What is staking? How proof-of-stake yield works, what it pays in 2026 and what it risks

Who this is for: Allocators, treasurers, founders and serious holders who want to understand where staking yield comes from, compare solo, liquid, custodial and ETF staking on a net-of-fees and exit-time basis, and judge the slashing, liquidity and regulatory risks before committing capital.

Proof-of-stake networks replace the energy-intensive mining of Bitcoin with a system where validators put up capital as a bond. Validators propose and attest to blocks, are paid for honest participation and are penalised for going offline or misbehaving. Since Ethereum’s switch to proof of stake in September 2022, staking has become the central question in how institutions think about holding ether: whether to stake, through whom, at what fee, and with what exit terms. The same questions now apply to Solana and to every other large proof-of-stake network.

The stakes rose in 2025 and 2026. US securities staff said in May 2025 that protocol staking is not a securities offering, and by the first half of 2026 staking yield was being passed to ETF shareholders by Grayscale, BlackRock and Morgan Stanley. Treasury companies such as Bitmine stake the large majority of their ether. At the same time, the first week of October 2026 produced the longest exit queue of the year when a single Lido node operator, MetaMask Staking, began exiting roughly 523,000 ETH as a precaution after a security incident. Staking has become both more institutional and more visibly dependent on operator behaviour and queue mechanics.

This guide explains how the yield is generated, the four main ways to stake and what each costs, how the entry and exit queues work, what slashing has actually looked like, how staking entered ETFs and tax law, and how to compare the options with a worked example on $1 million of ether.

Staking by the numbers

How the yield is generated

Two sources feed staking rewards. The first is issuance: the protocol mints new tokens and distributes them to validators as payment for securing the network. On Ethereum, issuance is deliberately low. Staking Rewards put the network’s annual inflation at about 0.86 percent on October 6, 2026. Because the issuance budget is set by a formula that grows with the square root of total stake rather than in proportion to it, each individual staker’s yield falls as more ether is staked. That is the main reason Ethereum’s validator APR has drifted from the 3 to 4 percent range of 2023 and 2024 to about 2.6 percent in October 2026, as the staked share rose past a third of supply (ethereum.org reported 35 percent staked and a 2.5 percent APR on October 6, 2026).

The second source is execution-layer income: transaction fees paid as priority tips to the block proposer, and maximal extractable value (MEV) captured when a proposer or its builder orders transactions profitably. This component is volatile. It spikes when on-chain activity is busy and shrinks when chains are quiet, which is why validator APR readings on explorers move from day to day while issuance barely changes. On Solana, the equivalent is priority fees and MEV tips distributed through Jito’s infrastructure, which Staking Rewards notes have become a growing share of validator income.

Solana’s higher headline yield is mostly a function of its issuance schedule. Staking Rewards reported an annual inflation rate of 3.65 percent on October 6, 2026, declining 15 percent a year toward a terminal 1.5 percent, against a 5.0 percent reward rate for stakers. Because about 69.5 percent of SOL is staked, each staker receives a smaller slice of a larger pie than an Ethereum validator does. For a refresher on how issuance schedules are designed, see the CCS guide on how to read a token supply schedule.

Nominal yield versus real yield

Rewards are paid in the native token, so the dollar value of staking income depends entirely on the token’s market price. Staking increases the number of tokens you hold, not their value. Dilution is the second subtlety: if a network issues 3.65 percent a year and you earn 5 percent by staking, your real gain in share of supply is roughly the difference, while unstaked holders are diluted by the full issuance rate. On Ethereum, where issuance is under 1 percent and part of each fee is burned, the drag is small and the yield is closer to a true cash flow.

The issuance debate of 2026

Ethereum’s issuance curve is now a live policy argument. In August 2026 six researchers including Justin Drake of the Ethereum Foundation published EIP-8363, a tapered issuance burn that would destroy a growing fraction of validator rewards as total stake approaches 60.25 million ETH, about half of supply, phasing in over 18 months, according to Cointelegraph’s August 5, 2026 report. Critics argued it would squeeze solo validators and institutional demand alike. The proposal is competing for inclusion in the Hegotá upgrade expected in the second quarter of 2027. Any staker with a multi-year horizon should treat the issuance schedule as a variable, not a constant.

