Solana Staking Report:
development vectors and analytics from epoch 1000

Contents

Key findings

How staking works on Solana

Solana is a Proof-of-Stake network: a validator earns the right to produce and vote on blocks in proportion to its stake — the amount of SOL delegated to it by holders. The network pays for this out of newly issued supply. To follow this report, five things are worth understanding.

1. Stake lives in dedicated stake accounts. This is not a checkbox on a wallet but a standalone on-chain account with its own address and balance. It carries two authority keys: the stake authority (staker), which manages delegation, and the withdraw authority (withdrawer), which moves funds out. We treat the withdrawer as the "owner": a single person, company, or fund may control many stake accounts but typically far fewer withdraw keys.

2. The lifecycle is measured in epochs (~47.5 hours, slightly under two days — the time the network needs to produce 432,000 slots). Delegated stake activates from the following epoch; deactivation likewise requires a full epoch of waiting. Stake that was deactivated more than 10 epochs ago (through epoch 990 inclusive) but never withdrawn is what this report calls a "zombie" — the money sits in the account and earns nothing.

Figure 1 — the stake account lifecycle, with the actual population of each state as of epoch 1000.

3. Rewards come from inflation and compound automatically. The network issues new SOL (currently ~3.75% per year, with the rate declining 15% annually toward a target of 1.5%) and distributes it in proportion to stake, adjusted for the validator's voting performance. The validator withholds its commission (0–100%); the remainder is added directly to the delegation ("auto-compounding"). MEV tips (Jito) are a separate stream: they are not capitalized into the delegation but arrive on the account balance as separate payments, and the stake owner must claim them manually. Some validators redistribute the MEV tips they receive as additional rewards, but that was outside the scope of this report.

4. Native versus liquid. With native staking you own the stake account yourself: there is no intermediary and no smart-contract risk, and in a number of jurisdictions native staking income is not taxed — but you must monitor the validator yourself (downtime or a commission switched to 100% eats the yield; in §6 we show 1,163 retail accounts stranded in exactly that way). With liquid staking, a pool stakes on your behalf and issues a token receipt (an LST, or Liquid Staking Token: JSOL, jitoSOL, mSOL, DzSOL and others) that can be put to work in DeFi. The pool takes on the job of ensuring that stake is delegated only to validators that meet its criteria and deliver real yield. Because the price of an LST rises each epoch by the staking rewards earned (net of the pool's fee), unstaking returns more SOL than you deposited. Pools use different smart contracts to manage delegations; the Solana Stake Pool Program can be considered the gold standard — it is used by JPool, JitoPool, TheVault and others.

5. Validator commission matters. Commission has a dominant influence on the APY delegators actually receive, since the validator may retain part of the staking rewards and/or the Jito MEV tips. That said, validator performance and downtime also weigh heavily on realized yield.

The network at a glance

Solana is one of the most heavily staked major networks in the world: 68% of total supply is locked into consensus. For comparison, Ethereum has 33% staked, Cardano 57%, and Tron 47%. Why participation runs so high is not purely a matter of conviction in the network: at 4% annual inflation, not staking means guaranteed exposure to the effects of that inflation (more on this in §9).

Stake census: what is happening across 1.5 million accounts

Start with the basic question: what state is every stake account on the network in? A census by lifecycle stage (definitions in §2) shows a healthy, growing system — 84% of accounts are active, inflows and outflows for the epoch are balanced, and the "graveyard" accounts for only 1.3%.

Account size: many people, but the money sits with a few

A stake account is a cheap object to create (the minimum reserve is 0.0023 SOL), which makes the size distribution extraordinarily long-tailed. The two mirrored charts below read as a single sentence: by account count the network is retail; by money it is institutional. 63% of accounts are smaller than 1 SOL and together hold 0.02% of the stake; at the other end, 872 accounts larger than 100,000 SOL hold 61.5% of it. To curb this "spam," SIMD-0490 was introduced in June 2026 (from epoch 989), setting a minimum size of 1 SOL for new native stake accounts; existing accounts below 1 SOL are left untouched.

