ETF vs Direct Staking vs Hybrid: An Institutional Allocation Framework

TL;DR

• A staking ETF offers brokerage access, daily liquidity, and zero operational burden, at the cost of control, key custody, and governance rights.
• Direct staking through a non-custodial validator such as Everstake keeps assets under the institution’s own custody, with full validator choice.
• A hybrid model divides exposure: ETF shares for liquidity sleeves, direct delegation for the core position.
• As of April 2026, several US ETH staking ETFs are live: Grayscale ETHE and BlackRock ETHB.
• The SEC and CFTC joint interpretive release of March 17, 2026 classified protocol staking rewards as non-securities, which opened the path for these products.

Three ways institutions access staking

Institutions choose among 3 access models for staking: an exchange-traded wrapper, direct non-custodial delegation, or a hybrid allocation across both. Each model is an access structure and each assigns control, custody, and responsibility differently.

The 3 models break down as follows:

1. Staking ETF. The institution buys exchange-listed shares. The issuer holds the underlying asset with a custodian, stakes a portion of it through contracted validators, and passes net distributions to shareholders.
2. Direct staking. The institution holds the asset itself or through its own custodian, then delegates validation rights to a non-custodial provider such as Everstake. Keys never leave the institution’s control.
3. Hybrid model. The institution holds both ETF shares and directly delegated positions, sized to liquidity needs and operational capacity.

The structural difference is who holds the keys and who carries the operational work. A pension fund restricted to listed instruments would require a different model than a fund with qualified custody and a digital-asset operations desk.

How staking ETFs work, structurally

A staking ETF holds the underlying asset, stakes part of it on the network, and distributes net staking rewards to shareholders, all inside a listed fund wrapper. 

As of April 2026, at least 3 such products trade in the US for ETH: Grayscale ETHE, Grayscale Ethereum Staking Mini ETF (ETH), and BlackRock ETHB, with more issuers awaiting approval.

The mechanics follow the standard fund pattern with 3 staking-specific layers:

• Creation and redemption. Authorized participants exchange baskets of shares for the underlying asset or cash. Filings such as the iShares Staked Ethereum Trust trust agreement of February 5, 2026 permit both in-kind and cash settlement.
• Staking allocation. Issuers stake only part of the holdings. ETHB stakes roughly 70% to 95% of its ETH, keeping the rest unstaked as a liquidity buffer against redemptions.
• Key control. The fund custodian, not the shareholder, holds the keys. Validator operation is contracted to staking providers selected by the issuer.

In-kind redemption returns the asset itself to the authorized participant, while cash redemption requires the fund to sell. The unbonding queue shapes both: Ethereum enforces an exit period ranging from about a day or two when the queue is empty to over six weeks during mass-exit events, so a fund cannot stake 100% of holdings and still meet daily redemptions.

The shareholder owns fund shares, not the asset. For how issuer demand reshapes the validator market, see staking ETFs and validator demand.

How direct staking works

Direct staking means the institution stakes at the protocol level (either running validators itself or delegating operations to an external operator like Everstake), while the assets remain under its own or its custodian’s keys. The non-custodial structure separates 2 roles: the institution controls the asset, and the validator such as Everstake performs the consensus work.

Direct staking comes in 2 operational forms:

• Solo staking. The institution runs its own validators. On Ethereum this requires min 32 ETH per validator, dedicated infrastructure, client maintenance, and around-the-clock monitoring.
• Delegated staking. The institution delegates to a professional operator. The operator runs the nodes, and the institution retains withdrawal rights and asset ownership.

Delegated non-custodial staking dominates institutional practice because it removes node operations without surrendering custody. Withdrawal credentials stay with the institution, so the validator can never move, lend, or rehypothecate the assets.

Each institutional position sits in identifiable on-chain addresses rather than in a pooled fund NAV, which simplifies audit, reporting, and counterparty analysis. Non-custodial institutional staking also preserves on-chain governance rights, including voting on networks where stake carries votes.

ETF vs self-custody: the control trade-off

The core difference between an ETF position and self-custodied direct staking is who holds the keys and who exercises the rights attached to the stake. Everything else, from slashing exposure to voting, follows from that single allocation of control.

