CME Group launches first futures market for AI computing power with Silicon Data partnership

CME plans to launch futures market for AI computing power

CME Group is launching the first futures market for computing power, in a sign that the AI boom is turning the output of sophisticated chips into a major asset class. The new contracts for the future rental of graphics processing units, or GPUs, which can take months to order and can swing sharply in price, will allow investors and technology firms to bet on or hedge the future cost of computing power. CME, one of the world's biggest derivatives exchanges, is partnering with Silicon Data, a firm backed by Chicago-based trading giant DRW that provides pricing indices and other market data about AI computing power. The new contracts will be based on Silicon Data's indices. "Compute is the new oil of the 21st century," said CME's chief executive Terry Duffy, adding that it is "becoming a fast-emerging asset class in its own right". The AI boom has fuelled massive demand for computing power, which labs use to train large language models such as OpenAI's ChatGPT and Anthropic's Claude. Chipmaker Nvidia is among firms to have highlighted the need for investment in compute to continue scaling AI technology and its capabilities. Silicon Data currently provides daily rental rates for A100, H100 and B200 chips on financial data platforms such as LSEG and Bloomberg. Futures markets allow market participants to lock in the price of a commodity to be bought or sold at a mutually agreed date in the future. The first futures contract linked to a physical barrel of crude oil, the world's largest commodity market, was traded in 1983 in New York. The firms aims to launch the contracts later this year to meet the significant demand from tech companies, banks, investors and others, according to Carmen Li, chief executive of Silicon Data. "There are billions, if not trillions, of dollars of contracts being signed," she said in an interview. "The desire to manage volatility and risk is real." In an interview with the Financial Times at the end of 2024, DRW founder Don Wilson said that compute could become the biggest commodity in the world and that he was exploring ways to trade it. Shares in major chip providers have rallied hard this year, as tech hyperscalers have committed hundreds of billions of dollars to grow their AI-related infrastructure, including chips. US government-backed Intel is up over 200 per cent, while AMD has doubled. Nvidia is up 17 per cent.

CME Group and Silicon Data to launch AI compute futures market - SiliconANGLE

CME Group and Silicon Data to launch AI compute futures market Silicon Data, the startup that provides market intelligence for artificial intelligence compute infrastructure, will provide the price indexes for a new futures market that will allow investors to hedge their bets on the semiconductor industry. CME Group's new "compute futures market" will give investors the opportunity to bet on the price of renting cloud-based computing power for AI workloads, the companies explained. Investors will be able to take out contracts and lock in a price for compute capacity based on Silicon Data's benchmark for graphics processing units. It provides a way for them to hedge against the rising costs of renting graphics processing units, at a time when demand for the chips is going through the roof. "GPU markets have historically lacked standardized reference pricing," said Silicon Data Chief Executive Carmen Li. "The launch of compute futures is an important step towards giving AI builders, cloud providers and investors more reliable tools for valuation, hedging and long-term planning." Futures markets are usually associated with commodities such as oil and other petroleum products, precious metals and basic foodstuffs, but they've also become common for assembled components in fast-growing industries, where prices and demand change rapidly. Silicon Data is a natural partner for CME Group in this endeavor. Last month it launched a new service called the GPU Forward Curve, which offers the first standardized look at the anticipated costs of GPU capacity both now and in the future. Its products also include a standardized GPU price index and a dynamic random-access price index. When it launched the GPU Forward Curve last month, Silicon Data said its intention was to transform AI compute into a financialized commodity, and it has wasted little time in making that vision a reality. The launch of the new index is especially timely because few industry observers, or investors for that matter, see demand for GPU rentals slowing down any time soon. Morgan Stanley analyst Shawn Kim said in a note to clients this week that demand for GPUs and also central processing units is going to explode in future as more autonomous AI agents come online. "The AI system in the future will look like a distributed system consisting of GPU racks for dense model compute ... [and] agentic CPU racks for orchestration, processing data and tool execution," he wrote. The price of memory chips has soared this year as AI infrastructure providers continue to gobble up the world's supply of GPUs and now, increasingly, CPUs as well. Hyperscale data center operators like Amazon Web Services Inc., Google LLC, Microsoft Corp. and Meta Platforms Inc. have all announced plans to increase their capital expenditures this year, leading to shortages of memory chips that are driving prices higher. Memory chip makers are benefiting enormously from this demand, with their stocks being among the biggest gainers in the year to date.