Who Controls Your Staked ETH: Lido vs Rocket Pool vs Frax

Every liquid staking pitch sounds the same. Deposit ETH, receive a token that keeps earning while you use it elsewhere, and skip the 32 ETH minimum and the hardware. Lido, Rocket Pool, and Frax all make that promise, and on the surface their products rhyme: you send ether, you get a receipt that tracks a validator’s rewards, and you can trade or lend that receipt whenever you want.

The pitch hides the only question that matters when something breaks: who actually controls the ETH once it leaves your wallet? Who picks the operators running the validators? Who can change the fees, pause a contract, or decide which transactions get built into a block? Who does a regulator call when it wants a sanctions list honored? Staking is easy. Control is the hard part, and the three biggest names in Ethereum liquid staking answer it in almost opposite ways.

2026 sharpened the contrast. In September, Lido began the largest validator consolidation in its history, squeezing hundreds of thousands of validators into far fewer, far larger, and mostly institutional operators. Rocket Pool cut its node bond in half and doubled down on letting anyone in the world run a validator. Frax finished renaming its token and folded staking into a chain it controls end to end. Same category, three different theories of who should be trusted with your money. All prices and figures below are as of September 28, 2026.

Staking is easy; control is the hard question

Liquid staking is now a huge share of Ethereum. About 35.7% of all ETH is staked, roughly 43.6 million ETH spread across close to 889,000 validators, earning a base consensus yield near 2.58% before tips and MEV. Most of that stake does not belong to the people running the validators; it belongs to depositors who handed their ETH to a protocol and took back a token. That handoff is where control quietly changes hands, and it is the part the marketing never mentions.

To compare Lido, Rocket Pool, and Frax honestly, judge them on five axes, not just yield. First, the operator set: who runs the validators, and can anyone join? Second, governance: who can change the code, the fees, and the operator list? Third, key custody: who can move, or lose, the underlying stake? Fourth, censorship resistance: who decides which transactions land in a block? Fifth, exit: can you actually withdraw, and how fast? A protocol can pay a great yield and still fail every one of the harder tests. Yield tells you what you might earn. These five tell you what you are actually trusting.

What you actually hold: stETH, rETH, and the Frax two-token trick

The receipt you get back is not the same across the three. Lido issues stETH, a rebasing token: your balance grows a little every day as rewards land, and it trades close to 1:1 with ETH. Most DeFi uses the wrapped version, wstETH, which holds a fixed balance and rises in price instead of quantity. Rocket Pool issues rETH, a reward-bearing token with no rebasing; it starts at parity and slowly appreciates against ETH as rewards accrue, so one rETH is now worth roughly 1.17 ETH.

Frax splits the job across two tokens. frxETH is a liquid, roughly 1:1 ether proxy that on its own earns nothing. sfrxETH is an ERC-4626 vault that concentrates all of the staking yield among the people who deposit into it, so it climbs against ETH the way rETH does. The design is not cosmetic. A rebasing token needs the protocol to push balance updates and generates a running stream of taxable events even if you never sell, while a reward-bearing token folds rewards into an exchange rate and tends to surface tax only when you dispose of it. It also shapes how the token behaves as DeFi collateral, where a wrong price feed can trigger liquidations. With ETH trading around $2,686 and stETH near $2,649 per CoinGecko, the three receipts look interchangeable on a price chart. What differs is who stands behind each one.

Lido: the giant that runs on a curated list

Lido is the giant. It accounts for roughly 23% of all staked ETH, close to 9.8 million ETH and about $26 billion in value, and it dominates the narrower liquid-staking segment with around 62% of it, according to Datawallet. Those numbers drive most of the coverage. The control fact behind them gets less attention: for most of its life, Lido did not let just anyone run a validator.

Lido’s stake has been allocated across a curated set of roughly 34 professional node operators, each admitted and removable by DAO vote. A newer Community Staking Module opened a permissionless path for solo stakers willing to post a modest bond, but the bulk of the ether still sits with the curated set. So when you hold stETH, your ETH is validated by a hand-picked list, not an open market. That is the trade Lido made for reliability and scale, and it is the reason its share has drifted down from a peak near 32% in late 2023 to about 23% today, partly through self-imposed restraint and partly because exchanges and institutional stakers have absorbed new deposits. The result is a paradox that defines Lido: it is simultaneously the most battle-tested option and the one whose sheer size makes decentralization advocates nervous.

