CME Group Launches First AI Compute Futures Contracts, Turning GPU Rentals Into Tradable Commodity

AI computing power is becoming a tradable asset class as CME launches futures contracts

Computing power is emerging as a new tradable asset class, with CME Group set to launch the first futures contracts tied to the cost of running the chips that power artificial intelligence. The exchange is partnering with Silicon Data to introduce two compute futures contracts on Oct. 5, pending regulatory approval, giving companies and investors a way to trade and hedge the price of AI computing capacity much as they do oil, electricity and other commodities. "For years, two companies buying the exact same GPU capacity could pay wildly different prices with no way to know who got the better deal. They will now have a benchmark to check that against," Carmen Li, CEO of Silicon Data, said in a statement. "Compute futures give the market something it's never had: a public, tradable reference price for the resource every AI system runs on." The contracts will allow buyers and sellers to trade against the rental cost of Nvidia's H100 and newer Blackwell B200 graphics processing units and will be based on Silicon Data indexes that track hourly GPU rental prices. Each contract will represent a month's rent for the Nvidia H100. The launch comes as Wall Street is finding new ways to finance and gain exposure to the enormous AI infrastructure buildout. Nvidia is working with some of the world's largest asset managers on an effort that could channel as much as $500 billion into AI infrastructure. Compute futures would add another layer to that emerging financial ecosystem. Rather than investing directly in data centers, chips or the companies building them, investors could gain exposure to the price of the underlying computing capacity itself, while AI developers and data-center operators could use the contracts to hedge their costs or revenues.