The age of stake: two waves of money

The activation_epoch field records the epoch from which an account began earning. This is the age of the account, not of the money — splits and merges "reset" a position's age, and closed accounts are invisible, so the chart shows accounts that exist today. With that caveat, the chronology of today's stake reads clearly as two waves. The first is the "year of big money" around epochs 350–399 (fall 2022, the FTX collapse): 49.6M SOL across just 14,000 accounts, averaging ~3,400 SOL apiece — large capital moving in. The second is the mass inflow of 2025–2026: epochs 800–999 hold 224M SOL (including growth in liquid staking, the number of pools, and Digital Asset Treasuries), more than half of today's stake.

Who holds the stake: network composition

To work out whose stake this is, we made two passes. First, every live delegated account (428.3M SOL) was classified through a cascade of eight label-based layers: pools, delegation programs, exchanges (by labels, keys, and validator names), stake on 100%-commission validators, service operators, "factories," and corporate vesting. Second, the large holders in the remainder (≥10,000 SOL, 1,386 wallets) went through manual verification — funding-source tracing and name-by-name checks — after which they were reassigned to exchanges and custodians, corporate vaults, programs, and pools. The resulting picture:

Inside the pool layer: the LST market

Liquid staking pools hold 35.6M SOL of native stake — 8.3% of the network. The market is concentrated but not monopolized: Jito leads with 27.7% of the pool layer, followed by Jupiter, DoubleZero, and Drift. Marinade appears twice — as the classic mSOL (6.6%) and as Marinade Native (2.7M SOL, outside this layer), which stakes on behalf of clients without issuing a token.

Solana's retail stakers: ~532,000–573,000 wallets

Why this needs counting. "How many people actually stake Solana themselves" is a question no explorer answers directly: aggregate metrics such as unique stakers (703,000) lump together individuals, exchanges, pools, and bots. We answer it by subtraction — removing from the network everything that is demonstrably not retail, and then studying what remains.

What remains after every filter is 910,815 accounts holding 169.3M SOL across 574,215 owners, of which 99.96% of accounts are pure self-custody (staker = withdrawer). From there, size decides:

Bottom line: roughly 532,000–573,000 wallets stake independently, holding ~29M SOL — about 7% of active stake. A caveat on scope is essential: this is native, self-directed retail. People earning staking rewards through LST tokens or exchange earn products are not in this number — their shares are buried inside the layers in §5 and are not visible at the wallet level.

A portrait of the retail staker

Retail stake accounts are young. 67% of retail SOL was activated after June 2025. The 2021 retail wave barely survived in money terms (79,000 accounts from that era hold just 0.84M SOL) — whereas the 2025–2026 inflow is substantial both in account count and in value.

Retail stakes through a wallet or interface, and trusts the provider. Mass retail concentrates on validators embedded in popular products: Ledger by Figment holds 120,336 retail accounts (averaging 74 SOL — Ledger Live users), and Everstake 203,911 accounts (averaging 34 SOL). Bitwise is the contrasting case: 7.4M SOL of "retail" spread across only 279 accounts (averaging 26,000 SOL) — that is the tail of the whale tier, not retail.

And retail gets a fair rate. The weighted yield on retail stake is 5.50% — the same as pools and programs earn (§8). 91% of retail SOL earns 5–6%. The exception is painful but small: 1,163 accounts holding 46,900 SOL are delegated to dead or delinquent validators and earn 0% — for years. That is the price of native staking's central drawback: you have to watch the validator yourself (§9).

Concentration: who owns the stake

Definitions. The Gini coefficient measures distributional inequality (0 — everyone holds equally, 1 — one entity holds everything); the Lorenz curve shows what share of stake is held by the smallest X% of owners; the HHI (Herfindahl–Hirschman Index) is the sum of squared shares, a classic measure of market concentration. Why it matters: the distribution of ownership is the distribution of economic power in the network, and it determines resilience to coordinated behavior by large holders.