Four ways to stake, and what each costs

Every route starts from the same protocol yield. What differs is who runs the validator, who holds the keys, what fee is taken and how quickly you can get out.

Solo and institutional self-staking

Running a validator directly on Ethereum requires at least 32 ETH, dedicated hardware, high uptime and operational discipline. Since the Pectra upgrade on May 7, 2025, EIP-7251 lets a single validator hold up to 2,048 ETH of effective balance with compounding withdrawal credentials, so a large holder no longer needs to run hundreds of 32 ETH validators. Pectra’s EIP-7002 also allows exits to be triggered from the withdrawal address rather than the validator signing key, which simplifies custody arrangements. Solo stakers keep the full validator APR (about 2.6 percent on October 6, 2026) but carry all slashing and downtime risk and face the queues directly.

Staking-as-a-service

Most large holders and all ETFs use professional node operators such as Figment, Kiln, Coinbase Cloud, Twinstake or Galaxy, which run the infrastructure while the client retains ownership of the ether through its own custodian. The SEC’s May 2025 statement described this as self-custodial staking with a third party. Fees are typically a percentage of rewards and are negotiated; Staking Rewards listed operator fees around 5 percent for providers such as P2P.org in October 2026, and Fidelity’s August 2026 ETF filing budgeted 15 percent of rewards for staking fees. Figment was the fastest-growing provider in 2025, reaching 4.5 percent of all staked ETH by August 2025 according to Dune data reported by CoinDesk. In August 2026 BNY said it would add staking to its custody platform with Galaxy as infrastructure partner, pending regulatory approval.

Liquid staking

Liquid staking protocols pool deposits, distribute them across a set of node operators and issue a tradable token that represents the staked position plus accrued rewards. Lido’s stETH is the largest. As of October 6, 2026, Lido reported 9.83 million ETH staked and a 2.2 percent APR after its 10 percent protocol fee, which is split between node operators and the DAO treasury. DefiLlama’s liquid staking tracker showed 17.33 million ETH ($47.0 billion) across all Ethereum liquid staking protocols on the same day, with Lido at 56.9 percent of that segment, Binance’s staked ETH at 21.5 percent, ether.fi at 11.0 percent and Rocket Pool at 3.0 percent.

Liquid staking tokens (LSTs) can be sold, used as collateral or deployed in DeFi while the underlying ether keeps earning. That convenience is also where the risk lives. LSTs can trade below the value of the ether they represent when many holders want out at once, as stETH did in mid-2022, and they add smart contract risk on top of validator risk. Lido’s share of all staked ETH has fallen from 32.3 percent in late 2023 to 24.4 percent in August 2025 (Dune via CoinDesk) and about 22.5 percent in October 2026 (9.83 million of 43.7 million ETH), partly because of deliberate community pressure over concentration. Lido has responded with Dual Governance, a Community Staking Module for permissionless operators, the stVaults framework of Lido V3 for institutions that want their own operator and terms, and a Curated Module v2 upgrade announced on July 28, 2026 that consolidates validators toward the 2,048 ETH ceiling and is projected to cut Lido’s validator count from about 880,000 to about 628,000.

Custodial staking

Exchanges including Coinbase and Kraken stake customer balances and pass through part of the rewards. The customer holds a claim on the exchange, not the validator. On October 6, 2026, Coinbase advertised 1.70 percent APY on ETH; CoinDesk reported in January 2026 that Coinbase’s retail commission runs up to 35 percent of rewards, which is consistent with the gap between the protocol APR and the advertised rate. Kraken showed 2.72 percent for bonded ETH and 6.88 percent for bonded SOL before commission on the same day. Coinbase Prime markets institutional staking integrated with its qualified custodian, Coinbase Custody Trust Company, and says it holds more staked ETH than any other institutional provider. The custodial route is the simplest and the one most exposed to counterparty failure, a lesson from the 2022 exchange collapses covered in the CCS guide on crypto custody.

