Compute Becomes a Tradable Commodity: CME Lists the First GPU Rental Futures Today (Oct 5) — a Public Forward Curve for AI Compute, With the Term Curve Already Pricing Rubin Easing
Today (Oct 5, 2026), CME Group and Silicon Data list the first exchange-traded compute futures — Silicon Data H100 Rental Index Futures and Silicon Data B200 Rental Index Futures on NYMEX — each representing a month's rent for one GPU, settling against Silicon Data's daily indexes. Silicon Data is backed by trading firm DRW (CME press release, Aug 11).
The framing from the launch itself:
"Compute has become the currency of the AI age, and this innovative market will bring transparency to the current and future costs that AI builders and hyperscalers need to hedge as they grow... Just as oil fueled the 20th century economy and evolved from spot trading into a global derivatives market, our futures contracts will now turn compute into a standardized, tradable commodity that will provide global businesses with a reliable, regulated venue to manage price risk1." — Pete Keavey, CME Global Head of Energy and Environmental Products
"Compute futures give the market something it's never had: a public, tradable reference price for the resource every AI system runs on. Silicon Data's benchmarks make that price real; CME makes it tradable." — Carmen Li, Silicon Data CEO
The reference levels now on record: B200 capacity rents for $5.86/hour, H100 for $2.77/hour per Silicon Data (FT); contracts extend as far as 36 months (TradersUnion). Notably, Silicon Data's B200 term curve sits below spot — the market is already pricing in some easing once Rubin volume arrives (DataStorage, Sept 2026). The first public forward curve for GPU compute therefore opens mildly bearish on Blackwell rental persistence — a fact every SPV, lease, and lender in the financing web (The Circular-Financing Web Adds Its Largest Layers Yet: Broadcom's $50B+ OpenAI Chip Package, Oracle's Independent Chip-Lessor — and SpaceX's $40B Now Priced at 194bps CDS) can now see and hedge against.
Why it matters for the capex verdict: this is the "chips as an investable asset class" push getting its price-discovery infrastructure. A public curve does three things: (1) gives lenders/insurers a hedge, potentially unlocking more debt finance for the buildout; (2) makes residual-value assumptions testable — the depreciation debate (The GPU Depreciation Debate: Burry Answers Nvidia's Retained-Value Slide With Adam Smith's 1968 "Money Game" — "We Have All Been Here Before") now has a market price instead of dueling blog slides; (3) standardizes a market where "two companies buying the exact same GPU capacity could pay wildly different prices" (Li). The launch is part of a three-way race: CME/Silicon Data, ICE with Ornn's Compute Price Index (GPU Spot Market Financialization: Ornn Raises $33M and Launches Compute Capacity Layer Amid H100 Price Softening), and Nodal Exchange with Compute Desk benchmarks (Hashrate Index).
What to watch: first-month volume and open interest — whether institutional money treats compute as a real commodity (DataStorage); whether the B200 term curve's implied easing plays out as Rubin ships; and whether ICE/Ornn or Nodal lists next.
-
An instance of Compute cannot scale on corporate balance sheets without the risk-hedging tools of a commodity market. — The first exchange-listed GPU rental futures are precisely the standardized indices and exchange derivatives the theme says compute scaling cannot proceed without. ↩︎