ETH Supply in 2026: Inflationary on Paper, Vanishing in Practice

Ethereum spent the third week of August 2026 doing something it had not done in months: moving. ETH jumped close to 20% in a single session on 19 and 20 August, briefly touching the low $2,300s, after a cascade of liquidations wiped out leveraged shorts in one of the largest squeezes of the year, according to Invezz. By 21 August the token traded near $2,402, with a market capitalization around $290 billion and a price still roughly 51% below its $4,946 all-time high, per CoinGecko.

Price is the headline. Supply is the story. Underneath the tape, Ethereum is running two supply narratives at once, and they point in opposite directions. On paper, the network is inflationary again: it now mints slightly more ETH than it destroys. In the market, the amount of ETH you can actually buy is shrinking, drained into validators, corporate treasuries and exchange-traded funds faster than sellers can replace it.

This is the supply paradox that defines Ethereum in 2026. The famous ultrasound money thesis, the idea that ETH would grow scarcer with every block, is broken at the protocol level. Yet a genuine supply squeeze is building at the market level, and a contested proposal called EIP-8361 could hard-wire scarcity back into the code. Here is how both ledgers actually read, and what they imply for price into 2027.

The stat that killed the ultrasound money meme

Start with the number that reset the narrative. Over the 90 days to 27 July 2026, Ethereum validators earned roughly 254,000 ETH in fresh issuance, while the EIP-1559 base-fee burn removed only about 5,200 ETH, per an analysis of the network trend by Crypto Daily. That is not a rounding error; it is a rout. Issuance outran the burn by nearly fifty to one.

Annualize it and Ethereum’s supply is expanding at roughly 0.23% per year, mildly inflationary, according to data aggregated by crypto.news. The canonical dashboard, ultrasound.money, still runs, but the meme it was built to celebrate has quietly inverted. It was designed to show a green counter of ETH vanishing block by block; for most of 2026 that counter has barely moved.

The cumulative math tells the same story. Since the Merge in September 2022, total supply has grown by roughly 950,000 ETH. The two most-cited trackers disagree on the absolute number, and the gap is worth understanding: ultrasound.money puts total supply near 122 million ETH because it sums execution-layer and consensus-layer balances, while CoinGecko lists circulating supply closer to 120.7 million. Neither is wrong; they measure different things. What both agree on is direction, and the direction is up.

How ETH supply actually works

Ethereum has no fixed cap. There is no 21 million ceiling and no halving calendar, the two features that make Bitcoin’s scarcity a matter of arithmetic. Instead, ETH supply is governed by three levers pulling against each other, and the balance between them changes with every upgrade.

The first lever is issuance. The protocol pays newly minted ETH to validators for securing the chain, currently on the order of 2,800 ETH per day in gross terms. The Merge cut that rate by roughly 88% when Ethereum abandoned proof-of-work in September 2022, embodying a design principle the core developers call minimum viable issuance: pay validators only as much as security requires, and not a wei more.

The second lever is the burn. Since EIP-1559 went live in August 2021, every transaction destroys its base fee, permanently removing that ETH from supply. More than 4.6 million ETH has been burned this way, worth well over $9 billion, per the running total on ultrasound.money. When the burn exceeds issuance, supply shrinks. That was the ultrasound money dream, and for stretches of 2022 and 2023 it was real.

The third lever is not destruction but immobilization: staking. ETH locked in validators is not burned, but it is off the market, and as we will see, that sink now matters more to price than the burn ever did. The base fee that feeds the burn, incidentally, is the same fee that debates over account abstraction and gasless transactions are trying to hide from ordinary users, which is one reason mainnet burn has become so hard to predict.

The contrast with Bitcoin is instructive, because it is deliberate. Bitcoin fixes its supply at 21 million and lets fees eventually replace block rewards; Ethereum refuses to pin a number, arguing that a hard cap would either overpay for security in quiet years or underpay for it in busy ones. The tradeoff is that ETH holders cannot point to a schedule and know their dilution in advance. Supply here is a policy, not a promise, and policy can change, which is precisely what the current fight is about.