Restaking

Restaking, popularised by EigenLayer, lets already-staked ether (or an LST) be pledged again to secure additional services for extra yield and extra risk. DefiLlama tracked $10.9 billion of restaking TVL on October 6, 2026, with EigenCloud at $7.1 billion and Babylon, which restakes bitcoin, at $3.5 billion. EigenLayer added bitcoin collateral alongside staked ETH in September 2026, as CCS reported. Restaking stacks slashing conditions from each service on top of base-layer slashing, so the yield premium is compensation for correlated risk. The CCS guide on what restaking is covers the mechanics, and our interview with Symbiotic co-founder Misha Putiatin discusses where the model goes next.

Entry and exit queues: the liquidity you did not price

Ethereum rate-limits how much stake can join or leave per epoch to protect consensus from sudden swings in the validator set. Since Pectra the churn limit is 256 ETH per epoch, which works out to about 57,600 ETH per day in each direction. When demand to enter or exit exceeds that, a queue forms. The queues are the single most under-appreciated variable in staking economics because they convert a liquid asset into a time-locked one at exactly the moment many people want liquidity.

The first week of October 2026 illustrated the point. On September 29 the exit queue held about 166,000 ETH. On September 30 and October 1, MetaMask disclosed a security incident in part of its infrastructure and, as a precaution, began exiting the validators that MetaMask Staking (formerly Consensys Staking) runs as a Lido node operator: about 523,000 ETH across roughly 17,000 validators, according to CoinDesk. The exit queue peaked at about 851,000 ETH on October 2 and stood at about 786,000 ETH with a 14 day wait on October 5. Lido said no action was required from stETH holders, pointed to its reserve fund of more than 6,750 stETH, and said the ether would be restaked over up to 45 days, with the last affected validators stopping by October 7. On October 6, validatorqueue.com showed an exit queue of 857,082 ETH with a 14 day 21 hour wait, plus a sweep delay of about 7.5 days before withdrawn ether reaches the withdrawal address.

The entry side was congested too. CoinDesk reported about 1.5 million ETH ($4 billion) waiting roughly 25 days to start staking on October 5, 2026, down from about 2 million ETH and a 35 day wait in early September. validatorqueue.com showed 1,432,630 ETH and a 24 day 21 hour entry wait on October 6. Lido’s April 2026 case study of exiting 6,918 validators for the departing operator A41 found the exit queue swung from an 18.6 day planning assumption to under one day while the activation queue stretched to about 43 days. Queue times are unpredictable, and a staking plan needs a liquidity buffer sized for weeks, not hours.

Solana works differently. Stake activates and deactivates at epoch boundaries, roughly every two to three days, with a cap of 25 percent of active stake per epoch, per Staking Rewards, so SOL unstaking takes days rather than weeks. Other networks impose fixed unbonding periods of two to four weeks. See the CCS Solana guide for its validator economics.

Slashing: what the record actually shows

Slashing is the penalty for provable misbehaviour: signing two conflicting blocks or attestations in the same slot, or voting in a way that contradicts a validator’s earlier votes. It is distinct from inactivity penalties, which slowly drain a validator that is offline and are usually small unless the whole network is struggling to finalise. A slashed Ethereum validator loses an initial penalty, is forced to exit, and after a delay of about 36 days receives a second correlation penalty that scales with how much other stake was slashed in the same window. Pectra reduced the initial penalty sharply for consolidated validators (to 1/4096 of effective balance under EIP-7251) so that a 2,048 ETH validator is not punished disproportionately for a single fault, while keeping the correlation penalty as the deterrent against coordinated attacks.

The historical record is reassuring but not empty. The notable slashing incidents on Ethereum have been operational accidents rather than attacks: an operator running the same keys on two machines during a failover, or a buggy migration. Examples include a roughly 75 validator slashing at the operator Staked in February 2021 and smaller incidents at Lido node operators in 2023, each costing a few tens of ETH. Set against more than 860,000 active validators in October 2026, cumulative slashing losses have been a rounding error. Reputable operators carry insurance or indemnities, and Lido keeps a stETH reserve fund for operator incidents. But the risk is yours unless your agreement says otherwise, and a correlated event at a large operator would be a different magnitude of loss than anything seen so far.

