Paradigm and Andromeda File Comment Letter on Compute Derivatives
Today, Paradigm and Andromeda, a global compute marketplace, filed a comment letter with the CFTC responding to its request for comment on the listing of derivative contracts that track the price of renting computing power.
As we make clear in our letter, compute is already one of the most economically significant commodities in the United States, accounting for as much as four percent of GDP. Yet almost no one—from the labs that buy capacity, to the data centers that sell it, to the lenders that finance it—can hedge the associated price risk. And until they can, that risk is passed to consumers, increasing cost and decreasing adoption for the entire AI buildout.
The CFTC, building on the foundation laid by President Trump’s AI Action Plan, recognizes the potential of these contracts. But the agency also raises a host of questions about the status of compute markets, including whether they are standardized and fungible enough to support a derivatives market, whether they can be priced accurately, and whether they fit within the CFTC's general regulatory framework. Fortunately, we believe the answers to all the major questions is a clear yes, so long as a few reasonable steps are taken by the Commission.
Standardization is a common problem, but addressable via benchmarks. The CFTC is right to note that compute varies by chip type, location, configuration, and a host of other factors. But that is not unique to this nascent market. Crude oil, for instance, varies by grade, sulfur content, and location, so the market is settled on a handful of benchmark contracts. Compute can work the same way—with published methodologies, transaction-based inputs, disclosed contributor concentration, and independent administration—operating within the CFTC’s existing framework.
Physical delivery can anchor the price. CFTC is also concerned about challenges related to physically delivering compute, which is what helps anchor the cash-settled price to the real-world market price. But here again, this is not unique and crude is an apt analogy. Fewer than one percent of crude futures end in delivery, but the possibility of actual delivery is what tethers the entire market to real-world demand. After all, traders don’t know which 1% of those contracts will be delivered, and so must act as if they all could be. Ironically, compute is actually better suited to physical delivery than oil or other extant commodities. After all, unlike the shipping and storing required for oil, a standardized cluster can be provisioned online and in minutes. It would be safe to say that compute is a better fit for derivatives markets than many existing derivatives markets; this fact should quell any fears that compute is not yet ready to support a derivatives market of its own.
Compute fits within the existing CFTC framework. Lastly, the CFTC asks whether compute markets require a bespoke regulatory regime. This question is reasonable but misplaced, especially given recent precedent by this Commission. Just last year, for example, CFTC withdrew product-specific guidance for carbon credit derivatives, concluding that new contracts “should be evaluated in the same manner as any other derivatives contract.” The same logic applies with equal force here.
Underlying our comments is the axiom that every market starts somewhere. The deep and established markets of today were once the thin and innovative markets of yesterday. The private sector has already germinated the beginnings of what we believe has the potential to be as critical and transformative for the economy this century as agricultural commodity markets were for the 19th century and energy markets were for the 20th. What is needed now are targeted, reasonable regulatory actions by the CFTC to help these markets grow.
Compute is poised to be the next great chapter in a long history of American capital markets, and we thank the Commission for taking up these questions. You can read our full comment letter here.