By owner, Solana staking is extremely unequal even by crypto standards: Gini = 0.9939. The median owner holds 0.81 SOL; the top 10 keys hold 19.1% of stake, the top 100 hold 53.6%, and 554 keys with positions of 100,000 SOL or more control 80.45%. The Lorenz curve lies almost flat against the axes — a visual portrait of a system in which half a million participants divide a few percentage points between them.

One important nuance of interpretation: a key is not a person, and the mismatch runs both ways (a custodian hides thousands of clients behind one key, while a single fund shards across hundreds of keys). Managed stake — 134M SOL under 295 operators using the staker≠withdrawer pattern — is where the boundary between "ownership" and "management" falls in these figures.

The economics of staking: measured yield and where issuance goes

Method. We do not take yield from calculators — we measured it: between the epoch 999 and epoch 1000 snapshots, every stake account's delegation grew by exactly that epoch's reward; each validator's rate was recovered as the mode of the rates across its accounts (robust to splits and merges), and the annual rate is that compounded over 184.4 epochs. Checked against an independent APY reference, the median discrepancy was 0.17 percentage points across 397 validators.

Who earns what is a question of owner class, and the answer is unexpectedly egalitarian: retail earns the same rate as pools and institutional programs.

Where issuance goes. Theoretical gross issuance runs at ~128,300 SOL per epoch. Its single largest interceptor is the set of 100%-commission validators: 69 validators hold 105.0M SOL (24.5% of active stake) and retain ≈6.0M SOL per year (≈32,600 SOL per epoch) — rewards that on an ordinary validator would have gone to delegators, but here stay entirely with the operator. The rest is ordinary validator commission (0–10% among the largest). This is not "leakage" but the cost of infrastructure plus a design choice about ownership — though the scale is worth knowing.

MEV is a separate line item, not part of delegation. Jito tips are not capitalized into stake: over the observation window, 322,000 accounts received 2,444 SOL in claims to their free balances. For a holder this is a second, less visible income stream; it is not included in the rates above — meaning total yield is slightly higher than shown, and the difference (~0.2 pp at the network level) matches the gap in the identity discussed in §9.

Outlook: inflation, dilution, and the evolution of staking

The central formula of this section fits on one line: staking yield ≈ inflation ÷ share staked. All issuance accrues to stakers proportionally; if 68% of supply is staked, each percentage point of inflation converts into ~1.47% of yield.

Solana on the PoS map: an outlier

Compare two axes — what share of supply is staked, and what inflation the network pays for it. Solana lands in the corner: participation on par with the most heavily staked networks, and inflation nearly double the median for major PoS chains (2.1%).

The issuance debate: from SIMD-0228 to SIMD-0550

The Solana community has long argued that the network "overpays for security." In 2025, SIMD-0228 went to a vote — a market-based issuance mechanism tying inflation to the share staked (the higher the share, the lower the issuance). Validators rejected it: the proposal cut directly into their income, and the model introduced uncertainty. In 2026 the question was approached from another angle: SIMD-0550, "Double Disinflation," leaves the formula untouched and instead doubles the pace at which inflation declines — from 15% to 30% per year, preserving the continuity of the curve. The authors estimate savings of ~18.9M SOL in issuance. Status as of this report: approved by both client teams (Anza and Firedancer) on July 6, 2026 — four days before our snapshot — and being prepared for a vote.

*At a 68% staked share and excluding MEV. The system's response is non-linear: falling yield will reduce the staked share, which in turn lifts the rate back somewhat — equilibrium should settle below today's level but above this naive calculation.

What this means for native and liquid

Today native staking dominates for rational reasons: no smart-contract risk, no tax on native staking rewards in a number of jurisdictions, and DeFi's alternative yields are not high enough to justify the added risk. But native carries a hidden cost — the duty of oversight: a validator can go offline (zero rewards for the period), switch commission to 100% for a short window, or quietly die. Our data shows that cost literally: 1,163 retail accounts holding 46,900 SOL have been delegated to dead validators for years, earning nothing (§6).