Solana does not slash automatically in the way Ethereum does; misbehaving validators have historically lost rewards rather than principal, so delegator risk there is dominated by validator performance and commission. For how networks reach agreement and punish faults, see the CCS guide on how blockchain consensus works.

Regulation: from enforcement to ETFs

For two years the US treated staking-as-a-service as a potential securities offering. On February 9, 2023 Kraken agreed to pay $30 million and shut its US staking program after the SEC alleged that its pooled staking product, which had advertised returns as high as 21 percent on some assets, was an unregistered offering. The same theory featured in the SEC’s June 2023 case against Coinbase and kept staking out of the spot ether ETFs approved in July 2024. Our guide on SEC versus CFTC crypto regulation covers the background.

The turn came on May 29, 2025, when the SEC’s Division of Corporation Finance issued a statement that protocol staking activities, covering solo staking, self-custodial staking with a third-party operator and custodial staking, do not involve the offer and sale of securities. It also covered ancillary services such as slashing coverage, early unbonding and aggregation to meet minimums. A further staff statement in August 2025 extended the analysis to liquid staking, and on September 26, 2026 the Division published FAQs saying staking receipt tokens are not securities when they function as proof of deposit of a digital commodity and the provider does not lend, pledge or rehypothecate the coins, as CCS reported. The Coinbase case was dismissed in February 2025. All of this is staff interpretation rather than rulemaking, and a future Commission could withdraw it.

Staking inside ETFs

The ETF market moved quickly once the staff position was clear. Grayscale became the first US issuer to enable staking in spot crypto ETPs in October 2025, distributing rewards from its Ethereum trusts and its Solana trust. VanEck registered a Delaware trust for a Lido Staked Ethereum ETF on October 2, 2025, a product that would hold stETH directly. Bitwise withdrew a staking proposal in September 2025. On March 12, 2026 BlackRock launched the iShares Staked Ethereum Trust ETF (ETHB) on Nasdaq with a 0.25 percent sponsor fee, waived to 0.12 percent on the first $2.5 billion for a year, alongside its unstaked iShares Ethereum Trust (ETHA), which then held about $6.5 billion. On July 28, 2026 Morgan Stanley launched ether and Solana ETFs at 14 basis points that pass 100 percent of staking rewards to investors. On August 12, 2026 Fidelity filed to stake up to 100 percent of the Fidelity Ethereum Fund (FETH), excluding reserves for redemptions and liquidity, with 85 percent of rewards retained by the fund and 15 percent going to staking fees, paid out as quarterly cash distributions. FETH had about $2.13 billion of cumulative net inflows at the time.

The ETF wrapper changes the risk shape rather than removing it. Shareholders cannot unstake, transfer or control the ether; the issuer decides how much to keep liquid for redemptions and bears the exit queue on their behalf. CoinDesk noted in January 2026 that Grayscale’s legacy Ethereum Trust charges a 2.5 percent fee against a staking yield of roughly 2.8 percent before the provider’s cut, so the fee decides the economics. The CCS ETF tracker follows flows, and the guide on how spot bitcoin ETFs work explains the creation and redemption mechanics staked products must work around.

Treasury companies

Corporate treasuries are now among the largest single stakers. Bitmine said on September 14, 2026 that it held 5.95 million ETH, about 4.9 percent of supply, with more than 5.06 million ETH staked, and projected $334 million in annualised staking revenue, roughly 2.5 percent on the staked position. For a company whose stock trades on ether per share, staking yield is the operating income. See the CCS guide on treasury companies.

Tax treatment of staking rewards

In the United States, Revenue Ruling 2023-14 holds that a cash-method taxpayer who receives staking rewards must include their fair market value in gross income in the year the taxpayer gains dominion and control over them, meaning the ability to sell, exchange or otherwise dispose of the tokens. The value is measured at that moment, not when the validation happened, which matters when a protocol or provider delays access to rewards. The ruling applies whether you stake directly or through an exchange. Those rewards then carry a cost basis equal to the amount recognised, so a later sale produces capital gain or loss on the difference.

