Injective launches 10x leverage market for Nvidia GPU rental costs

A new market for AI compute

Injective has launched a perpetual futures market that allows traders to bet on the hourly rental costs of Nvidia's most popular data-center chips. This new financial instrument is tied to the Super Intelligence Compute Index (SICI) and went live on the mainnet on October 26. The market uses USDC as the margin for participants who want to take positions on whether AI computing costs will rise or fall.

The index is designed to reflect the reality of how AI workloads move across different generations of hardware. Instead of tracking just one chip, the basket includes five different models: the H100, B200, H200, A100, and the consumer-grade RTX 4090. By including the 4090, the index accounts for developers who rent smaller cards for open-source models and inference tasks.

This move follows a previous initiative by Injective in August 2025, where they launched a contract tied specifically to H100 rental rates. The shift to a five-chip basket is intended to provide a more comprehensive view of the market. It helps participants understand the broader landscape of availability, performance, and price across the entire Nvidia ecosystem.

How the index is calculated

The benchmark is built and priced by LO:TECH, a capital markets firm based in London. The firm gathers reference prices in dollars per GPU-hour from various compute marketplaces. To ensure accuracy, the methodology pools both on-demand and spot listings while counting each venue only once per chip type.

Before data enters the final calculation, the observations undergo validation and outlier checks. The weights of the index are fixed within a specific methodology version, with a heavy lean toward hardware used for large-scale training. The H100 SXM 80GB makes up 40% of the index, while the B200 accounts for 25% and the H200 holds 15%.

The remaining weight is split between the A100 SXM 80GB and the RTX 4090, each carrying a 10% weight. LO:TECH strikes the index every hour in UTC and uses the 19:00 UTC reading as the daily fixing. The final published figure is expressed as a trailing seven-day average in index points.

Hedging and speculative use

Injective frames this new instrument as a practical tool for hedging against the rising costs of AI development. Developers who are facing higher rental bills can use the market to offset that risk. Conversely, GPU suppliers who are exposed to falling rental revenue can use the market to protect themselves against a downturn in the industry.

The launch comes at a time when investors are questioning if AI cloud providers can generate enough profit to cover their massive infrastructure costs. For example, research firm Gartner recently warned that more than 80% of specialist AI-focused cloud businesses could stop operating independently by 2030 if they cannot generate sufficient returns. This creates a high-stakes environment for anyone relying on rented compute.

The product also serves those looking to speculate on the economic health of the AI sector. As electricity consumption at AI-focused data centers jumped 50% in 2025, the cost of power and hardware has become a primary bottleneck. This market allows traders to profit from the volatility of these underlying costs without needing to own the physical hardware themselves.

The rise of compute derivatives

The launch of the SICI index is part of a growing trend toward linking finance to real-world assets. Injective reports that 47% of its active perpetual markets are now tied to real-world assets across 135 different instruments. These include equities, commodities, and even data from companies like SpaceX, Anthropic, and Perplexity.

Regulatory interest in these types of compute-linked derivatives is also increasing. The Hyperliquid Policy Center recently filed a submission with the Commodity Futures Trading Commission seeking regulated contracts for GPU rentals. They argue that perpetuals can solve recurring exposure issues for data-center operators, lenders, and AI developers alike.

In the broader credit market, hardware is already being used as collateral for loans. For instance, Bullish recently announced a $100 million facility to finance loans secured by GPUs. These developments suggest that the financial infrastructure for AI is maturing, moving from simple hardware sales to complex, derivative-based financial products.

Liquidity and future outlook

While the launch provides a new way to trade compute costs, the success of the market depends on its liquidity. Injective has not yet quantified how much adoption the market saw at launch. For a benchmark to be useful for hedging, it must be deep enough to absorb large flows without causing the spreads to widen significantly.

Traders will also be watching how the seven-day averaging method affects the index during periods of high stress. If rental prices swing sharply due to a sudden shift in AI demand, the averaging might smooth out the data too much for some hedgers. The goal is to provide a stable enough reference price to be useful while still reflecting the underlying market reality.

Moving forward, Injective has indicated that another on-chain finance product will arrive in the coming days. As the market for AI hardware continues to evolve, the integration of compute costs into decentralized finance will likely become a standard way to manage the risks of the AI era.