The Lido Core consolidation, and what it does to decentralization

The biggest structural change to Lido in years landed this quarter. In July 2026 the Lido DAO approved Curated Module v2 (CMv2), and the first live validator consolidations executed in September, with asset manager Bitwise carrying out the first one. The plumbing comes from Ethereum’s Pectra upgrade of May 2025, whose EIP-7251 created a new validator type that can hold up to 2,048 ETH in effective balance instead of the old 32 ETH cap. Lido is using it to merge roughly 265,000 legacy validators into far fewer, far larger ones.

The scale is hard to overstate. The migration restructures about 8.4 million staked ETH, worth roughly $16.5 billion, cuts Lido’s own validator count by about a third, and drags the whole network from around 880,000 validators toward roughly 628,000, trimming attestation traffic by about 29% per epoch, per The Block. The share of network stake sitting in auto-compounding validators nearly doubles, from about 32% to 52%, and CoinDesk reports that full migration is targeted for the first quarter of 2027.

Here is the control tension. Consolidation is genuinely good for Ethereum’s performance: fewer messages, less bandwidth, faster finality, and rewards that compound automatically instead of being swept out. But it concentrates. Each surviving operator now runs a bigger slice of the total, so a bug or a key compromise at one operator touches more stake than before, and larger validators shoulder proportionally more of the data-availability sampling load that Fusaka introduced. It is telling that Bitwise reached CMv2 operator status through its 2026 acquisition of Chorus One, an existing Lido operator, as Crypto Briefing noted; the drift is toward regulated institutions running larger validators. CMv2 does add a real check by forcing all 34 curated operators to post ETH bonds for the first time, putting their own capital at risk alongside yours. Efficiency up, operator count down, institutional weight up: that is the whole trade in one sentence.

Dual governance: can stETH holders really fire the DAO?

If a curated operator set and a concentrating validator base worry you, Lido’s answer is dual governance. For years the awkward truth was that LDO holders, a comparatively small group, governed a protocol that runs on nearly 9.8 million ETH of other people’s stake. Dual governance, approved on June 30, 2025 and live since July 4, gives stETH holders a brake they never had.

The mechanism runs through an escrow contract that accepts stETH, wstETH, and withdrawal tickets. Once more than 1% of the stETH supply sits in that escrow, veto signaling kicks in and delays any DAO motion by 5 to 45 days, scaling with how much stake objects. If more than 10% piles in, a rage-quit triggers: governance freezes entirely until the dissenting stakers finish withdrawing, under a timelock that can stretch from 60 to 180 days, per Lido’s own explainer and reporting from Unchained. Stakers cannot pass proposals, but they can block them and leave. That is a real check, and it is a very different kind of governance from the sort you can quietly accumulate votes in or borrow your way into; the same theme runs through our look at the governance you cannot flash-loan. It is worth being honest about the limit, too: dual governance is a defensive tool, useful in a crisis, not a day-to-day say over fees or operator selection.

Rocket Pool: permissionless by design, small by consequence

Rocket Pool starts from the opposite premise. There is no curated list and no DAO vote to become an operator: anyone with enough ETH and some RPL collateral can spin up a node and start validating. The network runs on roughly 2,000 independent operators and about $1.1 billion to $1.2 billion in value, which makes it the third-largest liquid staking provider and, by most measures, the most trust-minimized of the three. Governance is split between a protocol DAO of RPL holders and an oversight DAO of vetted members, a design meant to keep any single group from steering the protocol alone.

The Saturn One upgrade, live on mainnet in February 2026, pushed that further. It cut the node-operator bond from 8 ETH to 4 ETH per validator, with the remaining 28 ETH sourced from the rETH pool, halving the capital an individual needs to run a validator. It introduced megapools that group many validators under one contract to save gas, and a withdrawal buffer to steady rETH liquidity, per the Saturn project site. A reworked commission model called UARS pays operators a base of around 5% plus up to roughly 9% more routed to those who also stake RPL, now paid in ETH rather than fresh inflation, as The Defiant details, and RPL emission is winding down. The catch is size: maximal decentralization and censorship resistance, but a small fraction of Lido’s stake, because open participation is simply harder to scale than a curated roster. If Lido optimizes for scale first, Rocket Pool optimizes for credible neutrality first and accepts the smaller footprint that comes with it.