The practical consequences are significant. A validator earning rewards every epoch generates thousands of small income events a year, each with its own basis, so record keeping is a real cost of solo staking. An LST that accrues value inside the token, whether rebasing like stETH or value-accruing like wstETH or rETH, raises the unresolved question of whether income arises on each accrual or only on sale; the IRS has not ruled on it. ETF shareholders receive fund-reported distributions, a large part of the appeal for taxable US investors. The Jarrett litigation, in which a Tennessee couple argued that staking rewards are self-created property taxable only on sale, remains the main challenge to the IRS position. The CCS guide on US crypto tax rules covers broker reporting and lot tracking.

How we got here: a timeline

Worked example: staking $1 million of ether four ways

How to evaluate a staking arrangement: a checklist

• What is the fee, and on what? A percentage of rewards (Lido 10 percent, Coinbase up to 35 percent, operators 5 to 15 percent) is different from a percentage of assets (an ETF’s 0.25 percent). Convert both to net yield at today’s APR before comparing.
• Who bears slashing? Read the indemnity. Operators often cover losses from their own faults but not from protocol bugs or correlated events. Lido’s reserve fund held more than 6,750 stETH in October 2026; ask what the equivalent is elsewhere.
• What is the exit path and how long does it take under stress? Ask for the provider’s record during the October 2026 queue, when exits ran about 15 days. An LST gives you a market exit; a custodial product gives you the protocol queue plus the provider’s own processing.
• Who holds the keys? Self-custodial staking with a third-party operator leaves ownership with you. Custodial staking does not. The SEC distinguishes the two, and so should your risk committee.
• How concentrated is the operator set? Lido spreads stake across dozens of curated and permissionless operators, yet a single operator’s exit moved the whole network queue in October 2026. Ask how much of your stake sits with one operator, one client software and one cloud region.
• How is MEV handled? Does the operator use relays, smooth MEV across the pool, or keep it? This can be a meaningful share of yield in busy months.
• What are the tax mechanics? Will you receive per-epoch income records, a rebasing token, or a fund distribution? Each has a different compliance cost under Revenue Ruling 2023-14.
• What happens to yield if issuance changes? With EIP-8363 and related proposals under discussion for Hegotá in 2027, ask whether the product’s economics survive a lower issuance curve.

Risks and open questions

Concentration remains the structural concern. Even at a reduced 22.5 percent share, Lido is the largest single staking venue on Ethereum, and the exchange and ETF channels funnel stake toward a handful of professional operators. A fault in one operator’s software or key management can now move the network-wide exit queue, as MetaMask’s precautionary exit showed. Slashing penalties that scale with correlation make the design tolerant of this, but tolerance is not immunity, and the community’s own proposals to rein in staking growth reflect unease about how much of the asset sits in a few hands.

Liquidity is the second risk. In a market panic the fastest exit from a staked position is to sell an LST or an ETF share at whatever discount the market demands. Products that promise instant redemption of staked ether are either holding a large unstaked buffer, which drags on yield, or taking liquidity risk on the client’s behalf. Ask which.

Regulatory risk has shrunk but not vanished: everything that opened staking to ETFs in 2025 and 2026 rests on SEC staff statements and FAQs, not legislation or Commission rules, and a change of administration could reverse it. Finally, yield itself is a policy variable. Ethereum’s issuance curve is openly contested and Solana’s inflation is scheduled to fall 15 percent a year, so any model that assumes today’s APR persists for a decade is wrong by design.

What to watch next

Glossary

Validator

A node that has posted stake and participates in proposing and attesting to blocks. On Ethereum one validator holds between 32 and 2,048 ETH.

Effective balance

The portion of a validator’s balance that counts toward its voting weight and rewards, capped at 2,048 ETH since Pectra.

Issuance

New tokens created by the protocol and paid to validators. The main source of staking yield on most networks.

MEV

Maximal extractable value: profit a block proposer can capture by ordering, including or excluding transactions. Part of execution-layer rewards.

Slashing

A penalty that destroys part of a validator’s stake and forces it to exit, applied for provable faults such as signing conflicting blocks.