The control split looks like this:

Control question	Staking ETF	Direct non-custodial staking
Who holds the keys	Fund custodian	Institution or its custodian
Who selects validators	ETF issuer	Institution
Who bears slashing risk	All shareholders, via NAV	The institution, position-specific
Who exercises governance votes	Issuer policy, often unused	Institution, where the network supports it
Exit mechanism	Sell shares on exchange	Protocol unbonding queue

Slashing in an ETF is mutualized: a penalty hits the fund NAV and spreads across every shareholder regardless of which validator misbehaved. In direct staking the institution carries its own slashing exposure and can manage it through validator selection, coverage terms, and key architecture.

Governance is the quieter half of the trade. Networks where stake carries voting power give direct stakers a voice in protocol upgrades, while ETF shareholders hold no on-chain rights at all.

The decision framework

The structure decision resolves across 7 operational dimensions, and reward rate is deliberately not one of them. Net distributions vary by fee stack and staking ratio in every structure, so a mandate-fit comparison is the durable basis for the choice.

Dimension	Staking ETF	Direct non-custodial	Hybrid
Mandate fit	Listed-securities mandates, 40 Act-adjacent vehicles	Mandates permitting direct digital-asset custody	Mixed or evolving mandates
Custody requirement	Issuer’s custodian	Institution’s own qualified custodian	Both
Regulatory treatment	Fund share, standard brokerage reporting	Direct asset holding, digital-asset accounting	Split treatment
Governance rights	None	Full, where the network grants them	Partial
Operational capacity needed	None	Delegation workflow, monitoring, reporting	Moderate
Liquidity profile	Intraday, exchange-traded	Protocol-defined unbonding	Tiered
Slashing exposure	Mutualized in NAV	Position-specific, manageable via validator choice	Split

Each structure wins under identifiable conditions:

• The ETF wins when the mandate only permits listed instruments, when positions must be rebalanced intraday, or when no digital-asset operations capacity exists.
• Direct staking wins when custody control is required, when governance participation is part of the strategy, or when position size justifies validator-level terms.
• The hybrid wins when a liquidity sleeve must coexist with a controlled core, which describes most multi-mandate allocators.

Asset managers building this capability can review staking infrastructure for asset managers for the delegation and reporting workflow behind the direct and hybrid columns.

Custody and control by structure

Custody location is the cleanest single test for separating the 3 structures. Each model places the keys, and therefore the counterparty exposure, in a different set of hands.

The side-by-side allocation of custody:

• ETF. The issuer’s contracted custodian holds the keys. The institution’s counterparty exposure runs through the fund, the custodian, and the issuer’s chosen staking operators.
• Direct. The institution or its qualified custodian holds the keys. The validator, operating non-custodially, never gains transfer authority, so validator counterparty exposure is limited to performance, not principal.
• Hybrid. Keys split across both arrangements, with exposure sized to each sleeve.

Segregation differs. An ETF position is a pro-rata claim on a commingled pool, while a direct position sits in dedicated on-chain addresses attributable to one owner.

Custodian-integrated delegation closes the operational distance between the 2 models. A validator infrastructure for custodians lets an institution delegate from accounts at its existing qualified custodian, keeping the control profile of direct staking with custodian-grade key management.

Compliance and regulatory treatment

Regulatory treatment differs by structure, and for regulated mandates the treatment often decides the structure before any other variable is weighed. An ETF share is a listed instrument with familiar reporting, while a direct position is a digital-asset holding with its own accounting and custody rules.

The treatment splits across 3 areas:

1. Reporting. ETF shares flow through existing brokerage and fund-accounting pipelines. Direct positions require digital-asset sub-ledgers, on-chain reconciliation, and reward-event tracking.
2. Accounting. Fund shares are marked like any listed holding. Directly held assets follow digital-asset accounting standards, with staking rewards booked as they accrue or as they are received, depending on jurisdiction.
3. Jurisdiction. The SEC and CFTC joint interpretive release of March 17, 2026 classified protocol staking rewards as non-securities in the US, which enabled the current ETF wave. Other jurisdictions, including the EU under MiCA, apply their own custody and disclosure regimes to direct holdings.

A mandate written for listed instruments may have no legal path to direct staking, which makes the ETF the only available structure. Conversely, a mandate requiring asset segregation or prohibiting commingled vehicles can exclude the ETF wrapper entirely.

Legal review of the specific mandate language is required before either conclusion is acted on.

