UCPI-A100-SXM4-80GB-LISTED Child Specification#

Status: proposed child specification, version 0.1.0-draft. Not launched. Prepared 14 September 2026. A child of the UCPI-LISTED-GPU family specification under the Urdais Compute Price Index family; everything not stated here is inherited from those documents unchanged. A different economic object from any accessible-offer child.

Instrument#

NVIDIA A100, SXM4 form factor, 80 GB HBM2e, full physical device, per-accelerator allocation class. The A100 SXM4 40 GB, the A100 PCIe 80 GB and the A100 PCIe 40 GB are different instruments and are never collapsed into this one.

Admitted upstream SKUs#

Price of Compute canonical SKU A100-SXM-80GB (the source separates A100-SXM-40GB, A100-PCIE-80GB and A100-PCIE-40GB). Admitted as Grade C identity.

Seller topology evidence, 14 September 2026#

Runpod: A100 SXM 80 GB listed per GPU with per-GPU host resources (Secure Cloud). Verda: 1x A100 SXM4 80GB instance. Hyperstack: A100 SXM 80 GB listed per GPU with per-GPU host resources. Massed Compute: A100 SXM4 (80GB) x1 exists, but the contracting legal entity is evidenced only by a copyright notice; excluded SELLER_LEGAL_IDENTITY_UNRESOLVED until its terms name the entity. Lambda: A100 SXM 80 GB sold only as 8x instances; excluded WHOLE_NODE_REQUIRED. Denvr: A100 SXM sold as eight-GPU nodes; excluded WHOLE_NODE_REQUIRED. Azure: ND A100 v4 eight-accelerator instances; excluded WHOLE_NODE_REQUIRED. Vast.ai: platform aggregate, excluded.

Version History#

0.1.0-draft, 14 September 2026: initial child. No production effective date.