Once inflation — and with it the nominal rate — drifts toward 2%, the calculus changes. The ability to earn a staking reward and deploy the capital in DeFi (via an LST) turns from a pleasant bonus into the decisive argument: at a 2% rate, every unit of yield counts. Liquid staking today holds just 8.3% of native stake — we expect falling inflation to be its principal tailwind, while the native segment should see rising demand for automated oversight and restaking tools (already visible in products such as Marinade Native, with 2.7M SOL in managed native staking without a token).

Dormant capital: zombies, forgotten money, and permanent locks

Solana's staking system has accumulated a substantial layer of money that has fallen out of circulation — through forgetfulness, by design, or by the owner's deliberate choice. We separate it into four categories.

The largest abandoned stake deserves a mention of its own: an account whose address begins with "8k4MoNEY" deactivated 563,253 SOL on April 26, 2023 — and has not made a single transaction since. More than half a million SOL has sat one click away from withdrawal for three years. The irony of the address is left to the reader.

Locks and vesting. Solana's native lockup lets an owner freeze withdrawals until a date or epoch, either with an optional custodian (who can release early) or without one (in which case no one can). 22.1M SOL is under active locks: institutional vesting ladders (the largest custodian releases tranches of 1M down to 500K SOL through the end of 2027), the legacy of the TDS22 program, and corporate transactions structured as "you own it but cannot dispose of it" — locked and staked so that vesting does not forgo yield.

Curiosities and records

Conclusion

Solana staking as of epoch 1000 is a mature two-story system. The ground floor is half a million retail participants with a median position under 10 SOL, staking through wallet interfaces and receiving a fair market rate. The upper floor is a couple of hundred institutional structures — exchanges, custodians, owner-operated validators, funds — that control four-fifths of the capital and have engineered sophisticated ownership schemes around themselves: 100%-commission validators as a liquid income tap, sharding across hundreds of keys, cold-storage factories, and vesting ladders.

Three things stand out as the agenda for the epochs ahead.

The inflation transition. SIMD-0550 could halve nominal yield within two to three years; liquid staking and tools that extract more from stake (MEV, restaking) stand to gain, while passive, unsupervised native positions stand to lose.

The cost of validating. Lowering the cost of running a validator could increase the number of validators, smooth the distribution of stake, and thereby reduce risk across the network.

"Corporate staking." Shifts in the behavior of SOL holders who delegate to 100% nodes, together with a possible migration of corporations from native to liquid staking, would reshape the structure of staking and reduce concentration within the wealthiest group of delegators.

Methodology and data

Data. Two complete on-chain snapshots of every Stake-program account: epoch 999 (~64% through) and epoch 1000 (15.0% through; slots 432,064,863–432,066,781; July 10, 2026; 1,497,112 accounts) — plus a synchronous snapshot of validator commissions and liveness, a validator directory (names covering ~68% of stake, geography, commissions), an epoch-to-date table validated against historical transactions, stake-type labels from an independent database (33 labels), exchange address labels (237), and a public PoS network benchmark.

Classification — a cascade of eight layers (§5); each account receives exactly one class, with full line-by-line classification retained. Yield — measured as the delta in delegations between snapshots (the epoch 999 reward), with each validator's rate taken as the mode of the rates across its accounts.

Owner — the unique withdraw authority.

Limitations. This is a snapshot, not a history: closed accounts are invisible, and splits and merges reset account age. Retail inside LSTs and on exchanges is not visible at the wallet level — our 532,000–573,000 is therefore a lower bound on the number of "people earning from staking." Categories for large holders are annotations pending manual verification.

JPool Research · in collaboration with the SolBrothers validator · epoch 1000 snapshot (July 10, 2026)