For teams who buy, sell and finance compute
Every population labeled, every number dated. Published values are final.
Compute Desk publishes the daily, blended GPU price indexes that compute futures on Nodal Exchange and Architect’s AIX will financially settle against. Compute Signal is the semantic data layer built by the same team, on the same definitions with the same data quality focus. Rental prices, settled transactions, hardware prices and token economics, in one place.
Hardware, rental and token prices in one place.
Compute Signal brings together GPU hardware prices, rental rates, completed transactions and token pricing. In the rental price view, advertised prices, private offers and completed deals are shown separately, making the difference between asking prices and agreed prices clear.
Prices at full granularity
Hyperscaler lists, neocloud asks, private offers and completed deals are labelled separately.
Sample size is visible
The number of observations behind a price is shown, so a small sample can be distinguished from a larger one.
Definitions and dates are transparent
GPU model, region, price type and as-of date appear alongside the data.
Data Covered
Rental prices
Prices per GPU-hour by model,, region, contract type and term, updated daily. Neocloud and hyperscaler prices are shown separately, alongside availability.
Transacted prices
Anonymized aggregates of completed deals, including medians and term curves. Individual deals are not identifiable.
Hardware prices
GPU purchase prices by family, condition and region, with buy-versus-rent comparisons by commitment length and hours of use per day.
Token economics
LLM API pricing, throughput relative to price, and usage trends across models.
Published indexes
A100, H100, H200, B200, B300 as well as family level US benchmarks. The only IOSCO-compliant GPU rental price indexes with a robust benchmark administration methodology.
Coverage today
Vendors Tracked
Rental price observations per day
Settled transaction rows
Hardware listings
Token models
Ask for the cut you need
Ask for H100 availability against listed prices, or the published index against neocloud asks. The model picks the series. The server fetches and fills every number, and the chart engine that draws the rest of the product draws the answer. You do not wait for us to publish the chart you need.
Who is this for
Physical Market
Neocloud and hyperscaler pricing teams repricing an offer against the cohort.
AI labs and inference companies grading a quote before they sign it.
Operators deciding whether to rent the hours or buy the hardware.
Financial
Lenders marking GPU-backed collateral every reporting period.
Funds and analysts entering compute markets who need a dated, citable mark.
Risk teams following the benchmarks that compute futures will settle against.
Both sides
Lenders marking GPU-backed collateral every reporting period.
Funds and analysts entering compute markets who need a dated, citable mark.
Risk teams following the benchmarks that compute futures will settle against.