The CFTC Opened Customer Funds to Tokenized Assets. Then It Asked a Harder Question.

On Thursday, the Commodity Futures Trading Commission appended four entries to its running crypto FAQ. The headline item is straightforward: futures commission merchants and derivatives clearing organizations may now invest customer funds in tokenized assets. The buried item is not. A single clause requires that firms keeping their books on public, permissionless blockchains maintain the ability to retrieve those records when the network — or its block explorer — is unavailable. No offchain duplicate is mandated. The capability is. The rest of the document is administrative. That clause is engineering. It is also the only sentence in the package that admits, in writing, that the infrastructure the industry sells as unstoppable has an availability profile, and the regulator has now assigned that profile to the regulated entity to insure. I have spent nine years reading regulatory text against deployment scripts. The pattern has not changed. The concession is the press release. The constraint is the substance. The ledger does not lie, but it forgets. The instrument in question is Regulation 1.25. It governs how FCMs and DCOs may invest the customer money they hold in segregation — funds that back futures positions and belong to someone else. It is not a list of good ideas. It is a closed menu of permitted instruments: Treasuries, money market funds, certain bank deposits, comparable low-risk collateral. Anything outside the menu is not a judgment call. It is a violation. The new FAQ expands that menu without widening it. Tokenized versions of already-permitted assets now qualify, provided four conditions hold. The underlying asset must already be permitted, so nothing new enters the list. The token must confer legal and economic rights identical to, or functionally equivalent to, the traditional version. Liquidity, concentration, and maturity limits must be satisfied. And the token must sit with an acceptable custodian. That is a composability test, not a novelty test. The regulatory question was never whether the instrument was a token. It was whether the token does the same job. The agency answered with substance over form. The chain of decisions leading here reveals the method. March 20: a baseline crypto FAQ. June 16: a request for information under Executive Order 14405. September 16: Chairman Michael Selig committing to make rules under existing authority. December 2025: a staff letter permitting Bitcoin, Ether, and payment stablecoins as margin collateral. This Thursday: tokenized assets plus onchain recordkeeping. Five steps. No legislation. The Clarity Act sits stalled in the Senate, and the agency has decided not to wait for it. The RFI responses cited on the record came from dYdX Labs, the Blockchain Association, and the Solana Policy Institute — a DeFi protocol, a lobbying coalition, and a chain-ecosystem policy shop. The feedback loop is real. The output is a staff FAQ, which is not a rule. Now the teardown. The term "functionally equivalent" carries the entire edifice and has no definition. A tokenized Treasury bill is straightforward: same coupon, same issuer, same maturity. A tokenized equity claim is a different animal, and the FAQ does not say where the line falls. In 2017 I reverse-engineered the vesting schedules of an ICO whose token purported to grant holders rights equivalent to a preferred share. They were not equivalent. The rights lived in a legal wrapper, not in the token, and the wrapper was amended twice before deployment. Functional equivalence is an audit finding, not a marketing claim, and the agency has left it to be adjudicated case by case. That yields flexibility for the regulator and ambiguity for the filer. Whoever bears the ambiguity bears the cost. The retrieval requirement is the honest part, and it is expensive. Public chains fail in specific, unglamorous ways. RPC providers go down. Block explorers delist. Archive nodes are not running where you assumed they were. Each of those is a records-availability incident, and Regulation 1.31 does not care why the record was unreachable. So a firm taking the pure-onchain path must build offchain retrieval capability anyway: an indexer, an archive, a mirror, and a documented procedure. The compliance stack ends up dual-layer. The onchain ledger becomes the source of truth. The offchain mirror becomes the source of proof. Two systems where the narrative promised one. I have watched this movie. In 2020 I scripted pool-balance monitoring for a yield farm whose APY was manufactured by emissions rather than fees. The reporting was elegant. The mechanism underneath was not. Institutional recordkeeping has the same shape, with the elegant half public and the load-bearing half hidden. The data volumes do not justify the architecture being sold. This is where industry rhetoric outruns its own arithmetic. A clearing organization's compliance record set is small: orders, confirmations, position reports, audit trails. Compression and batching reduce it further. That is nowhere near the regime dedicated data-availability layers were designed for. Most rollups today do not generate enough data to justify their own DA committees, and a futures commission merchant's retention obligations will not either. The only onchain data market that has ever produced sustained fee pressure is inscription activity on Bitcoin, and no regulator asked for it. The sophisticated response to this FAQ is not a new DA chain. It is a retention policy. There is a structural preference hiding inside the availability clause. Require retrieval redundancy specifically from users of public, permissionless chains and require nothing comparable from permissioned deployments, and you have not banned permissionless infrastructure. You have taxed it. The tax is paid in archive nodes, indexers, and legal review. Enterprises respond to taxes by substituting. Expect the first wave of compliant deployments to look suspiciously consortium-shaped. The stablecoin exclusion is the actual editorial. Payment stablecoins are explicitly excluded from eligible customer-fund investments. Read past the headline: the document grants tokenized versions of approved assets a path in and holds payment stablecoins out of the same door in the same paragraph. That is not an oversight. It is a distinction the agency chose to draw while stablecoin legislation settles. Anyone trading the regulatory-opening narrative on the assumption that it covers everything should price the carve-out. One detail is smaller and more telling. Tokenized government money market funds require a written confirmation letter from the custodian. Ask what that letter is for. It exists because the mapping between the token and the underlying fund is a trust claim, not a proof. In 2021 I traced a collection's deployer wallet to three previously banned addresses, and its origin story collapsed in a week. A custodian's confirmation letter is the same class of artifact: a claim about history, signed by a party with an interest in the outcome. Where proof is available, regulators accept proof. Where it is not, they substitute a signature. That tells you where the commission believes the weak joint sits. On recordkeeping itself, Regulations 1.31 and 45.2 were declared technology neutral. No discrimination against onchain form. This reads as minor and is not. It is a tier-one regulator treating a blockchain record as a record, full stop, with no special pleading in either direction. Where the bulls are right, and I will grant it plainly: neutrality is not nothing. The reflexive critique that regulators will never accept public chains is now falsified by the text. The commission did not demand a permissioned ledger, a private chain, or a vendor's enterprise product. It accepted the public chain and priced the risk into the operator's own resilience obligations. That is the correct allocation. The party that chooses the architecture owns its failure modes. Compare that to 2020, when DeFi protocols externalized every failure onto depositors and called it decentralization. The regulator did the opposite. The bulls are also right that the demand-side effect is real. Customer funds held at futures commission merchants and clearing organizations form a pool measured in the hundreds of billions. Any widening of the permitted menu routes structurally more capital toward tokenized issuance and qualified custody. Infrastructure that can satisfy the retrieval standard inherits that demand. The blind spot is durability. This is staff guidance: not a rule, not a statute, not binding. The document says so about itself. An FAQ can be superseded by an FAQ. A change of leadership, a change of administration, or one adverse case can erase it without a hearing. A compliance program built on the current text is built on a promise the agency has explicitly declined to make legally enforceable. Institutional capital does not allocate on staff opinion. It allocates on rulemaking. Watch for the conversion, not the announcement. If the commission converts this FAQ into notice-and-comment rulemaking, the permitted-asset menu changes permanently and custody competition resets around the retrieval standard. If it does not, this is a memo whose shelf life equals the chairman's tenure. The ledger does not lie, but it forgets — and an unreachable record and a nonexistent record look identical to an examiner. Which one will the industry have to prove, in five years, actually happened?