Frax: the most centralized bet, and the most honest about it

Frax runs the smallest validator footprint and is the most team-directed of the three. Its liquid staking splits into frxETH, a liquid ether proxy that earns nothing by itself, and sfrxETH, the vault that concentrates the yield among its depositors. A later frxETH v2 turned validator provisioning into a node-operator lending market, where operators borrow ETH against collateral to run validators rather than the protocol spreading stake across an open set. None of this pretends to be a validator democracy; Frax staking is a product the Frax core team runs.

In 2026 that direction became explicit. Under a plan called North Star, Frax renamed its FXS governance token to FRAX, announced on December 30, 2025 with the swap executed at the end of April 2026, and made FRAX the gas token and sole commodity token of its own layer-2 network, Fraxtal, with a defined issuance schedule that starts near 8% and steps down by a point a year to a 3% floor, per the Frax governance forum and PANews. Bybit handled the 1:1 token swap and Binance listed the new FRAX and integrated Fraxtal in January 2026. Staking is now one piece of a vertically integrated stack that the team steers from top to bottom. That is the highest protocol and team risk of the three, but it is also the most nimble, and Frax is unusually upfront that this is the bargain: you are betting on a team and an ecosystem, not on an open operator set.

The trust spectrum, side by side

Put the control model of each protocol next to the others and the differences are sharper than any yield table. This is the comparison that actually decides who holds power over your stake.

Censorship resistance: who can filter your transactions

Running a validator is not the same as deciding what goes inside the block it proposes. On Ethereum, most validators outsource block building through MEV-Boost, which routes more than 90% of blocks, and the builder market behind it is concentrated, with the top three builders assembling the large majority of blocks. Relays sit in the middle, and a relay can choose to filter transactions, for example ones touching sanctioned addresses. That is where the operator-set question stops being abstract and starts deciding whether your transactions are guaranteed a path onto the chain.

A permissionless operator base is more likely to include operators who run censorship-resistant relays or build blocks locally, because no central party can force a compliance policy on them. A curated or heavily institutional set is more exposed to compliance-driven filtering, because regulated firms have every incentive to honor sanctions lists. That makes Rocket Pool structurally the most censorship-resistant of the three, and it puts a question mark over a post-consolidation Lido whose stake increasingly sits with regulated operators like Bitwise. Ethereum is building protocol-level defenses against exactly this, such as inclusion lists that force proposers to include eligible transactions, but they are not fully in force yet. The dynamic rhymes with the concentration debate in proof-of-work; we mapped the same tension in Bitcoin’s mining pool leaderboard blind spot and in the question of who actually secures the network. Client diversity is a related pressure point: a single consensus client and a single execution client each sit above 50% of the network, so a bug in either could ripple through every operator at once, no matter how decentralized the roster looks on paper.

Why Lido’s size is Ethereum’s problem, not just yours

Concentration in staking is not only a personal risk; it is a network risk, and it has a well-known threshold structure. A single entity that controls more than a third of stake can, in theory, delay finality; one past two-thirds could finalize the chain on its own. Lido at about 23% of all staked ETH, and roughly 62% of the liquid-staking segment, sits close enough to those lines that its operator and governance choices are systemic, not just commercial.

Ethereum co-founder Vitalik Buterin has warned for years about the danger of “the potential for a single liquid staking token to take over the ‘money’ network effects from Ethereum itself,” a concern he raised without naming any one provider, reported by The Block. The consolidation sharpens the point rather than softening it: fewer, larger operators mean each one carries more of the load, so the blast radius of any single failure grows even as the raw validator count falls. This is exactly why dual governance exists, and why the debate over whether Lido should cap its own share keeps resurfacing. A protocol big enough to threaten the chain it depends on has to keep spending real effort to prove it will not, and stakers are right to treat that effort as part of the product they are buying.

The numbers: market share, yield, and the risk-free rate

Control is the story, but the figures frame the stakes. Here is where the three protocols and their tokens sit as of September 28, 2026.

Token prices come from CoinGecko, which puts LDO near $0.46 with a market cap around $382 million and RPL close to $1.96 with a cap near $45 million. The governance tokens are a rounding error next to the stake they oversee, which is part of why dual governance mattered in the first place: whoever controls a cheap token should not casually control expensive stake. And the yield needs context. With the Federal Reserve’s H.15 release showing the 3-month Treasury bill at 4.24% as of September 24, a dollar investor earns more risk-free in T-bills than staked ETH pays in its base yield. Staking is an ETH-denominated coupon layered on top of ETH price exposure, not a dollar yield, a point that got louder after the Fed’s hawkish turn, which we covered in crypto’s FOMC reaction. When the spread over cash is thin, fees, token design, and who controls the validators matter more than the last basis point of headline yield.