Nvidia GPU rentals become a tradable commodity on 5 October

From 5 October the cost of renting an Nvidia chip becomes a tradable commodity, listed on the same exchange as crude oil. It also becomes a public number that anyone can read. Renting an Nvidia H100 for an hour costs whatever your supplier says it costs. There is no published price. Two companies buying identical capacity can pay very different rates, and neither of them will know. That changes on 5 October. CME Group, the exchange that lists crude oil and corn, will start trading two futures contracts tied to the hourly rental price of Nvidia chips, subject to regulatory review. Each contract represents one month's rent for a single GPU, CNBC reported. One tracks the H100, the chip most AI systems run on today. The other tracks the Blackwell B200 that succeeds it. It is a small product with a large implication. The input every AI company depends on is about to get a public reference price, and a forward curve showing what traders think that price will be next year. What is actually being traded Nobody takes delivery of a graphics card. Both contracts are cash-settled against indexes published by Silicon Data, a New York firm that has spent two years tracking what GPU capacity actually rents for. They will be listed on NYMEX, the CME's energy exchange. That detail is not incidental. The executive quoted in the launch announcement is Pete Keavey, the CME's global head of energy and environmental products. He made the comparison explicit. Oil fuelled the 20th century economy and grew from spot trading into a global derivatives market, he said, and these contracts "turn compute into a standardised, tradable commodity". The buyers and sellers are easy to picture. A data centre operator owns servers and earns rental income, so it can sell futures to lock in revenue. An AI developer pays those rents, so it can buy futures to cap its costs. Wall Street built the financing layer first This is the second piece of machinery to arrive in a fortnight. The first was Nvidia recruiting six of the largest names in finance for a $500bn funding package aimed at the AI buildout. Silicon Data chief executive Carmen Li drew the distinction herself, speaking to Bloomberg. The Nvidia announcement is a financing layer. This one is a risk management layer. A market of that size, she argued, cannot function without somewhere to hedge and discover prices. That is a fair description of what the last month has looked like from the outside. Enormous sums have been committed to compute with almost no public information about what compute costs. Lenders have already been improvising around the gap. Lambda sold a $917m leveraged loan backed by chips whose residual value nobody could independently price. OpenAI went hiring a power-trading lead because electricity, at least, already has a market. Gavin Baker of Atreides Management, who led the funding round announced the same day, put it in farming terms. Futures markets let farmers finance next season's seed and equipment rather than guess, he said. "You can't build against a price you can't see or lock in." The company selling the price just raised $30.5m Silicon Data announced an initial closing of $30.5m on the same day, led by the Valor Atreides AI Fund. It raised $4.7m in March 2025, so the round is roughly six times the size of its seed 17 months later. The money funds four things: the pricing benchmarks, an institutional data business, risk infrastructure for derivatives and credit, and a performance product called SiliconMark. Li told Bloomberg the ambition is to be "the independent referee" for the compute stack. SiliconMark is the one worth understanding, because it exists to fix a problem that could break the futures contracts. Two clusters built from identical chips do not deliver identical output. Networking, topology and configuration change what you actually get. That is a problem for anyone hedging. If the index tracks a standard hour of H100 time and your cluster underperforms it, the hedge stops matching the exposure. Silicon Data says normalising for performance also opens the door to physical delivery later. Almost everyone in this market owns a piece of the referee The investor list deserves reading carefully, and it is public. CME Group itself invested. So did DRW, Jump, Wintermute and Tectonic, all trading firms. So did VanEck, F-Prime, Samsung and Further. In other words, the exchange holds equity in the company whose index its contracts will settle against, and several likely participants hold equity in it too. Li said as much on Bloomberg, describing those investors as heavily her clients at the same time. This is not unusual in commodity benchmarks, where the firms that need a price often fund the people who publish it. It is worth stating plainly all the same. A benchmark is only as good as its independence, and the market it serves is currently very small. Regulators have views on this in other markets. Europe has policed financial benchmarks under dedicated rules since the Libor scandal, precisely because a reference price shapes contracts far beyond the people who set it. Compute pricing has not attracted that attention yet. What the curve will show before the earnings do The most useful thing Li said had nothing to do with her product. Asked what would tell her the AI buildout had tipped into oversupply, she named three indicators, and all three become visible once this market exists. The first is spot prices. A sustained fall means demand is weakening or supply is outrunning it. The second is the shape of the forward curve. It currently slopes upward, which means buyers are paying a premium to lock capacity in for longer. The third is residual value. If used servers start fetching less on the secondary market, that is the market marking down its expectations of what those machines will earn. Her own reading of the data is not bearish. A100 and H100 rental prices have risen about 20% since January, she said, and have been broadly flat for the past 20 days. Older chips keep earning because smaller models still run on them. Depreciation, she argued, is not the same as worthlessness. Ships and aircraft depreciate too, and their residual value is still a function of the cash they can generate. Why this matters in Europe European AI companies have spent this year raising against compute they have contracted but not yet earned from. Nscale is preparing a US listing that values it at $51bn, with most of that revenue still ahead of it. A published forward curve changes how those numbers get argued about. Contracted compute revenue can be checked against what the market expects compute to be worth when the contract runs. The same applies to the fight over building the capacity at all. More than 500 US jurisdictions now restrict or ban data centres, and the economic case for each one rests on assumptions about future compute prices that nobody could previously see. The caveats are real None of this works without liquidity. A futures contract with few participants produces a price that is easy to move and hard to trust, and this one has to bootstrap a market from nothing. The launch is also still conditional. Both contracts are pending regulatory review, and the CME and Silicon Data first announced the partnership on 12 May, nearly five months before the date they have now set. And a public price cuts both ways for the people who wanted one. Compute buyers get to stop negotiating blind. Compute sellers, including several companies whose valuations assume rental rates hold, get a number the market can disagree with in real time. Oil got its benchmark in 1983, more than a century after the first well. Compute is getting one about three years into its boom. That is either a sign of a market maturing unusually fast, or of how much money is now betting on a price nobody could previously check.

