Confidential GPU RFP generator

Turn workload, capacity, security, and residency requirements into a consistent supplier brief. Review the generated language with your engineering, security, procurement, and legal teams.

Generated supplier brief

CONFIDENTIAL GPU INFRASTRUCTURE REQUEST

Capacity
- GPU: NVIDIA H100
- Quantity: 8 GPUs
- Region: Europe or United States
- Initial term: 3 months
- Required start: As soon as possible
- Workload: Private AI inference

Required supplier response
1. Identify the exact GPU SKU, form factor, count per node, topology, host CPU, system memory, storage, and network.
2. State current available quantity, data-center location, delivery date, billing basis, minimum term, setup fees, and normalized price per GPU-hour.
3. Identify the supported CPU trusted execution environment: Intel TDX or AMD SEV-SNP.
4. State whether production GPU confidential-computing mode is enabled for this exact configuration.
5. Provide the supported firmware, VBIOS, driver, guest image, and confidential-computing compatibility references.
6. Describe CPU and per-GPU attestation, evidence format, verifier responsibility, freshness checks, reference measurements, and device-membership policy.
7. Describe how keys or secrets are released only after a successful verifier decision, including fail-closed behavior.
8. Identify the physical processing location, backup locations, contracting entity, subprocessors, deletion procedure, and available audit or compliance evidence.
9. Describe SLA, support, maintenance, replacement-device, scaling, and incident-response procedures.
10. Separate provider-confirmed facts from planned, optional, or unsupported capabilities.

Please attach source documentation and a current commercial quote.

Continue planning