The FAQ dropped without a press release. Two pages, a definitional table, and one sentence buried below the fold: whether a staking token is treated as a security depends substantially on who controls the redemption path. I pulled the document at 06:14 Sรฃo Paulo time and ran the criteria against the two largest liquid staking derivatives by locked value. Coinbase's cbETH. Lido's stETH. The FAQ files them under the same interpretation. Structurally they are not the same asset. Not adjacent. Not comparable. The market read the headline, bid the sector, and moved on. The headline is not the specification. The specification is the redemption path, and the redemption path is where capital dies. I have audited redemption logic before. The ledger does not lie, only the logic fails.
Context: Two Architectures, One Label
The SEC's staking FAQ is a classification guidance document, not enforcement. It sets out the criteria by which a staking token โ a receipt representing staked ETH โ may be deemed an investment contract. The determinative factors cluster around custody, control, and the mechanics of exit. That framing matters more than the label it produces, because the label hides the architecture.
cbETH is a custodial staking token. Coinbase holds the underlying ETH, runs the validators, and issues cbETH as a wrapper. The FAQ states that under its described model, the issuer may not additionally pledge or rehypothecate the underlying asset beyond the staking operation itself. Ownership does not transfer to Coinbase per the protocol's own claim. The user holds a receipt. Unwrapping cbETH back to ETH requires a qualified account, and after unwrap, the holder still faces the network-level unstaking request. Two gates. Two queues. One of them is Coinbase's KYC perimeter.
stETH, the Lido receipt, is a non-custodial protocol token. A user deposits ETH into a smart contract and receives stETH at a 1:1 initial ratio, with rebasing applied to reflect validator rewards. The ETH is locked to validators selected through Lido's operator set. Exit is through a withdrawal queue โ a first-in-first-out structure that processes unstaking requests against available validator exits and protocol liquidity.
Same surface function. Radically different failure modes. This is the distinction the FAQ gestures at and the market ignores. Code is law, but implementation is reality.
Here is what the table in my analysis produced, and why the two columns do not reduce to one regulatory category.
| Dimension | cbETH (Coinbase) | stETH (Lido) | |-----------|------------------|--------------| | Architecture | Centralized custodial | Non-custodial smart contract | | Ownership of underlying | Retained by protocol per claim | Held against protocol contract | | Exit path | Unwrap (qualified account) โ unstaking request | Withdrawal queue | | Issuer asset restrictions | No rehypothecation described | Locked to validators | | Novelty | Wrapper + custody | Early LSD paradigm |
The FAQ treats the asset restriction as a shared safeguard. It is not. A contractual restriction on Coinbase is a promise enforceable by law. A restriction in Lido is an invariant enforced by bytecode. One can be amended by a board. The other can be amended by a governance vote that most stETH holders will never read. Neither is the same guarantee, and the FAQ does not distinguish them.
Core: The Redemption Path Is the Asset
I want to walk through the exit mechanics line by line, because this is where the two products diverge and where the SEC's classification stops being useful.
Start with cbETH. To convert cbETH back to liquid ETH, the holder calls unwrap. This requires a qualified account โ Coinbase's term for an account that has cleared its compliance perimeter. Once unwrapped, the holder returns to staked ETH exposure, not free ETH. Then they must submit a separate unstaking request to exit the validator set entirely. The network-level withdrawal queue on Ethereum is throttled by the number of validator exits processed per epoch. If exits are congested, the holder waits. Days. Possibly weeks. The cbETH holder absorbs two sequential delays and one compliance gate.
The compliance gate is the part the market underweights. A qualified account is a moving target. It depends on jurisdiction, on Coinbase's risk posture at the moment of request, and on the regulatory environment in the jurisdiction where the account is domiciled. A holder in a jurisdiction that Coinbase exits โ or a holder whose account status changes โ does not have a redemption path. They have a claim on a token whose only liquid expression is the secondary market.