Why Dencun and Fusaka rewired the burn math

So why did the burn collapse? Blame Ethereum’s own success. The Dencun upgrade in March 2024 introduced blob transactions (EIP-4844), giving layer-2 rollups a cheap dedicated lane for their data. Almost overnight, the activity that used to congest mainnet and feed the burn migrated to L2s paying a fraction of a cent. Daily burn fell from thousands of ETH to as little as 50 to 70, per crypto.news.

That exposed a structural tension the ultrasound thesis had glossed over: a cheap, scaled Ethereum burns less than a congested, expensive one. The network’s success as global settlement infrastructure works directly against its scarcity as an asset. The more useful Ethereum becomes, the less ETH it destroys, which is an uncomfortable thing to design around.

Fusaka, which reached mainnet on 3 December 2025, tried to have it both ways. Its headline feature, PeerDAS (EIP-7594), cut the data load on home validators by roughly 87.5% so the network could scale blob capacity further. But Fusaka also shipped EIP-7918, a blob base-fee floor that forces rollups to pay a minimum price tied to the execution base fee, precisely so that scaling does not zero out the burn. Fidelity modeled that the floor would have added around $78.6 million in burn across 93% of days since 2024, per a supply review published by Bitget.

It is a partial fix. For Ethereum to turn deflationary again on fees alone, average mainnet gas needs to sustain something above roughly 16 gwei; for much of 2026 it has sat below one. The next scheduled upgrades, Glamsterdam (targeted for late 2026) and Hegotá (2027), are focused on proposer-builder separation and censorship resistance, not monetary policy, per the Ethereum roadmap. Which is exactly why a monetary-policy proposal arrived from outside the standard fork schedule.

A short history of ETH’s supply

Five upgrades explain almost everything about where ETH supply sits today. Each one moved a lever, and together they turned a steadily inflating proof-of-work coin into the strange hybrid of 2026: inflationary at the protocol, scarce in the market.

EIP-8361: the plan to make ETH scarce again

On 4 August 2026, a group of Ethereum Foundation-adjacent researchers, including Justin Drake, published a draft that would rewrite ETH’s monetary policy. Submitted through Jérôme de Tychey, president of Ethereum France, EIP-8361 is called the tapered issuance burn, and its logic is elegant, and to some, alarming, per CoinDesk.

The mechanism: as the share of staked ETH rises, the protocol burns a growing fraction of each validator’s rewards. The deduction starts at 0% and climbs to 100% as staking approaches roughly 60.25 million ETH, about half of today’s supply, worth some $112 billion at the time of the proposal. At that point, consensus-layer issuance effectively drops to zero. The change would phase in over roughly 18 months rather than flip overnight.

The intent is to restore the ultrasound money property without touching EIP-1559, by attacking issuance instead of leaning on the burn. It is minimum viable issuance taken to its logical endpoint: if security is already over-provisioned by too much staking, stop paying for it. de Tychey has warned that on the current trajectory, more than 70 million ETH could be staked by January 2028, which the authors argue is more security than the network needs and more dilution than holders should quietly accept.

The deeper argument behind the proposal is about the security budget. Ethereum pays for its consensus in freshly minted ETH, so every validator added past the point of adequate security is, in the authors’ view, dilution without benefit: the network buys safety it already had. Drake and the minimum-viable-issuance camp would rather cap that spend and let the market, not the protocol, decide how much ETH chases yield. Critics counter that security is never truly finished, and that throttling rewards could push stakers to unwind at the worst possible moment.

For now, EIP-8361 is only a draft. It did not receive proposed-for-inclusion status for the Hegotá upgrade, and it carries no schedule. (Owing to a numbering clash, some coverage refers to it as EIP-8363.) But it has already done something no fork has: it turned Ethereum’s monetary policy into a public brawl.

The builders’ revolt

The loudest opposition came from the people who build on Ethereum’s yield. Stani Kulechov, founder of the lending protocol Aave, was blunt: “Unfortunately this proposal doesn’t achieve the outcome it tries to achieve and is actually hurtful for Ethereum,” he wrote, adding that “Ethereum should not be punished for its growth,” per The Crypto Times.