What the SEC actually decided about control

United States regulators have quietly made control the legal test, too. On August 5, 2025, the SEC’s Division of Corporation Finance said that “Liquid Staking Activities … do not involve the offer and sale of securities” under federal law, a statement that lifted a large cloud over stETH, rETH, and sfrxETH for US users, published on sec.gov. The relief is conditional, and the conditions are all about control.

To stay outside securities law, the Division said, a provider must keep to administrative and ministerial functions. It cannot decide “whether, when, or how much” of a depositor’s assets to stake, it cannot “guarantee or otherwise set the amount of rewards,” and merely selecting node operators or holding custody does not count as the kind of managerial or entrepreneurial effort that would make the token a security. Discretion is the tripwire: the moment a provider starts steering outcomes, it can fall back outside the safe harbor. SEC Commissioner Hester Peirce framed the whole arrangement as “a variant on the longstanding practice of depositing goods with an agent who performs a ministerial function in exchange for a receipt that evidences ownership of the goods,” in remarks reproduced by Free Writings. The irony is sharp: the legal comfort depends on these protocols not controlling your stake beyond the plumbing, which nudges all three toward the ministerial, non-discretionary design they each claim to run. That same permissive posture underpins the staking ETFs now reaching US investors, part of the wider wave we tracked in the 2026 ETF approvals map.

How to choose, and the failure modes to price in

No liquid staking token is risk-free, and the risks differ by protocol. Price them before you chase a few extra basis points of yield.

The choice follows from what you actually want. If you want the deepest liquidity, the widest DeFi integrations, and institutional-grade rails, Lido is the default, provided you are comfortable holding a claim on a curated, consolidating operator set and are willing to lean on dual governance as your backstop. If your priority is decentralization and censorship resistance, or you want to run your own validator with only 4 ETH, Rocket Pool is the clearest fit, and you accept smaller size and thinner liquidity as the cost. If you are making a broader bet on the Frax and Fraxtal ecosystem and you do not mind a team holding the wheel, Frax’s sfrxETH concentrates yield inside a stack built for that thesis. The honest summary is that you are not just picking a yield; you are picking who you trust, and with how much of your ETH. Read the receipt, then read who stands behind it.

Frequently Asked Questions

Which is the most decentralized: Lido, Rocket Pool, or Frax?

Rocket Pool is the most decentralized of the three. Anyone can run a Rocket Pool validator without approval, and after the Saturn upgrade the node bond is just 4 ETH, so its roughly 2,000 operators form an open set. Lido relies mainly on a curated list of about 34 operators, and Frax runs a small team-selected set, so both concentrate control more than Rocket Pool does.

Does the Lido Core consolidation make my stETH riskier?

It cuts both ways. The consolidation makes Ethereum more efficient and forces Lido operators to post ETH bonds for the first time, which adds accountability. But it also packs more stake into fewer, larger operators, so a failure at any one of them now affects more ETH. Your stETH is not obviously safer or riskier day to day; the change is that the network is more concentrated behind it.

Can I lose my staked ETH with these protocols?

Yes, in specific ways. Your token could trade below ETH in a market panic, an operator could be slashed for misbehavior, a smart-contract or oracle bug could be exploited, or governance could push a harmful change. Withdrawals now exist on Ethereum, which removes the 2022-style trapped-liquidity scenario, but none of these protocols is risk-free, and Frax carries the most single-team risk.

Is liquid staking legal in the US after the SEC’s 2025 guidance?

The SEC’s Division of Corporation Finance said in August 2025 that liquid staking activities do not involve the offer and sale of securities, as long as the provider stays administrative and ministerial and does not exercise discretion over staking or guarantee rewards. That gave US users real clarity, but it is staff guidance tied to those conditions, not a blanket exemption, so how a specific product is run still matters.

Why is staking yield lower than a Treasury bill right now?

Ethereum’s base staking yield falls as more ETH is staked, and with about 35% of supply staked it has compressed to around 2.58% before MEV. Meanwhile the 3-month Treasury bill yields about 4.24% after the Fed’s hawkish turn. So in dollar terms a T-bill pays more; staking makes sense as a yield on ETH you already hold and expect to appreciate, not as a substitute for cash yield.

By Yuki Tanaka, staking and mining correspondent at HOGE Wire.