Accessing the New Compute Markets: Perpetuals vs Traditional Futures

Architect is preparing to launch Compute Futures, a derivatives contract designed to bring price transparency and risk management to companies exposed to GPU price volatility. Built on a suite of compute indexes, these contracts allow participants to trade GPU hours as a hedgeable commodity. To understand the opportunity Compute Futures represents, it helps to understand the two primary structures through which these new power markets can be accessed.

Traditional futures contracts

Traditional futures are financial agreements to buy or sell a specific commodity or security at a predetermined price on a set future date. At expiration, the buyer is obligated to purchase and take delivery of the underlying asset, or the contract is cash-settled.

These instruments serve two core audiences. Hedgers use them to protect against adverse price movements. Speculators use them to take directional views on an asset's price and gauge market sentiment.

The analogy to physical commodities is intentional. Consider a grain operator worried about an oversupply depressing the value of their crop. By entering a futures contract to sell at a fixed price, they lock in revenue regardless of what happens to spot prices. The same logic applies to a company with significant GPU compute exposure: a futures contract offers a way to fix costs and reduce price uncertainty.

Let's look at an example using a neocloud that is exposed to GPU rental price risk. The neocloud has debt financing costs that often span 3–5 years. Spot compute prices fluctuate with cyclical demand, supply shocks, and hyperscaler pricing, and if the price falls too low, revenues will fall and they won't be able to service their debt.

If the neocloud could sell a futures contract, they would be protected from price fluctuations. If the price were to drop, they would realize gains on their futures hedge that would offset their loss of revenue. Similarly, if the price increased, they'd lose money on the futures trade, but the increased rental price would lead to higher revenues.

What distinguishes traditional futures from bespoke agreements like swaps is standardization. Contract terms are uniform, they trade on regulated exchanges, and settlement flows through clearinghouses — institutions specifically designed to manage counterparty risk. This infrastructure gives traditional futures their reliability and the liquidity that customers need to enter positions at fair prices.

Architect's Compute Futures are built exactly on this model, and the futures contracts are based on real-time supply and demand of people buying and selling Nvidia H100 and H200 chips in the spot market. As volumes and participants grow, futures markets not only offer an invaluable risk management tool, but they will also provide more insight into how the price of compute will change over time.

Perpetual futures contracts

Perpetual futures gained popularity with the rise of 24/7 digital asset markets and occupy a fundamentally different design space. The distinguishing feature is that there is no expiration date.

Without an expiration, there is no settlement strike price, and traders can hold positions indefinitely. This makes perpetuals particularly attractive to speculators seeking continuous, flexible exposure. This is why they have become the dominant derivatives instrument in crypto markets.

To prevent the perpetual contract price from drifting too far from the underlying spot price, these instruments use a funding rate mechanism. The funding rate is calculated from the difference between the perpetual price and the spot price.

If there is a gap between the contract price and the spot price, one side of the trade pays the other at regular intervals. If the perpetual trades at a premium to spot, buyers (longs) pay sellers (shorts); if it trades at a discount, shorts pay longs. This continuous funding mechanism is what anchors perpetuals to reality without the hard constraint of an expiration date.

Why this matters for compute

The impending launch of Architect's Compute Futures signals that GPU compute is maturing into a tradeable commodity asset class. Companies building on AI infrastructure now face the same price risk that oil refiners, airline operators, and agricultural businesses have navigated for decades — and the derivatives markets are beginning to respond.

Traditional futures, with their standardized terms, exchange-traded structure, and clearinghouse settlement, are a familiar fit for institutional hedgers. These are data centers, cloud providers, and large AI companies that need to lock in the cost of compute months or years in advance.

Perpetual futures may cater to a more actively managed, speculative audience. These may be traders who want to express a view on GPU prices without committing to a fixed expiration, or who need the flexibility to scale in and out of positions as market conditions evolve.

Both structures are arriving at a moment when the cost of AI compute is one of the most economically significant and least transparent prices in the global economy. The development of liquid, standardized derivatives markets around GPU capacity is not just a financial curiosity but the actual infrastructure to make compute a proper commodity — with all the price discovery, risk transfer, and capital efficiency that entails.