How much SOL is locked in Solana rent? Network-wide numbers

The short answer

An estimated 3.4 million SOL is locked in rent across Solana's roughly 1.14 billion accounts, about 0.53% of the 636.9 million SOL total supply. Token accounts are close to half of it: the SPL Token program's 683 million accounts hold about 1.39 million SOL at the old rate, and Token-2022 accounts roughly 0.2 million more. These are estimates built from Solana Foundation state data from July 2026, not an on-chain total, and the method is below so you can check every step.

I build SOLTidy, a set of tools for reclaiming rent, and I kept running into the same gap: plenty of per-account figures, no sourced network-wide ones. This post is my attempt to fix that. Figures I measured myself are marked as such, with the date; everything else links to its source.

What counts as "SOL locked in rent"?

SOL locked in rent is the sum of every account's rent-exempt minimum: the balance each account must keep for as long as it exists. Rent is a refundable deposit, not a fee. The minimum is (128 + data bytes) x lamports_per_byte, where 128 bytes is a fixed per-account overhead, and the Solana Foundation's Reduced Rent page calls it "a fully refundable bond that pays for the storage an account occupies on every validator". The mechanics are in what Solana rent is.

Two things follow from that definition:

• It is a floor, not a balance. A wallet holding 10 SOL counts here only for its 0-byte minimum. Anything above the minimum is ordinary spendable SOL.
• It follows the rate when each account was funded. Accounts keep the lamports they were created with. Almost all of today's accounts were created at 6,960 lamports per byte, so they hold more than the current minimum.

What is the rent rate right now?

As of November 4, 2026, the rate is 5,080 lamports per byte, the second step of SIMD-0437. SIMD-0437, written by Igor Durovic of Anza, lowers lamports_per_byte from 6,960 to 696 in five separately feature-gated steps: 6,333, 5,080, 2,575, 1,322 and 696. It superseded SIMD-0436, which proposed a single cut to 3,480.

I checked the activation slots on mainnet by reading the feature accounts listed on the Reduced Rent page:

Step	Rate (lamports/byte)	Feature gate	Mainnet status, read November 4, 2026
Original	6,960	none	Until epoch 1028
1	6,333	4a6f7o7iTcA8hRDCrPLkSatnt5Ykxiu36wo5p1Tt12wC	Active at slot 444,096,000, epoch 1028, September 3, 2026
2	5,080	61BtM7BkDEE8Yq5fskEVAQT9mYA8qCejJWoLe5apqg81	Active at slot 446,256,000, epoch 1033, September 11, 2026
3	2,575	Ftxb3ZKq7aNqgxDBbP7EonvR2RszZk9ctjdsTX38kQaz	Not activated
4	1,322	GsUBNYNDPdMLHPD37TToHzrzcNcjpC9w5n1EcJk5iTaM	Not activated
5	696	mZdnRh9T2EbDNvqKjkCR3bvo5c816tJaojtE9Xs7iuY	Not activated

The Rent sysvar (SysvarRent111111111111111111111111111111111) currently reads lamports_per_byte 5,080 with an exemption threshold of 1.0, the format introduced by SIMD-0194, which folded the old 3,480 lamports per byte-year times a 2-year threshold into a single 6,960 constant. The Foundation expects steps 3 to 5 with Agave 4.4, and Solana Compass reports that no activation date had been published for them as of late October 2026. A sixth fallback gate can put the rate back to 6,960 if state grows too fast.

What that means per account, from getMinimumBalanceForRentExemption on November 4, 2026:

Account	Data size	Minimum today	Minimum before September 2026
Wallet (no data)	0 bytes	650,240 lamports (0.00065024 SOL)	890,880 lamports
SPL token mint	82 bytes	1,066,800 lamports (0.0010668 SOL)	1,461,600 lamports
SPL token account	165 bytes	1,488,440 lamports (0.00148844 SOL)	2,039,280 lamports (0.00203928 SOL)
Token-2022 account, immutable owner	170 bytes	1,513,840 lamports (0.00151384 SOL)	2,074,080 lamports
Stake account	200 bytes	1,666,240 lamports (0.00166624 SOL)	2,282,880 lamports

The full list is in the rent-exempt minimum by account type, and the Solana rent calculator reads the live rate for any size.

How many accounts are on Solana?

There are roughly 1.14 billion accounts on Solana, based on the most detailed public snapshot I could find. In July 2026 the Solana Foundation published Rent Reduction on Solana: A Data-Backed Analysis by data researcher Umberto Natale. Its treemap of account data size at epoch 997, which began on July 4, 2026, labels the SPL Token program with 683 million accounts and 60.14% of all accounts. Dividing one by the other gives about 1.136 billion accounts in total.

