The Architecture of CLARITY

May 26, 2026

The Senate substitute to H.R. 3633 — the engrossed amendment in the nature of a substitute introduced by Sen. Scott on May 12, 2026, that the Senate is now calling the “Digital Asset Market Clarity Act” — is 309 pages organized across nine titles. The first time you read it through, the impulse is to map it section by section: Title I for securities, Title II for BSA, Title III for DeFi, Title IV for banking, and so on. That reading is not wrong, but it misses the doctrinal architecture entirely. The bill’s most consequential moves are structural, not substantive. CLARITY does not abolish Howey. It does not promulgate a new test for what counts as an investment contract. It does not draw a bright-line rule between security and commodity that a court could apply on first principles. What it does — and what every practitioner working on a token launch, a DeFi protocol, a custody arrangement, or a bank’s digital-asset desk needs to understand before reading anything else — is route around the existing framework by overlaying a new statutory category on top of it and rerouting most of the consequences.

This post is the doctrinal scaffold for the series that follows. The thesis is straightforward and worth stating up front: CLARITY’s central innovation is the construction of a three-tier statutory regime — network token, ancillary asset, digital commodity — sitting atop the existing securities laws, paired with a fourth construct, the decentralized governance system, that operates as a separate legal person and refuses to be deemed an unincorporated association of its participants. The §4B disclosure regime grafted onto the Securities Act of 1933 is what makes that three-tier scheme operational. Every other title in the bill — Title III’s DeFi line, Title IV’s bank-permissibility expansion, Title VII’s bankruptcy reform, Title VI’s developer protections — is downstream of this architecture. Reading CLARITY as a series of standalone subject-matter reforms will leave you confused about why pieces fit together the way they do. Reading it as a single doctrinal engine that classifies first and then routes consequences accordingly will not.

The Trichotomy

Start with the categories. CLARITY’s §2 defines “digital commodity” by reference to the new §1a of the Commodity Exchange Act added elsewhere in the bill; the Commission’s universe is everything that is digital, recorded on a distributed ledger system, and not a security. So far this is unremarkable — the categorical move is borrowed wholesale from the 2024 FIT21 framework. The novelty is in the further subdivision. A “network token,” defined in §4B(a)(7) of the Securities Act as added by §102, is a digital commodity “intrinsically linked to a distributed ledger system and that derives, or is reasonably expected to derive, its value from the use of such distributed ledger system.” A network token is, by statutory operation, “treated as a non-security solely for purposes of the Federal securities laws,” subject to a narrow set of disqualifying conditions discussed below. The *ancillary asset*, also defined in §4B(a)(1), is the network token that has not yet reached maturity: a network token “the value of which is dependent upon the entrepreneurial or managerial efforts of an ancillary asset originator or a related person, as those concepts are further specified by the Commission by regulation.” The bill borrows the Howey vocabulary (entrepreneurial or managerial efforts) and uses it to mark the ancillary asset’s developmental phase rather than to determine whether the asset is a security at all.

In other words: every network token is potentially an ancillary asset in its infancy, and every ancillary asset is expected to mature into a fully decentralized network token whose value is no longer dependent on identifiable promoter effort. The bill builds in a rebuttable presumption to that effect in §4B(b)(5): a network token is presumed to be an ancillary asset until the originator or a digital asset intermediary submits a certification, supported by reasonable evidence, “sufficient to demonstrate that the network token is not an ancillary asset.” The Commission has sixty days to object; absent an objection, the certification takes effect automatically. The presumption runs *toward* regulation, but the mechanics make graduation cheap and quick.

So, the CLARITY framework is doing exactly what the SEC’s 2019 Framework for “Investment Contract” Analysis of Digital Assets purported to do, and what former Director Hinman gestured at in his June 2018 speech at the Yahoo Finance All Markets Summit — except that the bill replaces a multi-factor evaluative framework with a statutory category whose entry and exit conditions are spelled out and whose default direction is non-security. Howey survives by name; its operational consequences for token markets do not. The maturity question — Hinman’s “sufficiently decentralized” intuition — becomes a structured certification with a sixty-day clock rather than a years-long enforcement-driven discovery of the answer.

