Overview
On the evening of November 11, an A-share company known for box-type power equipment and charging piles disclosed a centralized procurement contract with an internet company for power cabins at computing centres, with an estimated total value of 250 million yuan. The scope is not a single cabin but a multi-category power-cabin system covering medium-voltage cabins and lithium-battery cabins, spanning medium-voltage distribution, auxiliary transformation, DC uninterruptible supply and storage, with the supplier responsible for overall integration and complete-set delivery. The company also reported cumulative winning bids in the computing-centre segment of over 1.5 billion yuan for January–September 2026, up more than 300% year on year.
Adjacent disclosures the same day point to broader demand spillover: a wire and cable maker won a 2026–2027 centralized procurement contract for ordinary low-voltage power cables from a property group worth 661 million yuan including tax, about 2.88% of its prior-year audited revenue; and the controlling shareholder of a lithium-battery materials company planned to increase its stake using a dedicated share-purchase loan and own funds, in a range of 500 million to 1 billion yuan.
The common backdrop is the packaged-power requirement created by AI computing infrastructure. A recent National Energy Administration meeting called for vigorous construction of new-type grids and new energy-sector business models, with computing-power and electricity coordination repeatedly emphasized. On the industry side, computing centres no longer buy power supply as individual transformers or switchgear panels; medium-voltage interconnection, transformation and distribution, uninterruptible supply, storage and monitoring are bundled into a single rapidly deployable system.
Background & Interpretation
1. Industry Background and Policy Context
Computing centres differ sharply from conventional industrial loads: rack power density keeps climbing, loads fluctuate quickly, continuity requirements are extreme, and build schedules are measured in weeks rather than years. Prefabricated, modular power cabins fit these conditions — assembly and pre-commissioning happen in the factory, leaving only placement and connection on site, which shortens delivery and reduces field-construction uncertainty. On policy, both the computing-power network and the new-type grid sit among the six-network priorities of the 15th Five-Year Plan. Source-grid-load-storage coordination, direct green-power supply and compute-power/electricity coordination require computing facilities to be not merely consumers but adjustable, market-participating load nodes. Recent measures on new-type grids, demand-side flexibility and renewable absorption provide the institutional setting for this business model.
2. Core Drivers and Underlying Mechanism
Moving from equipment sales to turnkey delivery changes three things. First, delivery boundaries: full integration means the supplier owns the interfaces, protection coordination and joint commissioning across medium-voltage distribution, auxiliary transformation, DC uninterruptible supply, storage and monitoring — responsibility expands from component performance to system performance. Second, value composition: system integration, engineering design and commissioning services take a larger share of contract value, so margin structure is no longer set by manufacturing cost alone. Third, customer structure: centralized procurement by internet and cloud players prioritizes delivery speed, availability and full-lifecycle operations, requiring cross-category coordination and repeatable scale. Including lithium-battery cabins in the same contract shows storage evolving from backup power into a constituent of computing-centre supply architecture — carrying uninterruptible supply while reserving an interface for demand response and peak shaving.
3. Market Structure and Industry Chain Impact
For power-equipment makers, computing centres have become a third large-scale demand scenario after renewable plants and grid tenders, and orders are scaling unusually fast. For storage and power-supply firms, entering the computing supply chain puts more weight on high C-rates, fast response and system-level safety certification. For cables, busways, power devices and thermal management, centralized procurement raises requirements for batch consistency and lifts entry barriers. At the same time, packaged systems squeeze purely trading-oriented channels, while suppliers with design, integration and operations capability are better placed to build durable customer relationships. The property-sector cable award also suggests leading equipment firms continue diversifying downstream exposure to cushion capex cycles in any single industry.
Implications & Outlook
Near term, the cadence of computing-centre capex and centralized tender timing is the core variable for order continuity; whether growth above 300% persists depends on sustained AI infrastructure investment. Over the medium term, technology choices — 800 V DC supply, prefabricated power cabins, grid-forming storage and liquid-cooling compatibility — will set how standardized and replicable these systems become. Longer term, if compute-power coordination links further with direct green supply and spot markets, computing-centre power systems shift from cost centres to adjustable assets participating in electricity markets. Risks include receivable and delivery concentration from centralized procurement, cyclical pullbacks in large-customer capex, and evolving safety standards and liability definitions at high power density.
Takeaways for Industry Participants
Equipment makers should extend capability from manufacturing into system design, interface standardization and on-site commissioning, building availability metrics and safety validation for computing scenarios rather than entering on low-priced components while carrying system-level liability. Storage and power-supply vendors should align early with DC supply architectures and demand-response interfaces, making products dispatchable system components. Computing operators writing tenders should specify lifecycle availability, expansion flexibility and operations response — not just upfront price. Component suppliers should prepare for the consistency and certification thresholds of centralized procurement. For the whole chain, the measure of this opportunity is not contract size but whether the supply system stays stable, scalable and settleable as computing loads swing rapidly.
This column compiles industry information and shares technical perspectives; it does not constitute investment advice.