The Brief
China's expanding market for artificial intelligence computing is drawing legacy industrial powerhouses, telecom carriers, and energy providers into the AI supply chain. State-owned telecom operators have rolled out token subscription packages, while hardware provider Lenovo has transitioned toward computing services. In parallel, energy firms and legacy manufacturers such as Hangzhou Iron & Steel and Litong Electronic are leveraging their electrical substations, land reserves, and power access to build and lease large-scale intelligent computing clusters.
Why it matters
The entry of heavy industry, energy suppliers, and telecom carriers into computing power shifts AI infrastructure from an internet-cloud monopoly into an asset-backed physical commodity. Companies with direct access to land, power grid connections, and telecommunications networks can lower infrastructure deployment costs, accelerating the integration of intelligent computing into the broader industrial economy.
China context
Under national strategies promoting new productive forces and the East-to-West Computing initiative, 'computing-electricity coordination' has emerged as a core focus. Industrial provinces and state enterprises are repurposing retired blast furnaces, industrial parcels, and existing power substations into high-density data centers, facilitating cross-regional computing dispatch from resource-rich western regions to eastern demand centers.
Editor's View
EDITOR'S VIEW — Analysis and inference, not factual reporting.
The commercial pivot of legacy manufacturers and state utilities highlights that China's AI infrastructure bottleneck is fundamentally physical and energy-related rather than purely algorithmic. Access to high-voltage power substations and industrial land has become a critical competitive asset. However, as hardware depreciates and low-cost token packages trigger price competition, these non-traditional computing providers will face tests regarding utilization efficiency and software ecosystem integration.
What to watch
- Pricing competition among telecom operators following the introduction of entry-level AI token subscriptions.
- Policy developments and grid support mechanisms governing cross-provincial computing-electricity scheduling.
- Sustained quarterly margins and capacity utilization rates for industrial manufacturers operating leased GPU clusters.
Key Takeaways
- 1China Mobile, China Unicom, and China Telecom have entered token retail, with China Mobile's intelligent computing revenue surging 279% in 2025.
- 2Energy operators like Jingneng Group and GCL Energy Technology are developing 'computing-electricity coordination' projects linking multi-regional power and data nodes.
- 3Legacy manufacturers, including Hangzhou Iron & Steel and Litong Electronic, are converting industrial land and substations into profitable AI data center hubs.
China's AI computing boom is no longer the exclusive domain of internet hyperscalers and dedicated artificial intelligence startups. According to a report by People's Daily, traditional state-owned telecom operators, energy firms, and legacy manufacturing corporations are rapidly repositioning themselves as key suppliers of intelligent computing capacity and AI token subscriptions.
In May 2026, China's three major telecom carriers—China Mobile, China Unicom, and China Telecom—launched mass-market token subscription plans, offering entry-level packages priced at 9.9 yuan (approximately $1.40) for 10 million tokens. Financial disclosures indicate that China Mobile's computing power service revenue approached 90 billion yuan in 2025, with intelligent computing revenue expanding by 279 percent year-on-year. For 2026, China Telecom has earmarked 25.5 billion yuan for computing infrastructure, representing 35 percent of its total planned capital expenditure. In parallel, hardware manufacturer Lenovo launched token subscription services for enterprise clients, with AI-related revenue reaching 33 percent of total sales during its 2025/2026 fiscal year.
Energy companies are similarly converting their electrical infrastructure into computing businesses. The Beijing Public AI Computing Platform, managed by a digital subsidiary of state-owned Beijing Energy Holding, is operating near capacity by scheduling computing resources across nodes in Beijing, Inner Mongolia, and Ningxia. Private clean-energy operator GCL Energy Technology has also transitioned toward computing power leasing, providing services for municipal governance clouds, healthcare systems, and tech enterprises. He Baohong, chief engineer at the China Academy of Information and Communications Technology, noted that computing-electricity coordination facilitates a structural loop between power networks and digital grids.
For traditional heavy industries, the transformation allows the monetization of legacy physical assets. Following the closure of its steel blast furnaces in Hangzhou, state-owned Hangzhou Iron & Steel Group repurposed 1,743 mu (about 116 hectares) of industrial land, factory buildings, and electrical substations to build two cloud data centers, contributing to a return to profitability for its listed unit in the first quarter of 2026. Similarly, precision metal parts manufacturer Litong Electronic, which entered the computing leasing market in 2023, now manages over 38,000 PFLOPS of computing capacity, posting a 1,088 percent net profit surge in 2025.