The currency of the AI age

Why it matters: A compute futures market would bring more transparency, new ways to manage volatility and even more investment into what's quickly becoming a critical part of the economy. The big picture: Futures markets allow companies to hedge against swings in prices. If you're an airline, you can buy oil futures to manage costs. A farmer might buy soybean futures to hedge against how well his crop does that year. * Historically, creating a more transparent market for a commodity has unlocked more investments. Plenty of traders and investors in oil and soybeans never own the underlying asset. The latest: The exchange operator CME Group said Tuesday that it plans to launch two compute futures contracts on Oct. 5, "pending regulatory review." The group is working with Silicon Data, a company that publishes indexes tracking compute pricing. * Each futures contract would represent a month's worth of rent for the Nvidia H100 or its newer Nvidia Blackwell B200, per the release. * The idea is to turn compute into a "standardized, tradable commodity," Pete Keavey, global head of energy and environmental products at CME Group, said in the release. State of play: Another company, New York-based OneChronos, is also working on a compute futures marketplace and is awaiting approval from federal regulators. * The company expects it to be operational this year, CEO Kelly Littlepage tells Axios. Reality check: The big problem with treating AI compute like a commodity is that it simply is not like a commodity, Littlepage says. A bar of gold is basically like any other bar of gold, but compute is different. * "Just from an Econ 101 stance, it actually fails every measure of being a commodity since it's not fungible. It's not storable. It's not transportable. There's actually nothing about it that you would consider a commodity." Zoom in: The startup's futures market is using what's known as "combinatorial" auctions to deal with the issue. And it's working with Paul Milgrom, an economist who developed a similar system to auction wireless spectrum -- and won a Nobel Prize for his work. Between the lines: The big players in the AI market have been talking about compute as an "investable asset class" -- literally the title of the Nvidia CEO's post Monday night about the $500 billion financing deal he's ginned up with six major financial firms. * At the Milken investor conference in May, BlackRock CEO Larry Fink had a similar pitch: "I actually believe a new asset class will be buying futures of compute," he said. "We just don't have enough compute power right now." The bottom line: AI compute is fast becoming the metaphorical currency of our time. Now, the race is on to make it more like a literal currency.

Kalshi's CEO is racing to build a futures market for AI's most precious resource, which could be worth $100 trillion by 2030 | Fortune

