ETH Reserves Hit Decade Low as Price Refuses To Follow

Holdings of Ether (ETH), the native token of the Ethereum blockchain, on exchanges stand at 14.6mn as of 18 Jun, the lowest level in nearly a decade. A year ago, that figure exceeded 21mn. The decline, which began in late 2023, accelerated sharply from July 2025 onwards. Roughly 31% of exchange supply has disappeared in 12 months, with no meaningful rebound.

On paper, this is the classic setup for a supply squeeze: fewer tokens available for sale means a mechanical price shock if demand returns. Except that ETH is trading around $1,750 today, nearly two-thirds below its August 2025 record, while the network's economic activity is surging. Stablecoin supply on Ethereum has more than doubled in two years, from $82bn to $172bn – the majority of all stablecoins in circulation.

Less available supply, more activity and a price that still won't follow. The ETH market has narrowed into a tight corridor, and the question this article seeks to answer is whether that corridor is the prelude to a violent repricing, or the symptom of a deeper problem in how the token captures – or fails to capture – the value of the network it is meant to represent.

The corridor

(Source: CryptoQuant & DeFiLIama)

The chart tells two different stories depending on the period. From late 2023 to mid-2025, reserves oscillate between 19.4mn and 21.2mn ETH with no clear trend, and the price follows its own cycle with no direct relationship to reserve levels. Then, starting in July 2025, the drawdown begins and never stops.

Two concurrent phenomena explain this shift. First, spot ETH exchange-traded funds (ETFs), which hold the token directly on investors' behalf, recorded their best month ever: in July 2025 they absorbed a net 1.62mn ETH ($5.43bn), driven by a rally of more than 50% and the convergence of three regulatory catalysts (the CLARITY Act, the GENIUS Act and SEC approval of in-kind redemptions, which let institutions create and redeem fund shares in ETH rather than cash). Almost simultaneously, Bitmine – a listed company holding ETH as a treasury reserve asset – began its ETH treasury strategy, with 625,000 ETH on its balance sheet by the end of July. Of the 6.66mn ETH withdrawn from exchanges between June 2025 and June 2026, Bitmine and the ETFs could explain a significant portion of the move, but with very different temporal dynamics. ETF demand concentrated in July and August 2025 and then ebbed from the autumn onwards. Bitmine's accumulation continued uninterrupted to today. The initial trigger was shared; the persistence of the decline appears to have been driven primarily by corporate treasuries.

Where did the ETH that left exchanges go? Roughly 33% of the total 122.2mn ETH supply is now staked, 12% remains on exchanges and 55% circulates elsewhere (self-custody, DeFi, unstaked treasuries). The corridor is narrow in both directions: little ETH is available for immediate sale, but little ETH circulates freely outside locked positions either.

Concentration risk

The standard reading of the corridor is simple: less available supply means any demand shock transmits faster to price. But a market this tight amplifies shocks in both directions, including those that come from within.

Start with the channel that should, in theory, absorb demand. The spot ETH ETF complex shows $11.2bn in cumulative net flows, but this total masks an extreme concentration. BlackRock's ETHA alone accounts for $11.3bn, more than the entire complex, because Grayscale's ETHE pulls the total down with $5.32bn in cumulative outflows since its conversion. Even the institutional channel, supposed to represent diversified demand, is concentrated on a single issuer.

This is where holder concentration becomes a symmetrical risk. Bitmine holds 4.7% of circulating ETH supply. The parallel with Strategy – the listed company, formerly MicroStrategy, that pioneered holding crypto as a treasury asset – is instructive. Strategy holds about 846,800 Bitcoin (BTC), nearly 4% of BTC supply, a comparable order of magnitude. And the precedent already exists: in late May, Strategy sold 32 BTC to fund a preferred dividend payment, its first Bitcoin sale in four years. The amount is anecdotal, but the mechanism is not. A dividend obligation can force a treasury holder to sell the very asset it is supposed to accumulate indefinitely.

For Bitmine, the exposure is more acute. BMNP, one of its preferred-share series, carries a deferral clause requiring the company to raise capital within 30 days if a payment is missed – a more constraining mechanism than that of Strategy's STRC preferred shares. And the impact of a forced sale would be proportionally heavier on an already compressed market. 5% of Bitmine's holdings represent 281,000 ETH, roughly $492mn at current prices and nearly 2% of total exchange reserves in a single transaction – an amount exceeding the worst single-day outflow ever recorded across the entire spot ETH ETF complex ($465mn). The market has never had to absorb an order of this size through this channel.

The paradox: record activity, falling price

If the supply compression does not explain the price stagnation, perhaps network activity should. Except it doesn't either, and that is the entire paradox.

