On 21 August 2026 the Commodity Futures Trading Commission published a six page request for comment on the listing of compute derivatives contracts (RIN 3038-AF77, 91 FR 54259). It puts twenty three numbered questions to the public about a class of contract that, by the Commission's own description inside the same document, does not yet have a settled underlier. Comments close on 20 October 2026.
The document has nothing to do with prediction markets. Read end to end, the request does not contain the phrase event contract, the phrase prediction market or the word gaming anywhere, and it never mentions Rule 40.11. No yes or no contract acquires a new obligation because of it.
We are writing about it anyway, because it is the cleanest measurement available of something this category argues about constantly. When a standard arrives after a product rather than before it, is that fate or a decision? Set this request beside the prediction market documents the CFTC published earlier in 2026 and the answer sits in the publication dates.
Twenty three questions and no contract to point at
The Commission announced the request on 19 August 2026 in release 9286-26, where Chairman Michael S. Selig framed it as a first step rather than a reaction, saying that "America cannot win the AI race without a robust derivatives market for compute." The underlier at issue is access to computing power, priced by the hour or by volume of model inference, and the Commission ties the exercise to the White House AI Action Plan of July 2025, one of whose recommended policy actions is to improve the financial market for compute.
What makes the document unusual is how candid it is about the state of the thing it is asking about. The full text records the Commission's preliminary understanding that compute markets are fragmented, that price formation happens mostly in opaque bilateral transactions, and that consensus is still forming on what the underlying commodity even is. It then states, in the Commission's own words, that compute may not yet show "fungibility, standardization, and sufficient liquidity". That is a regulator writing down that the product is not ready, and asking the listing questions in the same breath.
The twenty three questions sit in four blocks: seven on the size and liquidity of the compute cash market, nine on oversight and susceptibility to manipulation, five on customer protection, and two on perpetual futures. A request for comment of this kind carries no legal effect at all, and this one was still treated as a significant regulatory action under section 3(f) of Executive Order 12866 and reviewed by the Office of Management and Budget.
Every one of them is a question a listing memo has to answer anyway
Strip out the word compute and the second block reads like the file a designated contract market assembles before it certifies anything. Question 2(a) asks what a contract settling to an index built from private bilateral prices would have to demonstrate to satisfy DCM Core Principle 3, the statutory duty at 7 U.S.C. section 7(d)(3) and 17 CFR 38.200 to list only contracts that are not readily susceptible to manipulation. Question 2(g) asks whether any compute price series exists that meets the criteria in Appendix C to part 38, the guidance the Commission adopted in 77 FR 36612 in June 2012, and what the Commission should do if the answer is no.
Question 2(b) is sharper still. Published compute price series are built partly from rates that the capacity providers themselves administer, so the request asks what would stop a provider from moving the settlement index by adjusting a posted rate, steering capacity toward or away from a venue whose trades feed the calculation, or choosing not to trade during the observation window. That is the settlement source problem stated by the regulator before a single contract references the index, and it is the same structural problem this site described when it looked at who actually produces the number an election contract settles against.
Question 2(d) goes to Core Principle 4 and 17 CFR 38.250, and asks whether a DCM should be required to hold an information sharing arrangement with every compute venue and every capacity provider whose data enters a settlement reference price. Readers who followed the sports contract argument will recognise it. The CFTC has already proposed to treat a formal information sharing agreement with a governing body as a factor weighing in an exchange's favour, which is the subject of our piece on the league agreement that counts. Here the same idea arrives as an open question before anybody has had to answer it in a filing.
Question 1(c) is the one to keep. It asks whether it would be appropriate to permit trading in a contract settling to a price the Commission may not be able to observe, verify or surveil. Every venue that lists anything eventually has to answer that about its own settlement source. This is the rare case where the question is on the public record before the answer has to be defended.
Five a year became sixteen hundred before anyone asked
Now the comparison. The Commission's first broad public question set on this category was the advance notice of proposed rulemaking titled Prediction Markets (RIN 3038-AF65), published on 16 March 2026 at 91 FR 12516, with forty questions and a comment window that closed on 30 April. We covered what came back from it in forty questions went out and two rules came back.
Footnote 9 of that notice is the number worth carrying around. From 2006 to 2020, designated contract markets listed an average of roughly five event contracts a year. In 2021 the figure rose to 131, stayed near that level until 2025, and in 2025 the exchanges certified approximately 1,600 event contracts for listing. The Commission was describing a more than three hundred fold increase over that baseline in the same paragraph in which it opened its first general enquiry into the category.
The scale is confirmed on the other side of the same year. The reporting proposal at 91 FR 40102, published on 1 July 2026 and analysed here in the regulator would see four hundred traders instead of a million, records that twelve DCMs offer or intend to offer these contracts, that seven new DCMs were designated from the start of 2025, that more than twenty applications were pending, and that during February 2026 one of the largest exchanges averaged roughly 91,000 event contracts a day with trading volume.