§4B as Bolt-On

The mechanism for routing consequences is Securities Act §4B, inserted by §102 of CLARITY immediately after §4A (the JOBS Act’s crowdfunding section). §4B(b) is where the doctrinal magic happens. Subsection (b)(1) provides that “the offer, sale, or distribution of an ancillary asset by, or caused by, an ancillary asset originator, including through an underwriter, shall be considered to be an offer, sale, or distribution of an investment contract involving an ancillary asset.” Read that carefully. The transaction is an investment contract. The transaction therefore implicates the Securities Act. But subsection (b)(2) immediately provides that the network token itself is “treated as a non-security” for purposes of §2(a)(1) of the 1933 Act, §3(a) of the 1934 Act, §2(a) of the Investment Company Act, §202(a) of the Advisers Act, and §16 of SIPA. The bill carefully distinguishes between the transaction — which can be an investment contract — and the asset — which, even when it is the subject of an investment contract, retains its non-security status everywhere else in the federal securities laws.

The doctrinal upshot is that Section 5’s registration requirement applies to the transaction, but the analytic surface area shrinks dramatically because Section 5 compliance for ancillary asset transactions is achieved through the §4B(d) disclosure menu rather than through the standard S-1/S-3/Reg A/Reg D pathway. §103 lays out the exemption-and-rulemaking framework that makes this concrete: an ancillary asset offering meeting the disclosure conditions of §4B(d) is exempt from registration, the Commission is directed to adopt implementing rules within 360 days, and the rulemaking must “tailor” the disclosure burdens to the technical and economic realities of distributed ledger systems. Practitioners who lived through the SEC’s no-action experience with the Munchee (DAO Order Sec. Act Release No. 81207 (July 25, 2017)), Telegram (SEC v. Telegram Grp. Inc., 448 F. Supp. 3d 352 (S.D.N.Y. 2020)), and LBRY (SEC v. LBRY, Inc., 639 F. Supp. 3d 211 (D.N.H. 2022)) line of cases will recognize what is happening: the registration question that those cases agonized over is rerouted into a disclosure question, and the disclosure question is given a finite scope.

Two further provisions complete the bolt-on. Subsection (b)(3) addresses the secondary market: the offer, sale, or distribution of a network token by *any person other than the originator or underwriter* is not the offer of a security under any of the federal acts or any “functionally equivalent” state law. This is the provision that, more than any other, kills the Reves v. Ernst & Young, 494 U.S. 56 (1990), family-resemblance overhang for secondary trading. Subsection (b)(4) addresses gratuitous distributions: airdrops, staking rewards, validator rewards, liquid-staking issuance, and protocol-rules-based programmatic distributions are presumed not to be offers, sales, or distributions of a security. The taxonomy of “gratuitous distribution” in §4B(a)(5) is broad enough to capture nearly every non-purchase token-acquisition mechanism — a taxonomy worth a standalone post (and getting one).

Anti-fraud authority is preserved expressly. §4B(b)(4)(B) clarifies that nothing in the gratuitous-distribution presumption “may be construed to limit, impair, or otherwise affect the anti-fraud or anti-manipulation authorities of the Commission, the Commodity Futures Trading Commission, or a State regulator.” The same savings language reappears throughout the bill. Practitioners should not mistake CLARITY for a deregulation bill. It is a re-routing bill. The Commission’s §10(b)/Rule 10b-5 jurisdiction over token markets is unaffected; the §17(a) anti-fraud authority over offers and sales is unaffected; the New York Martin Act survives where consistent with the federal scheme. What changes is the registration overhang, not the anti-fraud floor.

The Disqualifying-Financial-Rights Test

The next layer of the architecture sits inside §4B(a)(7)(B). To qualify as a network token — and therefore as a presumptive non-security at the asset level — a digital commodity must not bear any of four “disqualifying financial rights.” First, a security as that term is otherwise defined, or anything functionally equivalent to one. Second, an interest representing or substantially equivalent to a debt or equity interest, an option on such an interest, liquidation rights, or “an entitlement to, or a reasonable expectation of, an interest, dividend, or other payment, or direct or indirect transfer of value, from a person other than a decentralized governance system.” Third, an interest in a §3(c)-excluded investment company. Fourth, an interest in a non-investment-company entity holding assets other than securities.