Kulechov’s concern is mechanical. If staking yield falls toward zero, the base rate that anchors DeFi lending disappears; borrowing ETH becomes pointless except to short it, and money markets that route billions in on-chain credit lose their reference yield. Zero issuance, in his telling, does not make ETH sound; it makes ETH’s financial layer inert.

Mike Silagadze, chief executive of the liquid staking and restaking firm ether.fi, was harsher still: “This is so disappointing on every level,” he posted, asking “Every builder on Ethereum opposes this. Why is this a focus?” His firm has a direct stake, since a proposal that halts new staking would freeze the very flows ether.fi intermediates.

de Tychey pushed back on the process, not the merits: “The window is closing,” he argued, and “Being proposed for inclusion is what opens the floor for feedback, not what closes it.” The subtext is a genuine philosophical split inside Ethereum: monetary purists who want ETH as hard as possible, against a growth-and-DeFi camp that treats yield as the network’s economic engine. The supply question, it turns out, is really a governance question.

The other ledger: who is locking ETH away

If the protocol is minting more ETH than it burns, why is the market acting starved? Because the second ledger, the one that tracks where ETH actually sits, is draining fast. Start with staking. As of early August 2026, about 41.4 million ETH was locked in validators, a record 33.98% of supply, per Coinpedia. Nearly one ETH in three now earns consensus rewards instead of trading.

That crowd has a cost. Because issuance is a roughly fixed pool split among all validators, more stakers means thinner slices. The 7-day staking APR has fallen to about 2.66%, a three-year low, down from a 5.06% peak in June 2023. Yield compression is exactly what EIP-8361’s authors point to as evidence the network is over-secured, and exactly what its critics say will get worse if issuance is burned away.

Where you stake increasingly shapes what you earn. Lido, once the dominant liquid-staking pool at roughly a third of all staked ETH, has slipped to around a quarter as institutions built their own validator operations, a shift we unpack in our comparison of Lido, Rocket Pool and Frax. The spread between the best and worst staking venues has widened enough that it is now worth shopping around, as our staking yield test across Coinbase, Binance, Kraken and OKX found.

Crucially, staked ETH is sticky. Exiting a validator means joining a withdrawal queue, and as ether.fi’s Silagadze has argued, people who stake ETH tend not to sell it. The exit queue sat near zero in August, while the entry side stayed busy earlier in the year, evidence that the flow is still mostly one way: in.

Staking is also no longer the end of the line for locked ETH. A restaking layer, led by protocols such as EigenLayer and by liquid-staking-and-restaking firms like ether.fi, lets the same staked ETH be pledged again to secure additional services, deepening the lockup and giving holders another reason to keep coins bonded rather than liquid. Every extra layer of yield stacked on top of base staking is one more incentive not to sell, which is a large part of why the float keeps thinning even in months when the burn does almost nothing.

Corporate treasuries and the Alchemy of 5%

The newest sink is the loudest. Over the past year, a wave of publicly traded companies has adopted ETH as a primary treasury reserve, the Ethereum answer to the Bitcoin treasury trade. Collectively, roughly 32 public companies now hold about 7.8 million ETH, more than 6% of supply, per CoinGecko’s treasury tracker, though the trackers lag the biggest holders’ own disclosures.

The whale is BitMine Immersion Technologies (BMNR), chaired by Fundstrat’s Tom Lee. As of mid-August, BitMine disclosed holdings of about 5.82 million ETH, roughly 4.8% of all ETH in existence, with more than 5 million of it staked, per Crowdfund Insider. Lee has framed the ambition as the Alchemy of 5%, an explicit goal of owning 5% of the entire Ethereum supply, with projected annual staking income of roughly $287 million once the stack is fully staked.

Behind BitMine sit SharpLink Gaming (SBET) with about 869,000 ETH, The Ether Machine with roughly 497,000, and Bit Digital with about 158,000. What makes these treasuries a double sink is that they do not just hold ETH off the market; they stake most of it, removing the same coins from circulation twice over, once as a treasury asset and again as validator collateral.