Now stETH. Exit runs through the withdrawal queue. The queue is a smart contract structure that matches unstaking requests against ETH becoming available from validator exits and from new deposits. In steady state, it is fast. Under stress, it is a line. The length of the line is a function of net exit demand, and net exit demand is a function of confidence. Confidence is the first thing to fail in a drawdown. This is exactly what happened in mid-2022, when stETH traded at a discount to ETH on secondary markets because the queue could not clear fast enough and holders wanted out before the line got longer. I built a mainnet fork during that period to model the failure. My conclusion then holds now: the discount was not irrational. It was the market pricing the time value of the queue.
The queue is not a feature. It is a liability with a maturity.
The FAQ's classification logic appears to assume that a staking token's liquidity is a property of the token. It is not. It is a property of the exit path, and the exit path is a function of protocol mechanics plus network-level throttling plus, for cbETH, a compliance perimeter that can close without notice.
Let me quantify the asymmetry. For cbETH, the effective exit latency equals compliance review time plus unstaking queue time. Compliance review is opaque and variable โ I have seen institutional onboarding take weeks. For stETH, effective exit latency equals queue depth divided by clearance rate, plus secondary market slippage if the holder chooses to exit via a DEX. The stETH path has more failure points but fewer hard gates. The cbETH path has fewer structural failure points but one gate that can become absolute.
Neither is safe in the way the market assumes. Both are safe in the way the FAQ implies.
Here is the deeper problem. The FAQ treats the two asset restriction clauses โ Coinbase's no-rehypothecation promise and Lido's validator lock โ as equivalent. My 2025 audit of a DeFi lending protocol under new Brazilian financial regulation taught me that this equivalence is a legal fiction. A contractual restriction is enforced after the fact, through courts, through discovery, through enforcement actions that take years. A bytecode restriction is enforced at execution, on every block, with no appeals process. When I patch Solidity to enforce a geographic restriction at the protocol level rather than the frontend, I am converting a legal claim into a computational fact. Most issuers never do this. The gap between the legal claim and the computational fact is where the exit risk lives.
So when the FAQ classifies cbETH and stETH together, it is classifying the legal wrapper, not the machine. A single line of assembly can collapse millions. The FAQ does not read assembly.
The secondary market illusion
Holders believe they can exit via the secondary market at any time. This is true until it is not. cbETH and stETH both trade on DEXs, on centralized venues, and in lending markets. The liquidity looks deep on a green day. It is reflexively provided. Market makers quote because they believe they can hedge their inventory by exiting the underlying. When the exit path congests, the hedge breaks, spreads widen, and the quoted liquidity vanishes precisely when holders need it.
I modeled this during the 2022 Compound V3 review, running the liquidation engine under extreme volatility on a local fork. The finding that transferred to LSTs is this: secondary market liquidity for a redeemable claim is a derivative of the redemption path's reliability, not an independent source of liquidity. When redemption congests, secondary liquidity is the first thing to disappear, because it is the first thing that was never really there.
The discount on stETH in 2022 was not a market failure. It was the market correctly pricing the redemption path. The FAQ, by classifying the token without classifying the path, encourages holders to believe the discount was an anomaly. It was the honest price.
Contrarian: The Classification Protects the Issuer, Not the Holder
Here is the angle the market is not pricing. The FAQ is not primarily a disclosure regime. It is a definitional regime. And definitional regimes benefit the party that can control the definition.
Consider who can modify the exit path after classification. For cbETH, Coinbase can change the qualified-account criteria, suspend unwrap in a jurisdiction, or reclassify which accounts are eligible โ all subject to its own compliance posture, and all within the contractual framework the FAQ describes. The restriction on rehypothecation is a description of the current model, not a cryptographic guarantee. The holder's protection is the enforcement apparatus, which is slow.
For stETH, the exit path can be modified by governance. Lido's parameters โ operator set, fee structure, withdrawal queue mechanics โ are all governance-controlled. A holder who does not vote is a holder who is governed. The FAQ classifies the token as issued and forgets that the issuance is continuously renegotiated by a token-governance process the average holder does not follow.
So the classification does not freeze the risk. It names a category and assumes the category is stable. The category is stable only at the legal layer. At the execution layer, both exit paths are live systems that can be changed by the party with the most influence over them.
Trust the math, verify the execution. The math here says the queue is a maturity. The execution says the maturity can be extended by a governance vote you will not see.