The third and fourth disqualifiers are largely housekeeping — they prevent the use of network-token wrappers to evade the Investment Company Act and its §3(c)(1) and §3(c)(7) carve-outs. The first is tautological. The doctrinal work is being done by the second, and specifically by the parenthetical “other than a decentralized governance system.” Read carefully: a token can confer “an entitlement to, or a reasonable expectation of, an interest, dividend, or other payment, or direct or indirect transfer of value” from a decentralized governance system without losing its network-token status. This means protocol-revenue-sharing tokens, fee-switch tokens, vote-escrow tokens, and similar mechanisms survive the test so long as the value flow runs *from* the DGS rather than from a centralized issuer or a related person. The doctrinal lever is the legal personhood of the DGS, which we get to in a moment.

A separate provision, §105(a), instructs the Commission to adopt rules within one year confirming that a network token “shall not be considered as providing a disqualifying financial right” if its market value is “primarily derived, or is reasonably expected to be primarily derived, from a distributed ledger system or from the broader adoption and use of such a system” — including where the ledger collects, receives, accrues, or distributes consideration from network functioning; where the token provides governance capabilities; or where its value appreciates or depreciates due to network use or ancillary-asset-originator efforts. This rulemaking instruction collapses the most-litigated Howey prong — “efforts of others” — by deeming network-derived value to be non-disqualifying as a matter of statutory direction. The Commission has discretion in the rule’s particulars but not in the basic principle.

§105(b)(2) adds what is probably the most consequential and least-discussed grandfather in the bill: a digital asset that was, on January 1, 2026, the principal asset of an exchange-traded product not registered under the Investment Company Act and listed on a national securities exchange is, by statutory operation, neither an ancillary asset nor a security. BTC and ETH receive this status automatically; any other token whose spot ETP was listed before the cutoff joins them. The asymmetry created is significant — a token *with* a spot ETP gets a regulatory free pass; a token *without* one has to clear the §4B/§105 maturity tests — and it creates an obvious incentive structure for issuers to push ETP listings onto national exchanges before the cutoff freezes the universe. That dynamic deserves its own post (and gets one in the queue).

The Decentralized Governance System as a Separate Person

The fourth structural element — and the one that does the most quiet doctrinal work — is the DGS construct. §2(5) defines a “decentralized governance system” as, with respect to a distributed ledger system, any “transparent, rules-based system permitting persons to form consensus or reach agreement in the development, provision, publication, maintenance, or administration of the distributed ledger system, in which participation is not limited to, or under the control of, any person or group of persons under common control.” Subparagraph (B) is the key: “the decentralized governance system and any persons participating in the decentralized governance system shall be treated as separate persons unless those persons are under common control or acting pursuant to an agreement to act in concert.” Subparagraph (C) accommodates wrapper entities — a state-law decentralized unincorporated nonprofit association (Wyoming’s DUNA), a Marshall Islands DAO LLC, or analogous foreign-jurisdiction vehicles — so long as the entity does not operate “pursuant to centralized management” and provided that “the delegation of ministerial or administrative authority at the direction of the participants in a decentralized governance system shall not be construed to be centralized management.” Subparagraph (D) closes with a rule of construction: a DGS “shall not be deemed to be a person or a group of persons acting under common control” for purposes of the Act.

The construct is doing four things at once. First, it solves the entity-classification problem that the *Sarcuni v. bZx DAO*, 664 F. Supp. 3d 1100 (S.D. Cal. 2023), and *CFTC v. Ooki DAO*, No. 2:22-cv-05416 (N.D. Cal. 2023), courts grappled with — the question of whether a token-holder-governed protocol is, by virtue of its operation, a general partnership of its token holders with joint and several liability flowing from any actionable conduct. CLARITY says no, by statutory direction, so long as the participants are not under common control or acting in concert. Second, it preserves token-holder governance — voting, treasury management, parameter setting — without converting that participation into the kind of “common enterprise” that *Howey* and its progeny treat as evidence of a securities relationship. Third, it makes the “value flow from a DGS” carve-out in §4B(a)(7)(B)(ii)(III) coherent — value can flow from the DGS to token holders without being treated as a distribution from a person, because the DGS and its participants are separate persons by statute. Fourth, it provides the doctrinal foundation for the §301 DeFi carve-out, which uses the same construct to draw the line between a “decentralized” and a “non-decentralized” trading protocol.

