Injective launches GPU rental index perpetuals with up to 10x leverage
Injective says its Super Intelligence Compute Index (SICI) perpetual market is live on mainnet. It lets traders take leveraged positions on a benchmark based on rental prices for five Nvidia GPUs. The index could help manage exposure to compute costs, but a live market is not necessarily liquid or a dependable hedge.
- SICI tracks a weighted basket of five GPU rental-price benchmarks.
- Injective says the market uses USDC margin and offers up to 10x leverage.
- Trading activity and key contract mechanics have not been disclosed.
What SICI measures
SICI combines rental-price observations for five Nvidia GPUs: the H100 SXM 80GB, B200, H200, A100 SXM 80GB and GeForce RTX 4090. The H100 has the largest weight at 40%, followed by the B200 at 25%, the H200 at 15%, and the A100 and RTX 4090 at 10% each.
LO:TECH, a London-based capital markets firm, calculates the index from rental listings on compute marketplaces. Prices are stated in dollars per GPU-hour, or the cost of renting one GPU for one hour. The listings include on-demand and spot prices. LO:TECH says each marketplace counts once for each chip, and that it validates observations and checks for outliers.
SICI is a relative index, not a direct quote for the current dollar cost of renting a GPU. LO:TECH sets each GPU’s reference price to 100 at the July 22, 2026 daily fixing, then combines the component levels using the stated weights. It calculates readings hourly in UTC and publishes a trailing seven-day average of the weighted basket. The 19:00 UTC reading is the daily fixing, or designated daily reference value.
A seven-day average can smooth sharp short-term moves, but it may also take longer to reflect a sudden change in current rental prices. That matters to businesses trying to match a hedge to their actual bills.
Some parts of the data process remain unclear. The methodology does not say whether observations are asking prices or completed rentals, how it combines multiple listings from one marketplace, or how marketplaces are selected. Asking prices do not always match the rates customers pay.
Injective says a methodology change would create a separate version of the index rather than rewrite the existing series. LO:TECH says “any print can be regenerated from archived data with an identical result.” That is the firm’s claim about reproducibility. The available description does not say whether the archived data include all original marketplace listings as well as the calculation inputs.
One date also needs clarification. Injective’s launch announcement is dated Oct. 26 but does not specify a year, while the methodology names July 22, 2026, as the rebasing point. Without the announcement year, the chronology cannot be confirmed.
A potential hedge, not a proven one
Perpetual futures are derivatives with no fixed expiry date. Leverage lets a trader control a position larger than the margin they post. Injective says SICI offers up to 10x leverage, with USDC as collateral. Leverage can magnify gains, but losses grow just as quickly. A position that moves against a trader can be liquidated.
Injective describes the market as a potential tool for AI developers and infrastructure providers exposed to changes in compute costs or GPU rental revenue. A developer facing rising rental prices could, in principle, take a position to offset some of the increase. A provider worried about falling rental income might take the opposite position. Whether either strategy works depends on the contract’s payoff mechanics and how closely SICI tracks the firm’s actual costs or revenue.
That gap is called basis risk: the difference between the benchmark used for a hedge and the price or expense a business wants to offset. GPU model, location, rental duration, cluster size and service terms can all affect a company’s bill. Public hourly listings may not reflect reserved capacity or longer-term agreements, which can have different prices and may not be public.
Injective has not disclosed SICI’s trading volume, liquidity, open interest or adoption. The available details also do not explain its funding, settlement, liquidation or index-disruption procedures. Those are no minor details for traders. They help determine what a position costs, how it closes and what happens if the benchmark’s inputs become unreliable.
Compute contracts and GPU-backed finance
SICI follows an earlier Injective effort to build markets around computing costs. In August 2025, Injective introduced an H100 rental-price contract with Squaretower, which supplied hourly price updates. The contract was available through Helix, a decentralized exchange. Injective also dates its AIX index perpetual to January 2025. That product combined AI-related tokens with equities including Nvidia and TSMC.
Interest in compute-linked derivatives is reaching regulators too. The Hyperliquid Policy Center submitted recommendations to the Commodity Futures Trading Commission (CFTC) on Oct. 20, proposing regulated contracts tied to GPU rentals and AI-model operating costs. The submission year is not specified, and no CFTC response or action is established here. The organization’s name alone does not mean the submission came from the Hyperliquid exchange.
The policy center recommended independent price contributors, filters for unusual observations, position limits and safeguards for perpetual funding rates. It also backed dated futures, which have a set expiry, for expenses with a known endpoint, such as a planned training run. A contract expiring when a project ends may fit that cost better than an open-ended perpetual. The right instrument still depends on the exposure and contract terms.
GPU-backed lending is another sign that computing hardware is becoming a financial asset, though it is separate from SICI. Bullish announced a $100 million facility for USD.AI on Aug. 28, but the announcement year is unspecified. USD.AI also announced a $98.1 million loan backed by 2, 304 Nvidia B300 GPUs in June, while investors separately funded a $34 million facility secured by 768 B200 units. The announcement details do not name those investors.
Bullish’s venture arm invested $4 million in USD.AI in September 2025. Bullish and USD.AI also announced plans for sUSDai trading pairs and a market-making program, but gave no listing date or pair details. These announcements point to broader efforts to finance and trade around computing infrastructure. They do not show that Injective’s index market has demand or liquidity.
Key questions about Injective’s GPU index
-
What does SICI track?
It tracks a weighted, relative index of rental-price observations for five Nvidia GPUs. Its value is not a current dollar quote for renting a GPU-hour.
-
How does the perpetual market work?
Injective says the mainnet perpetual market uses USDC margin and offers up to 10x leverage. The available details do not specify funding, settlement, liquidation or index-disruption rules.
-
Can SICI hedge a company’s real compute costs?
Potentially, but that depends on the contract’s mechanics and how closely the index tracks the company’s actual rental rates. The seven-day average and differences in rental terms can create basis risk.
-
How active is the market?
Injective has not disclosed volume, liquidity or open interest, so its adoption and practical value as a hedge cannot yet be assessed.
GPU-linked derivatives could give businesses a new way to manage exposure to a major operating cost. But a benchmark is only useful if its data are sound, it matches real rental agreements and there is a market around it. Until Injective publishes more about SICI’s liquidity and contract mechanics, the launch is a notable piece of financial infrastructure, not proof that businesses can already hedge GPU costs cleanly onchain.
