Your chips, clearly recorded
Confirm model, quantity, source, ownership and deployment location. Client-owned and purchased chips both require compatibility checks.
Compute services for enterprises. Hosting and asset management for chip owners.
Clients retain ownership; business income is settled as agreed.
July 2026 · Compute measured in FP16
National AI compute
Under national monitoring
National compute hubs
National data-center clusters
Qunxiang Cloud connects chip resources, data-center infrastructure and business demand for compute. Bring eligible new chips or purchase chips under an agreed plan. We organize servers, rack space, power, networking and software for actual computing workloads.
After onboarding, we handle agreed deployment, maintenance and service delivery, with asset records, operating information and monthly statements. You retain ownership. Income is settled under the agreement and depends on actual business; continued hosting or buyback follows the applicable contract.
Confirm model, quantity, source, ownership and deployment location. Client-owned and purchased chips both require compatibility checks.
Organize the agreed servers, facilities, power, networking and software, followed by daily maintenance.
Organize resources for enterprise and AI workloads, including agreed inference and application-running tasks.
Account for actual income and fees in monthly statements. Continued hosting or buyback follows chip source and contract terms.
Start with an inventory identifying the assets, their owner and deployment.
Organize the operating environment and connect acceptance with onboarding.
Provide agreed professional hosting around device and workload operation.
Organize cluster resources around actual workload requirements.
Distinguish business income, fees, client taxes and available balances.
Plan the next asset arrangement under the applicable cooperation documents.
Connect physical chips, hosting and settlement through asset inventories, deployment records, operating status and monthly statements. Service scope, data integration and deliverables follow the agreed plan.
Meet upstream compute procurement needs through partner data centers, organizing GPU clusters, servers and operating environments for project delivery.
A planned platform for enterprises, developers and AI teams, offering agent services, token APIs and compute rental.
A planned ecosystem supporting developer and OPC communities with compute and incubation, connecting product development with joint operations.
For cooperation projects including Xinjiang data centers, GPU resources, servers and operating environments are organized around project requirements.
Eligible new client-owned chips are connected after model, ownership and compatibility checks. Ownership remains with the client; purchase-contract buyback does not automatically apply.
Purchased chips are paid for, allocated, registered, delivered and accepted before onboarding. Allocation and physical delivery take place within 15 working days from the day after full payment; activation is confirmed separately.
Distributable income is consolidated as agreed. Earned income, monthly statements and withdrawable balances are tracked separately. Taxes are listed separately; business records and cooperation documents govern actual settlement.
Client pre-tax settlement
Hosting operations fee
Prior-month statements are issued; clients may request available withdrawals.
Bank payment is made within 7 working days from the day after a complete, valid request.
The operations fee is not charged twice on the same income and is zero when no distributable income exists. Extra repairs or special services require prior written agreement.
Browse regional nodes and the clusters and servers linked to your assets. Global coverage is a planned expansion.
Review chip quantities, models, associated servers and online or maintenance status using connected operational data.
Review business income, statements, available withdrawal balances and withdrawal records, with earned and withdrawable amounts distinguished.
Indicative chip-only budget
Indicative chip-only budget
Indicative chip-only budget
The relevant seller pays 97% of the actual original purchase price, with a one-time 3% depreciation deduction, under the purchase and hosting documents. Clients retain chip rights when hosting continues. This does not automatically apply to client-owned chips.
Discuss nationwide compute, chip hosting, hardware, enterprise AI or hardware recycling.