Here are the key figures in one place:

Figure	Value	Source and date
Total accounts	about 1.14 billion (derived)	SF treemap, epoch 997, July 2026
Live account data (accounts_data_len)	335.7 GB	SF analysis, early July 2026
AccountsDB storage files	495 GB	SF analysis, early July 2026
SPL Token program accounts	683 million (60.14% of accounts), 107 GB (31.71% of data)	SF treemap, epoch 997
Token-2022 accounts	92.0 million (8.10%), 17.9 GB (5.32%)	SF treemap, epoch 997
OpenBook program accounts	4.46 million (0.39%), 95.1 GB (28.32% of data)	SF treemap, epoch 997
Metaplex Token Metadata accounts	82.8 million (7.29%), 33.2 GB (9.88%)	SF treemap, epoch 997
Net new accounts	about 1 million per day, about 0.3 GB per day	SF analysis, June to July 2026
Accounts in May 2024	over 500 million (as reported)	Anatoly Yakovenko essay, May 31, 2024
Total SOL supply	636,944,314.8 SOL	My getSupply call, November 4, 2026
Circulating SOL supply	590,206,701.8 SOL	My getSupply call, November 4, 2026

Two cautions on the treemap. It shows exactly 8.10% and 92.0 million for both Token-2022 and the BPF Loader, which looks like a labeling issue, so I treat the Token-2022 count as less reliable than the SPL Token one. And the chart's byte figures appear to be estimated storage bytes rather than raw data length: 683 million accounts x 165 bytes is 112.7 GB, not 107 GB. I use account counts rather than bytes wherever I can.

How much SOL is locked in rent across the network?

About 3.4 million SOL, and two independent routes land on that figure.

Route 1: from state size. Every account contributes 128 bytes of overhead plus its data. With about 1.136 billion accounts and 336 GB of data, nearly all funded at 6,960 lamports per byte:

1. Overhead: 1.136 billion x 128 bytes = 145.4 GB.
2. Total rent-bearing bytes: 145.4 GB + 336 GB = 481.4 GB.
3. Rent: 481.4 billion bytes x 6,960 lamports = 3.35 million SOL.
4. If the 336 figure is GiB rather than GB (360.8 billion bytes), the result is 3.52 million SOL.

Route 2: from published reclaim figures. Coverage of the rollout reported that step 2 made about 612,000 SOL reclaimable and that all five steps would free about 3.06 million SOL, attributed to the Foundation's upgrade page. Step 2 cuts the rate by 1,253 of the original 6,960 lamports per byte, which is 18.0%, and the full schedule cuts it by 90%. Working backward, 612,000 / 0.18 and 3.06 million / 0.9 both equal 3.4 million SOL. A separate Helius estimate, attributed to Ghost, of "over 3 million SOL" refundable at the final rate is consistent with the same total.

So my working figure is 3.4 million SOL, with a plausible range of 3.3 to 3.5 million, as of mid-2026. That is about 0.53% of total supply. Three things move it over time: about 1 million net new accounts per day add to it, accounts opened since September 3, 2026 are funded at the lower rates, and owners withdrawing excess rent or closing accounts take SOL out of it.

How much of that is in token accounts?

Token accounts hold an estimated 1.6 million SOL, about 47% of all rent, even though they are only about a third of the data. The reason is the 128-byte overhead: a 165-byte token account pays rent on 293 bytes, so small, numerous accounts carry a lot of overhead per byte stored.

Group	Accounts (July 2026)	Rent per account at 6,960	Estimated rent held
SPL Token program	683 million	0.00203928 SOL	about 1.39 million SOL
Token-2022	92.0 million (see caveat)	about 0.00225 SOL (195-byte average)	about 0.21 million SOL
All token accounts	775 million		about 1.6 million SOL

These assume every account was funded at the old rate, which was true for almost all of them in July 2026. The Token-2022 average size comes from the treemap's 17.9 GB across 92.0 million accounts. The SPL Token program count includes mint accounts as well as token accounts, but mints are one per token and only 82 bytes, so token accounts dominate it.

Where those token accounts come from is an open question with one useful clue. The Foundation analysis estimates, with a suffix-based heuristic it calls indicative, that roughly 30% of SPL Token space is associated with pump.fun-style tokens. On July 11, 2026, classic SPL token accounts were 8.60 of the 10.23 GiB requested by all new accounts that day, about 84% of gross demand. The everyday version of that story, one account per new token you touch, is told in where your token accounts come from.

How much SOL is in empty token accounts?

No credible network-wide figure exists, and I would be wary of any site quoting one. Counting empty token accounts means reading every token account's balance, which requires a full getProgramAccounts scan over the token program; public RPC endpoints refuse it, and I did not find a published Dune or explorer count either.