When the price of jet fuel skyrocketed at the outset of the Iran war, it scrambled the business outlook for airlines -- but not all of them. It turned out carriers like Lufthansa had purchased hedging contracts that ensured that over 80% of their upcoming fuel purchases will be locked in at pre-war prices. Today, the growing mass of companies that consume huge amounts of compute -- which many describe as the new oil -- likely wish they had a similar option to hedge against fluctuating costs. They may soon have one. According to Kalshi CEO Tarek Mansour, compute -- a term that describes the chips and electricity powering the AI revolution -- will eclipse oil as the world's most valuable commodity, and spur a futures market for hedging it. On a recent TBPN podcast, Tarek predicted that compute will be a $10 trillion industry by 2030. He added that, if compute follows the pattern of derivatives markets for other commodities, its futures market will grow to 10-15 times the size of the underlying spot market -- meaning compute futures will one day be worth $100-$150 trillion. If Mansour's prediction is even remotely correct, compute futures represent a massive opportunity for whoever can build that market. In July, Kalshi itself announced a new series of events contracts and data tools that it says can be the foundation of a compute derivatives market. Kalshi, though, isn't the only firm looking to seize that opportunity. The derivatives giant CME Group revealed in May that it plans to roll out a product later this year in partnership with an AI data firm, while stock exchange giant Intercontinental made a similar announcement the same month. But even as compute futures represent a huge opportunity, the history of commodities markets shows the process for building such products can be slow, complicated and uncertain. Here's how that process is likely to play out in the compute field. How exactly do you hedge compute costs? In the 1970s, a series of economic shocks jolted oil markets, wreaking havoc on industries like airlines, trucking and tourism, where profit margins are closely tied to the price of fuel. This spurred a demand for a futures market in oil like the ones that had long existed for corn and precious metals, and that can provide a hedge against sudden price fluctuations. Despite this demand, when it came to oil, the financial companies that sold futures for other commodities faced a challenge: Unlike wheat or gold, there was no consensus on what constituted a standard barrel. Over time, the financial markets did come to agree on standards like Brent (named for a type of oil first pumped from a North Sea field of that name) and West Texas Intermediate, but it took years. This is the situation that now confronts companies looking to hedge their spending on AI compute, where hourly prices can fluctuate as much as 137% over the course of a year, according to research firm Allium. The volatility is especially high when it comes to renting compute powered by the newest models of computer chips, whose availability is frequently subject to manufacturing and supply chain constraints. According to Kalshi's Head of Research, Nicole Kagan, this is a big reason why the financial industry, which has long offered hedging tools for other commodities, has been slow to develop a futures market for compute. "Compute is very different in that it's very opaque. The way compute is priced is via B2B executed contracts from suppliers like Nvidia directly with corporations like HP, which then go and sell them on. So it's very difficult to even understand what the expected pricing is on that thing, right?" said Kagan. Kagan added that the task of developing hedging tools for compute is more difficult still because it is not a commodity like oil or wheat whose basic properties don't change. In the case of compute, newer chips result in higher efficiency -- but the change in efficiency is hard to predict, which complicates any attempt to predict future prices of compute. Despite these challenges, Kalshi believes its prediction markets provide a way to do just that, and is currently listing wagers for five types of chips. Like every other contract it lists, bettors are invited to take one side in a yes/no outcome. For instance, Kalshi users can currently bet on whether the average hourly cost to rent Nvidia's H200 chip over the course of August will be above or below $5. As of August 6, the cost of buying the yes side of that bet was 30 cents and, as with every contract, would pay out $1 if a bettor is correct and $0 if not. To determine the final outcome of the bet, Kalshi relies on a firm called Ornn, which publishes a popular dashboard that shows the cost of renting various hardware. Kalshi is also using its chip-related prediction markets to produce so-called forward curves, which rely on past data in order to plot price movements as far as a year out. The company published some of these curves, which suggest the cost of compute will be fairly constant, in a recent report. The report also describes other prediction markets that could provide signals about the future cost of compute, including markets related to geopolitics or to the cap-ex spending of giant AI users like Google and Meta. The coming battle for a multi-trillion dollar market Kalshi is not the only prediction market service offering compute-related contracts. Its main rival, Polymarket, recently launched similar products but, so far, the two firms' combined offerings are not even a fraction of the oil futures market, where over $70 billion of futures contracts trade hands daily. "Kalshi's GPU rental markets recorded $4.4M in notional volume through July 27 against $285K on Polymarket, roughly 15 times more. Kalshi also covers five chips, compared with Polymarket's three. The positions traders are still holding are equally small, with about 517K contracts open on Kalshi and about 130K outcome shares on Polymarket, backed by $125K in cash," the research firm Allium reported in July. Allium's report also notes that Kalshi's contracts have yet to deliver much insight into where prices are heading. For instance, two days before a contract is due to close, the price has typically deviated from the actual closing price by around 10% -- an outcome, says Allium, that indicates the contracts are not ready for prime-time as a serious hedging mechanism. It's early days, of course, and both the popularity and predictive power of Kalshi's compute-related contracts are likely to rise. But it is no sure thing as the history of new commodities markets attests. One prominent failure in this field is an attempt by a West Coast stock exchange to create a futures market for California almonds. Even though almonds are the state's most valuable crop, the push to launch almond futures never panned out as the industry struggled with standardization questions, and as the market failed to attract a critical number of the speculators who provide essential liquidity. Given the sheer size of the compute market, it's likely that financial firms will figure out how to develop a futures market before long. The question is whether that will be Kalshi or a large or more traditional player. Darrell Duffie, a finance professor at Stanford University's business school, says it will be the latter. He observes that successful futures markets like the one for oil cater to a massive customer base of both retail traders and institutions, and that exchanges like CME and NYSE-parent Intercontinental -- the two big incumbents that announced their intention to get into compute -- are best designed to do this. In addition to the retail futures market, there is another type of popular derivative known as OTC swaps that are used by institutions, and that entail creating one-off contracts between two parties. These type of swaps are likely to become common in the world of compute in coming years, but Duffie says it is deep-pocketed big banks that are likely to dominate this trade. "There is a narrow lane for Kalshi to offer forward curve contracts for compute in their prediction market but they are not as well positioned as the OTC swap market intermediated by bank dealers and the exchange traded futures market," said Duffie. "Kalshi does not have the infrastructure and participant capital necessary to safely handle high-volume risk transfer for compute." Kalshi, meanwhile, says it is still in stage one of its plans when it comes to compute futures, and that it is not daunted by bigger and more established competitors. On Wednesday, the company announced it has brought back Jeff Bandman, a former senior CFTC official, to run a division focused on futures markets. Kagan, the company's head of research, also noted that Kalshi is not seeking to compete across the entire emerging compute market. Rather, she says, the company's ability to quickly create new compute-related markets will give it a data advantage that will benefit Kalshi's business directly, but that will also feed into the financial industry more broadly.

