Fengzhushou Co., Ltd. has signed a computing-power service agreement valued at RMB 5,716,992,000, or approximately RMB 5.717 billion, with a customer identified only as D Company. The Shenzhen-listed company, whose stock code is 301382, disclosed the contract in a September 28 announcement filed through CNINFO.
The agreement covers comprehensive computing services over 60 months. Its stated amount is tax-inclusive but excludes cabinet service fees, which will be charged separately. The disclosure is significant for its scale, but the contract is not presented as a guaranteed earnings stream. Fengzhushou specifically said the total is neither an earnings commitment nor a forecast, and that its eventual effect on fiscal-year results remains uncertain.
The deal places a listed company at the center of a long-duration computing arrangement while leaving important commercial details undisclosed. The filing withholds the identities and business information of both D Company, the service customer, and C Company, the separate server-purchase counterparty, citing commercial-secrets protections.
A broad service package with batch-based billing
The service scope includes CPU, memory, GPU, storage and bandwidth, along with related computing-power support services described in the filing as cabinet services. Rather than treating the arrangement as a single delivery event, the parties will deliver computing services in batches and bill those batches accordingly.
Each batch carries a locked service period of 60 months beginning when billing starts. If a batch remains inside that locked period when the overall cooperation period ends, the agreement extends until that batch period is completed. Any extension after the cooperation period would require further negotiation and mutual agreement.
The cabinet component follows a different settlement basis. Fees are to be calculated according to actual usage, cabinet rated power and the number of cabinets that D Company actually activates. That structure means the headline contract value does not include every charge associated with the service. It also makes the disclosed amount difficult to treat as a simple fixed-price measure of future business.
Fengzhushou is the service provider under the agreement, while D Company is the customer. The company said it must complete delivery of all computing services within 22 weeks after signing. The service agreement uses monthly post-payment settlement, with each calendar month serving as the settlement cycle. In practical terms, the company disclosed a requirement to prepare and deliver the contracted resources before receiving monthly payments, creating an execution and working-capital dimension to the arrangement.
The deal comes amid a broader EastFrontier focus on the physical requirements behind China’s AI economy. Earlier coverage of China’s push toward subsea data centers and offshore wind examined how operators are looking beyond conventional facilities as computing demand grows. Fengzhushou’s filing does not identify a data center, workload, AI model or end-user application, so the agreement should not be tied to any particular project or technology platform.
A separate RMB 3.96 billion procurement commitment
Alongside the service agreement, Fengzhushou signed a separate server-procurement agreement with C Company valued at RMB 3,960,000,000, or RMB 3.96 billion, including tax. C Company is to deliver servers in batches according to Fengzhushou’s requirements. The purchase contract is distinct from the RMB 5.716992 billion service contract and should not be treated as part of that amount.
The two agreements also had different approval mechanics. Fengzhushou said both proposals were reviewed and approved by its fourth-term board audit committee and its 18th board meeting on September 28. The service agreement became effective after the parties affixed their official or contract seals. The server-purchase proposal, however, still required consideration by shareholders before it could become effective.
The filing described a funding plan combining internal funds, finance leases and supporting bank credit. For the separate main server-procurement agreement, Fengzhushou estimated that internal funds could represent 5% to 20% of funding and finance-lease financing 80% to 95%. Those figures are estimates for the proposed funding mix, not evidence that the financing has already been raised or secured.
The capital structure matters because the filing separately describes a server-procurement agreement, a 22-week requirement to complete delivery of the contracted computing services, and funding through internal funds, finance leases and supporting bank credit. Monthly post-payment terms could leave Fengzhushou carrying equipment and operating costs before collections arrive. The filing does not disclose cabinet counts, computing capacity, equipment utilization or customer pricing beyond the contract’s stated total and fee treatment.
The filing provides no comparison with Tencent, Alibaba or Baidu and does not identify the customer’s intended use. The supported point is narrower: Fengzhushou disclosed a package of computing resources and related support, not a named model deployment or completed capacity rollout.
Execution risk remains central to the disclosure
Fengzhushou said C Company and D Company are legally existing companies and are not persons subject to enforcement for dishonesty. It also said neither has a relationship with Fengzhushou’s controlling shareholder, actual controller, directors or senior executives, and that the company had no similar transactions with them during the previous three years. These are statements in Fengzhushou’s disclosure, not independent verification of the counterparties’ financial strength or ability to perform.
The company also identified several risks, including financing, delivery, customer default, service operations, cash flow and equipment utilization. Those warnings are material to how the RMB 5.717 billion figure should be read. A signed agreement establishes contractual terms, but it does not establish that revenue has already been recognized, that cash has been collected, that servers have been deployed or that computing capacity is operating for customers.
Fengzhushou said the project could improve its computing-service system, computing-power carrying and dispatching capabilities, and comprehensive service capability. It also said the arrangements could support varied customer scenarios, strengthen customer stickiness and create synergies with its digital-goods, Internet-of-Things traffic, household-market and AI-device businesses. Those are company-stated expectations rather than independently established outcomes.
The filing’s risk language places execution ahead of headline value. Fengzhushou must meet a 22-week delivery requirement, arrange equipment through a separate procurement agreement, manage monthly post-payment settlement and fund the project through a combination of internal resources, leases and bank credit. The customer identities remain confidential, and the disclosure does not provide enough information to assess demand, utilization or counterparty concentration.
That discipline is also consistent with EastFrontier’s earlier coverage of DeepSeek hiring to build computing infrastructure, where the infrastructure question was distinct from claims about model performance. In Fengzhushou’s case, the verified development is the signing and structure of two contracts. Their eventual contribution to revenue, profit and operating scale remains to be determined through delivery, billing, collection and utilization.
For now, the announcement defines a potentially substantial five-year operating framework, not a completed financial result. The company’s own caveat is decisive: actual effects on future fiscal-year performance are uncertain. Investors and customers evaluating the agreement will therefore need to watch the less visible milestones behind the headline, including shareholder approval for the server purchase, delivery within the stated timetable, financing execution, monthly settlement and the eventual use of the contracted computing resources.