None of that is an accusation. It is a sequence. The questions about listing standards for event contracts were asked while the market was running at ninety one thousand contracts a day. The questions about listing standards for compute derivatives were asked while the Commission was still writing down that the commodity is not fungible yet.
Customer protection has its own heading in the compute document
The contrast is starkest in the third block. Section 3 of the request is titled Customer Protection of Market Participants and runs to five questions, including 3(d), which asks what unique protections and prophylactic measures are appropriate or necessary for the protection of retail users, compared with other derivatives markets. There is no listed compute contract and no retail user of one, and the question is already in the Federal Register.
The advance notice on prediction markets has no customer protection heading at all. The word retail appears three times in it, twice inside the margin question under Core Principle 11 and once in a question comparing institutional with retail trading. Nothing in it asks what a deposit limit, a cool off period or a self exclusion tool should look like on a venue where a contract can resolve in minutes.
The rule that would ask is still a sentence. In his remarks to the Innovation Advisory Committee on 20 August 2026, Chairman Selig described a roadmap of three items. Two of them, the public interest proposal at 91 FR 35806 and the reporting proposal, were on paper. The third, amendments to parts 38 and 40 carrying consumer protection requirements in product governance, market design and incentive programs, existed only as that paragraph. It still does. As of 25 September 2026 the Federal Register carries nothing under it, which is the state we described in the consumer protection rule exists only as a sentence, and nothing in the Commission's Federal Register output since has changed it.
So the sequence for retail protection, stated plainly and with dates: for a product class with no listed contract, the question was published on 21 August 2026. For a product class that certified sixteen hundred contracts in one year, it has not been published yet.
The market with no volume got the longest comment window
Comment periods are the cheapest possible measure of how much room a regulator thinks a question deserves, and all four are a matter of record.
- Advance notice on prediction markets, published 16 March 2026, comments closed 30 April 2026. Forty five days.
- Public interest determinations proposal, published 12 June 2026, comments closed 27 July 2026. Forty five days.
- Event contract data reporting proposal, published 1 July 2026, comments closed 31 July 2026. Thirty days.
- Compute derivatives request for comment, published 21 August 2026, comments close 20 October 2026. Sixty days.
The shortest window of the four went to the rule for the market already running at ninety one thousand contracts a day, and the longest to the market that has not opened. That is what catching up looks like from the inside. When a category grows faster than its rulebook, the rulebook stops being drafted and starts being patched, and patches come with thirty day clocks.
None of this is legal authority over an event contract
It is worth saying again, because the temptation to borrow authority from an unrelated document is real. The compute request creates no duty for any prediction market, interprets nothing in the Special Rule, and does not touch the definitions under consultation in the 40.11 proposal we examined in the gaming definition as a measuring stick. An operator who cites it as a source of obligation is wrong.
What the two categories genuinely share is the rulebook underneath. Both are listed by designated contract markets under part 38. Both reach the market through self certification under 17 CFR 40.2 or through the slower approval route at 17 CFR 40.3, a door that, as we found in no event contract has taken since 1995, nobody in this category has used. Both have to satisfy Core Principle 3 with the same Appendix C reasoning. The questions transfer even though the authority does not.
An exchange can choose the order too
The reason this belongs on a site about voluntary standards is that the choice of when to answer is not the regulator's alone. A venue preparing a new category can produce the same file before listing that the Commission is asking compute participants to produce before trading. Nothing prevents it, nothing requires it, and the difference between the two sequences is most of what our fifth directive is about: "Standards are set against the jurisdiction that regulates next, not only the one that has not yet."
Concretely, the compute request is close to a template. Before its next new category goes live, a venue can put on its own public record: what the settlement source is and who administers it; which parties are positioned to move that number and what stops them; what surveillance covers the underlying market and whether an information sharing arrangement exists with the venues that produce the price; what the deliverable supply or its cash market equivalent looks like; and which protections apply to a first time retail user of that specific product rather than to the account in general.
Those five items are not our invention and they are not a certification requirement. They are questions 1(c), 2(b), 2(d), 1(g) and 3(d), lifted from a document the Commission published for a different commodity. Answering them in advance costs a venue a memo. Answering them afterwards, in a comment file, on a thirty day clock, while the product is already trading, costs considerably more, and the category has a year of evidence for that price.
Where these documents are
The request for comment is at 91 FR 54259, published 21 August 2026, with the complete text available through govinfo. The advance notice on prediction markets is at 91 FR 12516 and the reporting proposal at 91 FR 40102. All comment dates above are taken from the Federal Register records for those documents, and the statement that no consumer protection proposal for event contracts had been published was checked against the Commission's Federal Register output through 25 September 2026.