What can be done is a sensitivity table. Each row assumes a share of the 683 million SPL token accounts is empty and that they were funded at the old rate:

Share empty (assumption)	Empty SPL token accounts	SOL held by them
10%	68.3 million	about 139,000 SOL
25%	170.8 million	about 348,000 SOL
50%	341.5 million	about 696,000 SOL
75%	512.3 million	about 1.04 million SOL

The rule of thumb is about 13,900 SOL for every 1% of SPL token accounts that is empty. For a sense of how this plays out wallet by wallet, SOLTidy's own data on how much SOL is stuck in wallets that used a cleanup tool has the median and spread across real wallets; I won't repeat it here.

There is a second pool that needs no assumption about emptiness. Every SPL token account opened before September 3, 2026 holds 0.00055084 SOL more than it needs today. If all 683 million were funded at 6,960, that excess comes to about 376,000 SOL, recoverable without closing anything through the token program's WithdrawExcessLamports instruction, which the Foundation describes in its guide to reclaiming excess SOL. Nothing moves automatically; the owner has to sign.

Why does locked rent matter for the network?

Locked rent matters because it is the price of state, and state is the one cost every validator carries forever. Every live account has to be stored, indexed and included in snapshots by every validator. The Foundation measured AccountsDB storage files at about 495 GB in July 2026, against Anza's recommended 1 TB of capacity, and fitted recent growth at about 0.23 GB per day.

Three points from that analysis are worth knowing:

• Most new state is short-lived. In the cohort studied, 75.5% of account-creation events closed within the same transaction, and only 5.7% of newly allocated bytes survived past 1 million slots, about 4.7 days. Arbitrage bots creating and closing token accounts drive much of that churn.
• Persistent state is concentrated. OpenBook holds about 28% of live data in under 0.4% of accounts, because order-book markets keep large queues on chain. Token accounts are the largest category by count.
• The deposit is the brake on bloat. The analysis estimates that even after the full 90% cut, filling the current storage headroom would need about $17 million of locked capital. Rent is the reason that number is not zero.

This is also why closing accounts is more than a personal refund. An empty token account is pure state: 165 bytes on every validator for an account that may never be used again. Closing it with the SPL Token program's CloseAccount instruction deletes those bytes from the live state and returns the deposit to the owner. That's the design working as intended, with the deposit acting as a bond on storage you hand back when you are done.

How much can one wallet reclaim?

One wallet can reclaim up to 0.00203928 SOL per empty SPL token account opened before September 3, 2026, 0.00148844 SOL per empty account opened since, and 0.00055084 SOL per older account it keeps. Here's how to get your own number rather than an average:

1. Run a read-only scan with the Solana wallet cleanup or the Token Account Closer. Scanning signs nothing and shows each account's SOL.
2. Close the empty accounts you no longer need. The closer handles SPL Token and Token-2022 accounts, also withdraws excess rent from older accounts you keep, and charges 5% of the SOL it recovers, taken inside the same transaction.
3. If you have only a handful of accounts, doing it yourself is perfectly reasonable: the spl-token CLI costs only network fees, and so does your wallet's own close option.
4. For accounts you hold tokens in, withdraw the excess instead of closing, as explained in how to get SOL back without closing token accounts.

The per-account arithmetic, including fees and the Token-2022 sizes, is in how much SOL you can recover from token accounts. If you traded on pump.fun, its extra accounts hold rent too.

What would change these numbers?

The figures above are a mid-2026 picture and will drift. Here's what to watch:

• Steps 3 to 5 of SIMD-0437. At 696 lamports per byte, a new SPL token account would need 203,928 lamports (0.000203928 SOL), and roughly 90% of the 3.4 million SOL would become withdrawable surplus. I will update this post if they activate.
• A rent increase. SIMD-0392 makes the rate safe to raise again for new accounts without breaking existing ones, should state grow too fast.
• A fresh snapshot. A newer account-size breakdown, or a public count of empty token accounts, would replace my estimates with measurements. If you have one, I would like to cite it.

Bottom line

Solana's roughly 1.14 billion accounts hold an estimated 3.4 million SOL in rent, about 0.53% of supply, and token accounts hold about 1.6 million SOL of it; two independent methods, one from Solana Foundation state data and one from the published SIMD-0437 reclaim figures, agree on that total. The rate is now 5,080 lamports per byte after steps taken on September 3 and 11, 2026, but existing accounts keep their old deposits until their owners act. Nobody knows how many token accounts are empty, though every 1% of them is worth about 13,900 SOL. Your own share is easy to find: scan with the Token Account Closer, close what is empty, and withdraw the excess from what you keep.