TL;DR
- Injective launched SICI, an onchain benchmark tracking prices for five Nvidia GPUs, alongside a perpetual market with up to 10x leverage and USDC margin.
- The index gives the H100 a 40% weight, combines GPU data, and publishes a seven-day trailing average with hourly prints and daily fixing.
- Injective says SICI gives traders, developers and infrastructure providers exposure to compute prices while creating potential tools for hedging GPU costs and revenues.
Injective has launched the Super Intelligence Compute Index, or SICI, an onchain benchmark designed to track the cost of renting AI computing power. According to the official announcement, the index is live on mainnet alongside a perpetual market offering up to 10x leverage with USDC as margin. SICI turns a basket of GPU rental prices into a tradable onchain reference, extending the range of markets available across the broader Injective ecosystem.
Injective Expands From Single-GPU Markets to a Compute Index
SICI tracks the hourly rental prices of five Nvidia GPUs at fixed weights: the H100 SXM 80GB represents 40%, followed by the B200 at 25%, H200 at 15%, A100 SXM 80GB at 10%, and GeForce RTX 4090 at 10%. The structure gives the H100 the largest influence while capturing both frontier training and lower-cost inference demand. The launch expands Injective’s existing RWA perpetual market infrastructure beyond equities, commodities, currencies and single-GPU exposure.
The index is built and priced by LO using reference prices gathered from compute marketplaces. On-demand and spot listings are pooled, while observations pass validation and outlier controls before entering the calculation. Each GPU was rebased to 100 at the July 22, 2026 daily fixing. SICI publishes a trailing seven-day average of the weighted hourly basket rather than a direct dollar rental quote, with an hourly index print and a daily fixing at 19:00 UTC.
Injective positions the market as a tool for both speculation and risk management. AI developers can face higher operating costs when compute capacity becomes scarce, while GPU providers can see rental revenues weaken when prices fall. A liquid compute benchmark creates a potential hedge for participants exposed to changing infrastructure costs or revenues. The product also complements Injective’s push toward AI-native financial infrastructure, where autonomous software can access markets and execute onchain transactions.
Injective says 47% of its active perpetual markets are tied to 135 real-world assets across multiple categories. SICI builds on earlier experiments, including Injective’s H100 rental market and its AIX index combining AI-related tokens and equities. The new benchmark broadens compute exposure from one GPU model to five and is available for integration across ecosystem dApps. The launch moves GPU pricing further into the category of financialized onchain infrastructure, while fitting Injective’s broader effort to develop markets that can also be accessed by autonomous AI trading agents.
