing capacity. CXMT Vice President Luo Xiaodong cited industry forecasts suggesting that global AI computing power could grow at a compound annual rate of 96.1 percent between 2024 and 2031, nearly doubling every year.
The company also reported a 73 percent increase in research and development spending during the first three quarters of 2026, reaching 11.07 billion yuan, approximately 1.66 billion U.S. dollars.
CXMT noted that servers now account for more than half of global demand for DRAM memory.
Computing infrastructure deployed in hours
Another development focused on making AI computing resources more accessible to businesses and research institutions.
Domestic AI hardware company Tecorigin introduced a modular computing container equipped with 32 to 256 Chinese-made AI accelerator cards and high-density liquid cooling.
The company says a 256-card configuration can support more than 20,000 concurrent users.
Its principal advantage is rapid deployment. Traditional intelligent computing centers can require construction periods of around a year, while modular systems can be largely assembled in advance.
Tecorigin representative Chen Xinjie said that up to 90 percent of the prefabrication process takes place at the factory. Standardized servers and energy equipment can be dispatched within 24 hours, with on-site commissioning taking as little as two hours, according to the company.
The solution is intended partly for research organizations and small and medium-sized enterprises that may lack the resources to construct large data centers.
Beyond import substitution
Technological changes are also extending into semiconductor manufacturing logistics.
JD Logistics presented warehouse automation solutions for semiconductor producers. According to the company, chip manufacturing processes are already highly automated, while warehousing and material handling still rely heavily on manual operations.
JD Logistics uses domestically manufactured chips in a substantial proportion of its Langzu-series robots.
Company representative Chen Jia said Chinese-made chips had significantly improved in performance, while domestic supply chains had become more stable and reliable.
The company also plans to deploy its self-developed robots in overseas warehouses, supported by localized services.
The broader economic implications
The technologies showcased in Beijing demonstrate that the expansion of artificial intelligence depends on progress across several interconnected industries.
Faster and more energy-efficient memory supports growing data-processing requirements. Modular computing systems may reduce the cost and complexity of deploying AI infrastructure. Automated logistics can improve efficiency across high-technology manufacturing and supply chains.
Together, these developments suggest a shift from replacing imported components toward developing integrated domestic technological solutions.
However, the performance figures presented at the exhibition are manufacturers' claims. Their wider economic impact will depend on real-world operation, deployment costs and sustained market demand.
As reported by SSTV+, IC China 2026 highlighted Chinese semiconductor and technology companies' efforts to expand domestic capabilities amid rapidly increasing global demand for AI computing power.