The construct is not unlimited. Subparagraph (B) preserves the common-control and acting-in-concert exceptions; subparagraph (C) excludes wrappers operated pursuant to centralized management. Practitioners structuring DAOs in CLARITY’s wake should expect a body of Commission rulemaking and eventual enforcement on the boundary between ministerial delegation (preserved) and centralized management (disqualifying). The drafting is sound enough that the boundary should be administrable, but it is not self-executing.

Coordinated Control and the 49% Test

§104 puts operational teeth on the DGS construct by defining “coordinated control” as the disqualifying condition for ancillary-asset status. The Commission is directed to adopt rules implementing five indicia: (A) the protocol and any distributed-ledger applications are open-source and publicly available; (B) no person or common-control group has unilateral authority to censor, restrict, or grant preferential treatment via system operation or hard-coded source-code privileges; (C) no person or common-control group has beneficial ownership of more than 49% of outstanding token units or 49% of governance voting power; (D) the system has reached an autonomous state — no person has unilateral authority to alter functionality or consensus rules; (E) economic independence — the value-accrual mechanisms intended to flow consideration to the ancillary asset are functional.

The 49% threshold is the only hard number in the entire decentralization framework, and it is doing a great deal of work. Token-supply concentration and governance-vote concentration are both captured (the test is disjunctive — either trips it), and “beneficial ownership” is the operative metric, which means scoped grants, vesting schedules, lockups, custody arrangements, and trust structures will all need to be analyzed for ownership-attribution purposes. The Commission has discretion in the precise rules of construction (the bill says the Commission “shall consider” the criteria) but not on the 49% number, which is hard-coded.

§104(b)(3) provides safe harbors. Subparagraph (B) reaffirms that a DGS is not a person or common-control group for §104 purposes. Subparagraph (C) is the cybersecurity emergency-measure safe harbor: a pre-defined, temporary, rules-based emergency measure exercised by an incident response or security council exclusively in response to a documented cybersecurity incident or imminent threat, pursuant to publicly disclosed on-chain authorization mechanisms, that is strictly limited in scope and duration and exercised without unilateral control by any single person, does not by itself constitute common control or acting in concert. This is a thoughtful drafting concession to the realities of protocol governance — most major DeFi protocols maintain emergency-pause multisigs, and CLARITY explicitly preserves them as compatible with decentralization status.

The §301 Line and the Rest of the Architecture

The DGS construct and the coordinated-control test repeat themselves in §301, which applies the same logic to DeFi trading protocols. A “decentralized finance trading protocol” is defined as a distributed ledger system through which participants execute transactions via predetermined non-discretionary automated rules without third-party custody. A “non-decentralized” version is one in which (i) a person or common-control group has authority to control or materially alter functionality, (ii) execution is not based solely on source-code-encoded rules, or (iii) someone has censorship authority. §301(a)(2)(C) carves out from the “non-decentralized” classification a long list of activities that should not be deemed control: compiling, relaying, searching, sequencing, validating, oracle services, computational work, bandwidth provision, and participation in security councils. The activity carve-outs are practitioner gold and merit a standalone post, but the architectural point for this post is that §301 reuses the same DGS-separate-person logic and the same control-versus-non-control distinction that §104 establishes. The trading-protocol layer is just the trading-protocol analog of the asset-classification layer.

The same architectural through-line runs through Title IV (banking) and Title VII (bankruptcy). §401 authorizes national banks, FHCs, federal credit unions, and (by parity) state banks to perform the full menu of digital-asset activities — custody, staking, lending, principal market-making, dealing, payment, brokerage, derivatives — and bootstraps those activities into “the business of banking” under 12 U.S.C. § 24 Seventh. The list is fourteen items long and explicitly excludes nonfungible assets. §401(h) eliminates prior-notice and approval requirements outside the existing organic banking statutes. This is the statutory equivalent of, and significantly broader than, the OCC’s 2020–21 Interpretive Letter 1170/1172/1174 sequence (modified by IL 1179 to reinstate prior approval). For a chartered institution, the universe of permissible digital-asset activity goes from contested-and-conditional to expressly enumerated.