The playbook is borrowed from the Bitcoin treasury trade that turned a software company into a leveraged proxy for BTC, but ETH adds a twist the Bitcoin version lacks: yield. Because staked ETH earns a return, a treasury company can present itself not just as a bet on price but as a cash-flowing operation, which is how BitMine frames its projected staking income. That yield is the sales pitch for the premium, and it is why a proposal to zero out staking rewards lands as a threat to a business model, not just to a number on a dashboard.

That reflexivity cuts both ways, and it is the clearest risk in the whole supply story. Treasury companies trade at a premium to the value of their ETH (their so-called mNAV), and that premium is what lets them issue stock to buy more. If it collapses, the flywheel can reverse: the buyers of last resort become forced sellers. BitMine’s authorization of a $1 billion share buyback in August, defending its own stock, is a sign that management is watching the same risk.

Where the float went

Add the sinks together and the picture sharpens. The table below is a set of overlapping lenses, not a clean sum: treasury ETH is often also staked, and ETF ETH is custodied separately, so the buckets double-count on purpose. The point is not the total; it is that the one bucket representing freely tradable ETH, coins sitting on exchanges ready to sell, has shrunk to a multi-year low.

*Derived from roughly $10.5 billion in ETF net assets at about $2,400 per ETH, not a directly reported coin count. Exchange reserves have been carved down from around 35 million ETH at the 2021 peak into the mid-teens of millions, a level not seen in years, per on-chain data from CryptoQuant highlighted by The Crypto Times; different trackers put the figure lower still, depending on which wallets they count. Against a shrinking float, ETF net assets of about $10.5 billion and cumulative inflows above $11 billion, per CoinMarketCap’s ETF dashboard, represent a persistent, price-insensitive bid.

What the SEC did to make staked supply investable

None of the ETF or treasury demand would be legal at scale without a regulatory decision that landed earlier in the year. On 17 March 2026, the SEC and the CFTC issued a joint interpretive release clarifying that staking rewards on 16 named digital commodities, Ethereum among them, are not securities, and that protocol staking, whether solo, custodial or liquid, does not trigger Securities Act registration, per an analysis by Sullivan & Cromwell.

That reversed the Gensler-era SEC position, which in 2024 had pushed ETF issuers to strip staking out of their filings. With the legal barrier gone, staking ETFs went live: Grayscale’s Ethereum product (ETHE) began distributing staking rewards, the first US exchange-traded product to do so, and BlackRock launched its iShares Staked Ethereum Trust (ETHB) in March 2026. Gross yields run about 3.1% to 3.3%, or roughly 1.9% to 2.6% net of fees, per Everstake.

The supply consequence is subtle but powerful. When an investor buys a staking ETF, the fund buys ETH and stakes it, so ETF inflows now feed the validator sink directly. Regulatory clarity did not just legitimize an asset; it wired Wall Street’s demand into Ethereum’s supply lockup. US holders should note that those staking rewards are generally taxable as ordinary income when received, a wrinkle covered in our 1099-DA-era tax guide.

Clarity has limits. The joint interpretation is guidance, not legislation, and a future SEC could revisit it; durable rules would require Congress to pass market-structure law, which remains unfinished. For now, though, the practical effect is settled enough that issuers have built products on it, and those products keep buying and staking ETH. Regulation, in other words, has become a supply variable, one more input that decides how much ETH sits idle inside a fund versus trading on an exchange.

Reading the August tape

This is the backdrop against which the 19 and 20 August squeeze detonated. The trigger was macro, not crypto-native: on 19 August the US Treasury said it would at least double the size of its liquidity-support buybacks for longer-dated government debt, raising the maximum per operation from $2 billion to at least $4 billion between early September and early November. Long-dated yields fell, the 30-year slipping from 5.26% to as low as 5.18%, and risk assets caught a bid, per Coinotag.

In a normal market, that might have nudged ETH a few percent. Against a thin float, it produced a 20% candle. More than $3 billion in leveraged positions were liquidated in 24 hours, roughly 92% of them shorts, one of the largest squeezes of 2026, per KuCoin. When freely tradable supply is scarce, marginal buying moves price violently, and spot ETFs chose that same week to post one of their strongest single-day inflows of the year.