There is a second contrarian read. The FAQ's focus on whether a staking token is a security distracts from the question that actually determines holder outcomes: is the exit path trust-minimized? A token can be a non-security and still have a redemption path controlled by a single compliance team. A token can be a security and have a redemption path enforced by immutable bytecode. The regulatory category and the technical safety are orthogonal. Conflating them is the error the market is making as it bids the sector on the FAQ headline.
I have made this argument before in the context of institutional custody. My 2024 review of BlackRock's IBIT multisig and cold storage disclosures found the same structure: the security model is only as strong as the key management, and the key management is only as transparent as the filing requires it to be. The FAQ is a filing. It is not an audit of the exit path.
The withdrawal queue as a stress instrument
Let me be specific about the stress scenario, because vague risk talk is how holders get liquidated.
Ethereum processes validator exits at a bounded rate per epoch. Assume net exit demand spikes โ a wave of stETH redemptions, a cbETH unwrap surge, or both simultaneously. The queue lengthens. Secondary discounts widen. Lending protocols that accept LSTs as collateral begin to see margin calls as the collateral discount widens against a fixed debt. Forced selling into a widening discount deepens the discount. The feedback loop is mechanical and fast.
In this scenario, the cbETH holder faces a compounded problem. If unwrap is gated by compliance and the secondary market has discounted, the holder has one real option: sell the discount. The stETH holder has the same secondary option plus the queue. Neither path is safe. The point is that the FAQ did not measure either path. It measured the wrapper.
Efficiency is not a feature; it is the foundation. The efficiency of the exit path is the only liquidity that is real.
I have watched three cycles now where a regulatory document was read as a risk clearance. Each time, the technical risk persisted and the market paid for the confusion. The 2021 NFT audit I did on OpenSea's v2 marketplace โ 400 hours tracing off-chain indexing against on-chain settlement โ taught me that the whitepaper's promise of atomic swaps did not survive contact with the EVM execution steps. The same gap exists here. The FAQ promises a classification. The exit path delivers a queue.
Takeaway: Classify the Path, Not the Token
The forward question is not whether cbETH or stETH will be deemed a security. The forward question is which issuers will convert their contractual exit promises into computational guarantees before the next congestion event forces the market to discover which of the two they actually bought.
I would watch three signals. First, whether any issuer moves unwrap permissioning from a compliance perimeter into an on-chain allowlist with published, versioned criteria. That converts a promise into a spec. Second, whether any LST governance proposal reduces the withdrawal queue's dependence on new deposits to clear exits โ because a queue that needs inflows to pay outflows is a structure with a hidden reflexivity. Third, whether lending markets that accept LSTs as collateral begin applying exit-path-adjusted haircuts rather than generic LST haircuts. When collateral haircuts diverge between cbETH and stETH, the market will have admitted what the FAQ did not.
Until then, the classification stands. The queue stands behind it. And when the congestion comes, holders will learn the distinction the FAQ skipped โ that a token is only as liquid as its worst exit ramp, and the ramp closes fastest when everyone runs for it at once.
History is immutable, but memory is expensive. The 2022 discount was the market paying tuition. The FAQ is offering a chance to skip the class. Most holders will take it. The ledger will record the exam anyway.
Appendix: What the FAQ Measured Versus What It Did Not
| Layer | FAQ Coverage | Execution Reality | |-------|--------------|-------------------| | Token category | Explicit | N/A โ legal construct | | Issuer restriction on rehypothecation | Assumed shared | Contractual for cbETH, bytecode for stETH | | Unwrap / exit gating | Not differentiated | Compliance perimeter vs queue | | Queue latency under stress | Not addressed | Bounded by epoch exit rate | | Secondary market liquidity | Implied as present | Reflexive, vanishes under congestion | | Governance mutability of exit path | Not addressed | Live for stETH, corporate for cbETH |
The table is the whole argument in one structure. The FAQ covers the top row and assumes the rest. The rest is where the capital is. A classification that does not model the exit path is a classification of the label, not the asset. Audit the path. Verify the ramp. The classification is the easy part.
Volatility is the tax on unproven utility. The discount on a congested redemption path is not volatility. It is the toll.