Compute is becoming
a financial market.
Tensors is building a market for AI compute,
connecting capacity available today with the infrastructure needed tomorrow.
Future demand.
Future supply.
A market between them.
AI infrastructure takes years to build. The commitments that finance it are often bespoke, hard to price, and harder to transfer.
Make those commitments transferable, and buyers gain flexibility. Developers gain demand they can build against.
Liquidity changes what gets builtFrom a future project
to a present market.
A prepaid commitment’s journey.
Six stages, one physical foundation.
A developer plans a new AI data center. Power, land, and hardware need capital long before the first workload runs.
In this example, buyers prepay for a portion of future capacity, contributing capital toward construction.
Delivery period, region, performance, and service levels become defined contract terms. Large commitments divide into smaller units.
Buyers can hold or transfer their contracts as needs change. The developer retains the original funding; a new buyer takes the future capacity.
Construction progress and delivery timing can change a contract’s value. Project prices can diverge from a broader compute benchmark.
During the agreed delivery window, the buyer receives qualifying compute capacity. Capital becomes infrastructure. A commitment becomes an AI workload.
Future infrastructure
Liquidity changes
what gets built.
From future demand to physical infrastructureA view of the
forward market.
Capacity, place, and time.
The beginnings of a common language
for future compute.
Prices in USD per normalized compute-hour
| Delivery | Compute class | Forward price | Selection |
|---|---|---|---|
| Performance A | $3.20 | ||
| Performance A | $3.10 | ||
| Performance A | $3.00 | ||
| Performance A | $2.92 |
Finance what comes next.
Help shape the market for compute—as a buyer, provider, infrastructure developer, or capital partner.
Talk to us