§701 then pulls ancillary assets and digital commodities into the Subchapter III stockbroker liquidation regime of Chapter 7 of the Bankruptcy Code. Ancillary assets are added to 11 U.S.C. § 741’s customer-property definitions, and digital commodity transactions are designated commodity contracts for purposes of the safe-harbor provisions (§§ 362, 546(e), 553, 556, 561, 562). The doctrinal point — and the through-line back to §4B — is that CLARITY treats ancillary assets as quasi-securities for customer-property and bankruptcy-safe-harbor purposes while treating them as non-securities for offering, trading, and reporting purposes. That asymmetry is not an oversight. It is the structural choice the bill is making: protect customers as if they hold investment property, but classify the property itself out of the registration regime.

What this Means in Practice

The practical consequences of the architecture, before we get into the standalone deep dives that follow in this series, are five.

First, the classification question becomes a routing decision rather than a legal conclusion. The question every issuer’s counsel will ask is no longer “is this an investment contract” but “is this a network token, an ancillary asset, or something else.” The first two categories are presumed non-securities at the asset level even when the originating transaction is a §4B(b)(1) investment contract; the third — anything that fails the network-token test, including any token with disqualifying financial rights — remains a security subject to the full *Howey* analysis and the standard registration/exemption framework.

Second, the registration question shrinks. Ancillary-asset offerings clear §4B(d) disclosure rather than full §5 registration. Secondary-market trading of network tokens is non-securities trading. Gratuitous distributions are presumed non-securities events. The compliance surface for token issuance is meaningfully smaller and more predictable. The compliance surface for fraud and manipulation is unchanged.

Third, DAO and protocol structuring becomes a structured exercise. The DGS construct, the 49% test, and the §301 carve-outs together provide a checklist for assessing whether a given arrangement clears the decentralization bar. Wyoming DUNA wrappers, Marshall Islands DAO LLCs, and similar vehicles are accommodated. Multisig emergency councils are accommodated. What is not accommodated is centralized management masquerading as decentralized governance.

Fourth, custody and intermediary regulation becomes activity-based. §301(c) requires the SEC to determine intermediary-registration consequences “only with respect to securities-related activities, based on the functions performed by the controlling person or group of persons, … without regard to technological form, distributed architecture, or purportedly decentralized characterization.” The bill is explicit that decentralization rhetoric is not a defense to control-in-fact; conversely, the absence of control is a defense regardless of decentralization rhetoric.

Fifth, every consequence flows from the classification. Banking permissions, bankruptcy treatment, BSA obligations, broker-dealer registration, anti-fraud reach — all turn on whether the asset is a digital commodity, ancillary asset, network token, security, or payment stablecoin. The five-way taxonomy is the bill’s organizing principle. Read every other title with that taxonomy in front of you and the structure resolves.

What this Series Will Cover

The posts that follow this one will work through the architecture in the order it was built. Next up: the §4B disclosure regime in detail — how the bill keeps *Howey* nominally while draining it of consequence — and the §4B(a)(7) disqualifying-financial-rights test that determines who is in the network-token category in the first place. After that: the §104 coordinated-control test and the 49% threshold; the DGS construct and the *Sarcuni*/*Ooki* overrule; the §301 DeFi line; the §401 bank-permissibility expansion; the §701 bankruptcy reform; the §404 stablecoin yield prohibition and its activity-rewards loophole; the §602 NFT safe harbor; the §505 tokenization-parity rule. Toward the end of the series, the contrarian posts: the §105(b)(2) ETF grandfather and the regulatory-arbitrage incentive structure it creates; what CLARITY pointedly does not address (tax, ERISA, residual CFTC jurisdiction); and the §904 Build Now Act housing rider that found its way into a digital-asset market-structure bill.

Written by David Lopez Kurtz