The Sorso View Compute Index · SVCI
The honest read on AI compute pricing.
Published every Monday. Median hourly rental rates across nineteen GPU cloud providers in five tiers. Every data point traceable to a public source URL with timestamp. Full methodology disclosed in the same place as the number.
Inaugural issue · Coming weeks
The first SVCI publishes the week of our launch, with full methodology and editorial commentary alongside it. Subscribers receive it in their inbox the morning it publishes.
What the Index measures.
On-demand, publicly-quoted, single-instance hourly rental rates for five hardware configurations across the nineteen-provider panel: H100 SXM, H100 PCIe, H200, B200, and A100 80GB. Captured every Sunday evening. Normalized to per-GPU-per-hour. Source URL and UTC timestamp logged for every price.
The most transparent and continuously measurable price signal in the AI compute market — and the reference point against which every other tier is priced.
Read the full methodology →
What the Index does not measure.
Private contract pricing (confidential and customer-specific). Realized enterprise rates (derivable from public financials but not directly observable). Committed-use and reserved pricing (terms vary too much across buyers to support a stable median).
Major operators who do not publish on-demand rates — including Nscale, IREN, IBM Cloud's enterprise tier, and Fluidstack — are acknowledged in editorial coverage but excluded from the Index calculation.
The line between what we measure and what we don't is published in the same place as the number itself.