(Source: Coin Metrics & TradingView)

Stablecoin supply on Ethereum has more than doubled in two years. Daily active addresses reached 1,447,500 on 16 Jan, the highest level in our observation window. The network's economic usage is at a record, and yet the price has lost nearly two-thirds of its value since its all-time high.

The structural explanation for this decoupling lies in the EIP-1559 burn rate – the share of every transaction fee that is permanently destroyed, or "burned," removing ETH from supply. Before the Dencun upgrade (13 Mar 2024), the network burned an average of roughly 3,100 ETH per day in base fees. After Dencun, that average fell to 572, a decline of more than 80%. Over the last 90 days, it is down to just 86 ETH per day, a fall of 97% compared with the pre-Dencun period.

(Source: Coin Metrics)

This decline is more than a technical detail. The EIP-1559 mechanism removes ETH from circulation, but it must offset the continuous issuance paid to validators to secure the network. As long as the burn exceeded that issuance, ETH was net deflationary – the "ultra sound money" narrative, the idea that supply would shrink over time, that underpinned part of its investment thesis. With the burn down to 86 ETH/day, that is structurally no longer the case. ETH has become inflationary again. Available supply on exchanges is contracting, but total supply is no longer shrinking.

Dencun did exactly what it was designed to do. It drastically reduced transaction costs for Layer-2 (L2) rollups – networks such as Arbitrum, Optimism and Base that bundle transactions and settle them on Ethereum's mainnet to cut fees – which now pay near-zero base fees to the mainnet. Activity hasn't declined. It has migrated. Users, volumes and economic usage continue to grow, but on layers where value no longer flows back to the ETH token through the burn mechanism.

What a token captures, and what it doesn't

Reducing the valuation of a Layer 1 to a simple revenue-to-market-cap ratio is an analytical error. Unlike a stock, a token does not necessarily entitle its holder to a share of the revenue generated by the network. The value of ETH is built on more diffuse mechanisms.

The first is economic security. The 39.6mn staked ETH constitutes a guarantee that makes attacking the network prohibitively expensive: the higher the staked value, the higher the cost of an attack. The second is its collateral function. ETH remains the reference asset for borrowing on the decentralized lending protocols Aave, Sky (formerly Maker) and Compound, a function comparable to that of high-quality collateral in traditional finance. The third is its monetary function: ETH is the gas token for the entire ecosystem, including most L2s. The fourth is the network effect. More users and developers reinforce confidence in the system's durability, a mechanism that appears in no income statement but directly determines the market's willingness to pay a premium for the asset that secures it.

Finally, and this is the most structural difference with Bitcoin, ETH's value proposition is not fixed. The transition to proof-of-stake, the introduction of EIP-1559, Dencun and upcoming upgrades (Pectra and beyond) directly alter the mechanics of issuance, burning and value capture. Bitcoin derives part of its premium from the immutability of its rules. ETH makes the opposite bet. Its governance can, in theory, restore a more direct link between activity and value, for instance, by redirecting part of L2 economics back to the L1. But that flexibility is also what makes its value-accrual model unstable, as the previous section illustrated.

What the corridor means for what comes next

Available supply is at its lowest, economic activity has not faded, and the price won't follow. This configuration has two coherent readings.

Scenario one: the compression translates into price. Three plausible catalysts could trigger it – a capital rotation from Bitcoin; a protocol upgrade that restores part of the link between L2 activity and L1 value capture; or an ETF inflow wave post-staking, should the SEC ever authorize direct staking within spot vehicles. In this case, the supply concentration ceases to be a risk and becomes an amplifier: any purchase of meaningful size produces a disproportionate price impact on a market this thin.

Scenario two: the corridor stays narrow, without repricing. Network activity continues to grow, but value remains captured by L2s. Staking yield continues its structural decline. Corporate treasuries find themselves sitting on an asset whose value-accrual model doesn't generate enough to justify the premium they paid. Supply concentration ceases to be a bullish signal and becomes a latent liquidity risk – a motionless market, vulnerable to the slightest forced sale rather than sustained by organic demand.

The signals that will decide: the post-Dencun burn rate is the most direct barometer. A protocol upgrade that durably lifts the burn above its current average (86 ETH/day) would signal a Scenario A in formation. ETF flows, which have turned net negative in recent weeks despite positive cumulative totals, will indicate whether institutional demand is returning or continuing to withdraw. And the share of staked supply, already at 33%, is itself a double-edged signal: a continued increase mechanically removes available supply (favourable to Scenario one), but also signals that less and less ETH circulates freely to absorb a selling shock (favourable to the risk in Scenario two).

The corridor, in itself, does not say which direction it will break. It only says that whichever way it goes, the move will be more violent than it would have been on a deeper market.