Validator selection in direct and hybrid models

Validator selection is the central diligence exercise in the direct and hybrid models, because the operator’s rigor determines uptime, slashing exposure, and continuity. The criteria institutions apply are the same ones ETF issuers use when contracting staking operators for their funds.

The evaluation checklist covers 6 areas:

1. Uptime and performance SLA. Documented historical uptime, target figures such as 99.9%+, and contractual remedies for misses on demand.
2. Slashing prevention measures. Double-signing controls, distributed key setups, transparent terms, and the operator’s slashing history.
3. Compliance posture. Audits such as SOC 2 Type II, jurisdiction of incorporation, sanctions screening, and counterparty documentation.
4. Key architecture. Confirmation that the design is non-custodial, with withdrawal credentials held by the institution and signing keys isolated from transfer authority.
5. Continuity and infrastructure. Geographic distribution, client diversity, failover design, and incident-response history.
6. Reporting. Reward attribution, on-chain proof, and data feeds compatible with the institution’s accounting stack.

Everstake operates against this checklist as a non-custodial provider with historical experience across 130+ networks, serving both direct institutional delegations and issuer-side integrations as in the case with Canary Capital or Zodia Custody. The overlap with issuer diligence is covered in staking ETFs and validator demand.

Why the future is likely a blend

Most institutions will end up running a hybrid allocation, because the 2 pure structures answer different needs that coexist inside one portfolio. The ETF sleeve answers liquidity and access, and the direct sleeve answers control and scale.

The typical progression runs in 3 stages:

1. Entry. First exposure arrives through ETF shares, since brokerage access requires no new infrastructure.
2. Capability build. The institution adds qualified custody, delegation workflows, and validator diligence, then moves a core position to direct non-custodial staking.
3. Steady state. The ETF sleeve persists for rebalancing and short-horizon needs, while the direct sleeve holds the strategic position with full custody and governance rights.

Allocation between sleeves then tracks operational capacity rather than market view. As reporting, custody integration, and validator relationships mature, the direct share grows, and the ETF share settles into a liquidity-management role.

The structures are complements, not competitors. Issuers themselves rely on the same non-custodial validator layer that direct delegators use, so a hybrid allocator is exposed to one infrastructure stack through 2 wrappers.

FAQ

What is the difference between a staking ETF and direct staking?

A staking ETF gives exposure through exchange-listed fund shares, while direct staking means holding the asset and delegating validation under the institution’s own keys. With Everstake as the validator in the direct model, the institution keeps custody, validator choice, and governance rights, none of which exist inside an ETF share.

What is non-custodial staking?

Non-custodial staking is delegation where the validator never takes possession of the assets. Everstake has operated this model across 130+ networks: the institution or its custodian holds the withdrawal keys, and the validator only performs consensus duties.

Who controls the assets in a staking ETF?

The fund’s contracted custodian controls the assets, not the shareholder. Everstake notes that shareholders hold a pro-rata claim on the fund, with validator selection and key management decided entirely by the issuer.

How do institutions choose a validator?

Institutions evaluate validators on uptime SLA, slashing protection, compliance posture, key architecture, continuity, and reporting. Everstake publishes performance data against these criteria, and the same checklist is applied by ETF issuers when contracting staking operators.

Does Everstake offer custody?

No, Everstake does not offer custody. Everstake is a non-custodial validator, so assets remain with the institution or its qualified custodian at all times, and Everstake holds no transfer authority over delegated positions.

What is a hybrid staking model?

A hybrid staking model combines ETF shares with directly delegated non-custodial positions inside one portfolio. Everstake typically sees the ETF sleeve used for liquidity and rebalancing, while the direct sleeve holds the core position with full custody control.

Does a staking ETF carry slashing risk?

Yes, slashing penalties on the fund’s validators reduce the fund NAV and are shared by all shareholders. 

Disclaimer

This article is provided for informational and educational purposes only. Nothing in it constitutes, or should be construed as, financial, legal, tax, or accounting advice, or as a recommendation, solicitation, or offer to buy or sell any digital asset, exchange-traded product, or other financial instrument.

References to specific products, issuers, or service providers — including staking ETFs, validators, and custodians — are illustrative only and do not constitute an endorsement or recommendation. Reward rates, fees, unbonding timelines, product availability, and regulatory treatment vary by network, jurisdiction, and structure, and may change without notice. Information is accurate as of the date of publication and may become outdated.