Churn limit

The maximum stake that may enter or leave the validator set per epoch, 256 ETH on Ethereum since Pectra, about 57,600 ETH a day.

Exit queue

Validators waiting to leave the active set because exits exceed the churn limit. Measured in ETH and in days of expected wait.

Sweep delay

The time after a validator exits before its balance is actually paid to the withdrawal address, about a week on Ethereum in October 2026.

Liquid staking token (LST)

A transferable token such as stETH or rETH representing a share of a staking pool and its accrued rewards.

Staking-as-a-service

A professional operator runs validators on behalf of a client who retains ownership of the staked assets.

Custodial staking

An exchange or custodian holds the assets and stakes them; the client holds a claim on the intermediary.

Restaking

Pledging already-staked assets or LSTs to secure additional services for extra yield and extra slashing exposure.

Dominion and control

The IRS test from Revenue Ruling 2023-14 for when staking rewards become taxable income: when the taxpayer can sell or dispose of them.

Staking receipt token

The SEC staff’s September 2026 term for a token evidencing a deposit of a digital commodity with a staking provider, not a security when it is a pure receipt.

Why it matters

Staking changes what a proof-of-stake asset is. Ether held unstaked is a commodity with no cash flow; ether staked is a productive asset whose yield can be compared with a bond, priced, hedged and packaged. That is the basis for valuing ether on cash flow, the reason staked ETFs matter to allocators who could not otherwise justify the position, and the reason treasury companies can describe staking income as operating revenue.

The events of 2025 and 2026 added a second lesson. The yield is modest, about 2.6 percent gross on Ethereum in October 2026 and falling as participation rises, so the fee, the exit terms and the operator’s competence decide most of what an investor keeps. A single operator’s precautionary exit can reset the network’s liquidity clock for weeks. Staking has matured into infrastructure, and like all infrastructure its value is in the details of who runs it and what happens when something breaks.

Sources

1. SEC Division of Corporation Finance: Statement on Certain Protocol Staking Activities, May 29, 2025
2. SEC: Kraken to Discontinue Unregistered Offer and Sale of Crypto Asset Staking-As-A-Service Program and Pay $30 Million, February 9, 2023
3. IRS: Revenue Ruling 2023-14, July 31, 2023
4. ethereum.org: Pectra upgrade (EIP-7251, EIP-7002, EIP-6110), accessed October 6, 2026
5. validatorqueue.com: Ethereum entry and exit queue dashboard, accessed October 6, 2026
6. Lido: Total ETH staked and APR, accessed October 6, 2026
7. Lido blog: Exiting ~7,000 Validators, a Case Study in Optimizing Ethereum Validator Exits, April 2026
8. DefiLlama: Ethereum liquid staking providers, accessed October 6, 2026
9. DefiLlama: Restaking protocols TVL, accessed October 6, 2026
10. Staking Rewards: Ethereum staking metrics, accessed October 6, 2026
11. Staking Rewards: Solana staking metrics, accessed October 6, 2026
12. Coinbase: Ethereum staking, accessed October 6, 2026
13. CoinDesk: Ethereum has a 25-day wait to start staking, with nearly 1.5 million ETH in line, October 5, 2026
14. Crypto Coin Show: MetaMask exits staking validators after detecting security incident, October 1, 2026
15. Crypto Coin Show: SEC clarifies staking receipt tokens are not securities when functioning as proof of deposit, September 26, 2026
16. Cointelegraph: Fidelity files with SEC to add staking to Ethereum ETF, August 12, 2026
17. Cointelegraph: Ethereum researchers want to rein in staking; critics say it could backfire, August 5, 2026
18. Cointelegraph: Bitmine projects $334M in annual staking revenue from $15.8B crypto treasury, September 14, 2026
19. CoinDesk: BlackRock debuts staked ether ETF as demand grows for yield in crypto funds, March 12, 2026
20. CoinDesk: Crypto ETFs with staking can supercharge returns but they may not be for everyone, January 25, 2026

Disclosure: This guide is for education only and is not investment, legal or tax advice.