What is the difference between an egg and a computer chip? There are several answers, but an important financial one is that recent fundamental market changes have been much more disruptive for egg buyers than computer-processing users.
Avian flu has caused egg prices to oscillate from less than $1.50 a dozen to more than $8 in the Midwest and has at times resulted in the disappearance of eggs from supermarket coolers and the bankruptcy of large producers. Now consider what happened with the price of computing services when China’s DeepSeek unveiled an artificial-intelligence model, developed on the cheap, that caused the largest single-day decline in the market capitalization of any stock — more than twice the previous record: Prices barely moved and there were no bankruptcies.
It was a sign that the sophisticated financial infrastructure insulating computing services — but not eggs — from price shocks is growing and innovating rapidly and could represent a significant emerging opportunity. It was also a reminder of how quickly the narrative around AI demand for processing power can shift, reviving interest in futures markets for computing services, a dream people have pursued since 1989.
One of the purposes of futures markets is to draw financial investors into the production process of commodities. Institutional investors can provide cheap, patient capital by rolling futures contracts — going long the nearest delivery month, short future months, and moving the position forward every month. This allows producers, processors and distributors to operate with less capital, meaning lower costs and less disruption to real economic activity from financial shocks. There aren’t any egg futures, and only 5% of eggs trade on a specialized spot market, the Egg Clearinghouse Inc. This is an important reason for the wild price swings and market dislocations.
The market for computer chips and processing power is considerably more sophisticated, and futures markets are developing rapidly. The demand for processing power from independent AI companies alone is more than 10 times the size of the egg market and projected to grow at 50% per year. There is also growing demand from computational biology, weather modeling and other fields. In five years, it’s plausible that people will spend more money on computing power than on oil — and, of course, oil futures are a gigantic part of financial markets.
Computing futures are not a new idea. In 1989, the Pacific Stock Exchange proposed a dynamic random access memory (DRAM) chip futures market. A combination of regulatory skepticism and market-maker indifference killed the project. The Chicago Board of Trade, Chicago Board Options Exchange and Minneapolis Twin Cities Board of Trade each had their own versions of the idea; all went nowhere.
In 2001, Enron Corp. announced a plan to create a DRAM futures market as the first step toward making markets for a whole range of computer components. Kenneth Wang, who headed the effort for Enron, said: “The [DRAM] chip market is a boom or bust cycle, with prices very high or very low. It’s the perfect market. And Enron has the credibility. ... We’d like to hedge maybe 60% of a computer, if possible, in the long term.” By the end of that year, Enron had filed for bankruptcy and no longer had the necessary credibility.
Groups including dealer Buckaroo.com, the Singapore Exchange Ltd. and futures trader and entrepreneur Gilbert Leistner have also tried and failed. Why did they fail, and why is this time different?
One simple reason is scale. The global market for semiconductors in 1989 was a quarter of the 2024 demand just from independent AI companies. It was not large enough to interest large market makers and institutional investors. Volatility plays a role as well. The potential for sudden large market swings from events such as the DeepSeek announcement makes computer-processing futures more useful for hedgers and more attractive to traders.
Also, the market for chips used to be dominated by a few large producers and users, which could sign long-term bilateral agreements. While chips are still made by a few large companies, computing services that use those chips are sold to users by a wide variety of large and small data centers with high costs of capital, high leverage and little marketwide information. Purchasers include many relatively small startups, also lacking marketwide knowledge.
The final piece of the puzzle is standardization. Like oil, with its range of types, grades and sources, computing power comes from a complex mix of chips in different configurations and locations. Creating oil futures that were diverse enough to be useful for market participants but small enough to concentrate liquidity on active contracts required a huge investment in physical and legal infrastructure. Computing doesn’t require the physical investment that went into oil storage, pipelines and other infrastructure. And modern technology allows exchanges to trade limited sets of standardized contracts that concentrate liquidity while still allowing tailored transactions.
An event to watch this week is the first open auction held by Compute Exchange, one of the pioneers in this area. If it attracts large interest and runs smoothly, it could support the idea that a major futures market could develop around computer processing.
This is of obvious interest for market makers, futures traders, hedge funds and institutional investors. For individual investors, there’s the potential for a new class of real assets available in exchange-traded funds and other products. Computer processing has the potential to have a lower correlation with the general economy than traditional real assets such as property and physical commodities, and perhaps more security and a better long-term expected return as economic value moves from physical things to intellectual assets.
Of course, a healthy dose of skepticism is appropriate for an idea with a 35-year history of unbroken failure. Or, just maybe, computing exchanges are due for a win.
A message from Advisor Perspectives and VettaFi: To learn more about this and other topics, check out some of our webcasts.
Bloomberg News provided this article. For more articles like this please visit
bloomberg.com.
More Artificial Intelligence Topics >