That is the bullish reading of a shrinking float: it turns modest demand into outsized rallies. The bearish reading is the same mechanism in reverse. The corporate-treasury flywheel that absorbed so much ETH depends on those share premiums holding; if they crack, the most aggressive buyers become sellers into a market with little depth to catch them. A thin float amplifies in both directions, and 2026 has shown mostly the up direction so far.

The bulls make a longer-term case that rests on exactly this supply setup. Tom Lee has argued that structural demand for ETH as the settlement layer for tokenized assets and on-chain finance justifies a valuation well above current levels, and that a shrinking float turns that demand into leverage on the price. Skeptics note that the same reflexivity powering the rally is untested on the way down, and that a market this thin has never had to absorb a wave of treasury selling. Both can be true at once; the float simply makes the swings larger in either direction.

Scenarios for ETH supply into 2027

Where does the supply story go from here? Three scenarios, framed around the two levers that matter most: protocol issuance policy and the size of the tradable float.

The signposts to watch are concrete. A staking ratio climbing toward 50% would put EIP-8361’s trigger in range and force the inclusion debate to a head. Fork decisions at Glamsterdam and Hegotá will show whether monetary policy re-enters the roadmap at all. Sustained mainnet fees above 16 gwei would revive the burn on their own, no proposal required. And treasury-company mNAV premiums are the single cleanest gauge of whether the corporate bid is still expanding or about to reverse.

What it means for holders

Strip away the noise and a few conclusions hold up:

• Ultrasound money broke, but scarcity did not disappear; it migrated from the burn to the float. Judge ETH by exchange reserves and the staking ratio, not the burn counter alone.
• Yield is compressing. At a sub-3% base rate, where and how you stake matters more than whether you stake, and staking ETFs trade convenience for a fee-shaved return.
• EIP-8361 is the biggest single swing factor in ETH’s monetary future, and the least predictable. It could double as a scarcity catalyst or a DeFi shock, and it is currently stuck in governance.
• The ETF and treasury bid is real but reflexive. It drained the float on the way up; it could refill it on the way down.

For most of its life, Ethereum’s supply debate was a story about the burn: a green counter ticking up, a meme about sound money. In 2026 the counter has stalled, but the more important number moved somewhere else, into the shrinking pile of ETH that anyone can actually buy. The protocol is minting more ETH than it destroys; the market has rarely had less of it to trade. Which of those two facts wins is, increasingly, what ETH’s price is about.

Frequently Asked Questions

Is Ethereum inflationary or deflationary in 2026?

Mildly inflationary. Net issuance runs around +0.23% per year because the EIP-1559 burn collapsed after the Dencun upgrade moved activity onto layer-2 rollups. Ethereum was net deflationary for stretches of 2022 and 2023, but since 2024 it has minted slightly more ETH than it destroys.

How much ETH is staked, and what is the yield?

About 41.4 million ETH, a record 33.98% of supply, was staked as of early August 2026. The 7-day staking APR had fallen to roughly 2.66%, a three-year low, because a fixed issuance pool is now split among more validators than ever.

What is EIP-8361?

EIP-8361, the tapered issuance burn, is a draft proposal that would burn a rising share of validator rewards as staking grows, cutting consensus-layer issuance to zero once about half of all ETH (roughly 60.25 million) is staked. It aims to restore ETH’s scarcity, but it is not yet scheduled for any upgrade and faces strong opposition from DeFi builders.

Why is the amount of ETH on exchanges falling?

Three sinks are absorbing ETH faster than it returns: staking (about 34% of supply), corporate treasuries (more than 6%), and spot ETFs. Exchange reserves have fallen from around 35 million ETH at the 2021 peak into the mid-teens of millions, a multi-year low that leaves a thinner tradable float.

Does the SEC treat ETH staking rewards as securities?

No. On 17 March 2026 the SEC and CFTC issued a joint interpretation classifying staking rewards on 16 digital commodities, including ETH, as non-securities, and confirming that protocol staking does not require Securities Act registration. That cleared the way for US staking ETFs from Grayscale and BlackRock.

By Marcus Halloran, senior markets editor at HOGE Wire.