CFTC Weighs AI Compute Futures as CME Eyes October Launch

The regulatory review could shape an emerging market that lets companies and investors trade and hedge the cost of increasingly scarce AI computing power. The US Commodity Futures Trading Commission (CFTC) is preparing to solicit public comment on futures contracts tied to computing capacity, a critical resource for artificial intelligence development, as major exchanges move to launch products tied to the emerging asset class. Bloomberg reported Monday that the regulator sent a request for comment to the White House Office of Management and Budget for review. The move could complicate the timeline for planned compute futures from CME Group and Intercontinental Exchange, whose products remain subject to regulatory approval. Once the White House review is complete, the CFTC is expected to open a public comment period, typically lasting 30 or 60 days, according to Bloomberg. The review signals that regulators are still weighing questions around a market that would allow participants to trade and hedge the cost of computing power. CME announced last week that it plans to launch two compute futures contracts on Oct. 5, pending regulatory approval, effectively turning AI computing capacity into a tradable commodity alongside oil and electricity. Market intelligence firm Silicon Data will provide the benchmarks used to price the contracts. The products are being launched as artificial intelligence reshapes the economy and investment landscape, driving a historic buildout of data centers and computing infrastructure. Recent estimates from TD Lombard, Goldman Sachs and Bridgewater Associates put AI infrastructure spending at roughly 2% to 2.5% of US GDP this year.

CME Group to launch GPU compute futures contracts in October By Investing.com

CHICAGO - CME Group and Silicon Data announced plans to launch two compute futures contracts on October 5, 2026, pending regulatory review, according to a press release statement. The contracts will track hourly rental costs for Nvidia H100 and B200 graphics processing units based on indexes published by Silicon Data. Each contract represents one month of GPU rental. Silicon Data is backed by trading firm DRW. The Silicon Data H100 Rental Index Futures will track the Nvidia H100 chip, while the Silicon Data B200 Rental Index Futures will track the Nvidia Blackwell B200. Both contracts will be listed on NYMEX. "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," said Pete Keavey, Global Head of Energy and Environmental Products at CME Group. Carmen Li, Chief Executive Officer of Silicon Data, stated that the futures will provide "a public, tradable reference price for the resource every AI system runs on." The contracts aim to provide hedging tools for companies managing GPU rental costs, which have experienced price volatility due to demand for AI infrastructure. The futures will allow businesses to lock in compute costs and provide visibility into future AI spending. CME Group operates derivatives exchanges across multiple asset classes including interest rates, equity indexes, foreign exchange, energy, and agricultural products. The company trades through its CME Globex platform and operates CME Clearing as a central counterparty clearing provider. This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.

Nvidia chips join oil in the futures markets

Training an AI model requires graphics processors, GPUs. Rather than buying them, most companies rent access by the hour from data center owners. That hourly rate varies widely and is negotiated over the counter, meaning client by client, with no one knowing what the neighbor paid. CME Group Inc and its partner Silicon Data will measure it daily and publish an index, pending approval from the US regulator. A futures contract is used to lock in today a price for a future date. The buyer hedges against a rise, the seller against a fall, and no chip changes hands because the contract is cash-settled, based on the difference between the fixed price and the observed price. The price, continuously quoted, becomes the reference the entire industry checks, including those that will never trade the contract. Two products to be launched in early October Two products will be offered, one on the H100 chip, the other on the B200 that succeeds it. Each covers 730 GPU-hours (365 days X 24h, divided by 12), or one month of renting a Nvidia chip, with maturities listed out to three years. "For years, two companies buying exactly the same capacity could pay radically different prices with no way to know who got the better deal," said Silicon Data Chief Executive Officer Carmen Li. The new index gives them a benchmark. The first in line are AI start-ups that buy compute, and cloud giants such as Amazon or Microsoft, which sell it. Investors, meanwhile, will be able to wager on the price of compute without buying a stock or hardware. A window of visibility in the fog Pierre-Yves Gauthier, chairman of AlphaValue, warns that this transparency "will probably highlight weaknesses in the hyperscalers' business model". As long as rates remained confidential, the true profitability of these activities eluded investors. The existence of a form of public price could revive an accounting debate, because cloud giants have recently tended to extend depreciation lives to soften the blow of colossal capex. It is also important not to forget the link to energy: an hour of computing is first of all an hour of electricity consumed. Data centers absorb about 7% globally and are boosting gas demand in the US, which could open arbitrage between the price of compute and the price of power, Pierre-Yves Gauthier notes. It will take a few weeks to determine whether the futures contracts offered by the CME will stir up the sector and how the market will actually embrace them. At this stage, their impact is purely conjectural.