ETH Supply in 2026: The Burn Broke, and Scarcity Got Political

For three years, the case for Ethereum’s monetary policy fit on a bumper sticker. Every transaction burns a little ETH, so if the network stays busy, supply shrinks and the asset grows scarcer over time. The community called it ultrasound money. In September 2026 that sticker is out of date. Ethereum now mints slightly more ETH than it destroys, the burn that was supposed to do the work has shrunk to a rounding error, and the question of how scarce ETH should be has moved from an automatic formula into a contested governance fight playing out on a GitHub pull request and a developer call.

That shift matters more than the usual supply-chart trivia, and not only to protocol engineers. The investment case for ETH as a productive, slowly deflating reserve asset rests on supply behaving a particular way. When the mechanism that produced the scarcity stops working, the scarcity becomes a decision rather than a guarantee. What follows is an analysis of where ETH supply actually sits right now, why the deflation engine broke, who is fighting over how to fix it, and what a tighter or looser supply would mean for traders heading into a Federal Reserve meeting that the market now expects to deliver a rate hike. The price was near $2,530 on 12 September, with a market capitalization close to $309 billion, according to CoinGecko, still roughly 49% below the August 2025 record.

The short version: Ethereum stopped minting scarcity and started legislating it

Three facts define ETH supply in late 2026, and they pull in different directions. First, at the protocol level Ethereum is mildly inflationary. Net issuance has run somewhere between roughly 0.2% and 0.8% a year since early 2024, depending on the measurement window, because validators are paid more new ETH than the base-fee burn removes. Second, the burn that once offset that issuance has collapsed: daily burn sits around 50 to 70 ETH, down from thousands a day at the peak, after Ethereum’s own scaling upgrades pushed activity onto Layer 2 networks. Third, the tradable float keeps shrinking anyway, because staking, corporate treasuries, and exchange-traded funds are locking ETH off the market faster than issuance adds it.

Put those together and the headline writes itself. The part of ETH’s scarcity that used to be automatic is broken, so the only way to restore it is to change the rules by hand. That is exactly what a proposal called the Tapered Issuance Burn would do, and it is exactly why some of the largest builders on Ethereum are fighting it. Supply analysis in 2026 is no longer an exercise in reading a burn dashboard. It is an exercise in reading a governance dispute.

Where ETH supply stands in September 2026

Start with the numbers, because the rest of the argument depends on them. The table below pulls the key supply figures as of early-to-mid September 2026. Trackers disagree at the margins (the circulating-supply and staked-supply series in particular drift between data providers), so treat these as close approximations rather than decimals carved in stone.

The picture is a paradox that the protocol cannot resolve on its own. On paper, more ETH exists every week. In practice, less ETH is available to buy. The first number is set by issuance and the burn. The second is set by where holders put their coins. Both are now moving against the thing that made ETH interesting as money, and fixing the first one requires a political choice that the second one makes harder.

What ultrasound money was supposed to do

The phrase ultrasound money was coined by Ethereum Foundation researcher Justin Drake, and it rested on two upgrades stacked on top of each other. In August 2021, EIP-1559 changed how gas fees work: instead of all fees going to miners, a variable base fee was burned, destroyed permanently, on every transaction. Ethereum logged its first deflationary day the following month, as CoinLedger notes in its history of the thesis. Then in September 2022, the Merge switched Ethereum from proof-of-work to proof-of-stake and cut new issuance by roughly 90%, because stakers require far less of a subsidy than miners did. Supporters called the combination a triple halving.

The logic was elegant. Issuance was now small and predictable. The burn scaled with demand. So in any period where the network was busy enough, the burn would exceed issuance and the total supply would fall. Bitcoin’s supply only ever grows, slowly; Ethereum’s could actually shrink when people used it. That is where the ultrasound branding came from, a cheeky claim to be even harder money than the sound money Bitcoiners talk about. For about eighteen months, the data cooperated.

The deflationary age, and the upgrade that ended it

From the Merge through early 2024, Ethereum was genuinely deflationary on a net basis. High activity on decentralized exchanges, NFT marketplaces, and DeFi lending kept mainnet gas prices elevated, the base fee stayed high, and the burn comfortably outran the small issuance to stakers. The supply curve bent downward. For a community that had spent years arguing that ETH was more than a tech stock with a token attached, the shrinking-supply chart was the proof.

Then came Dencun, activated on 13 March 2024. Dencun introduced blobs, a cheap new data lane designed specifically for Layer 2 rollups to post their compressed transaction data. It worked spectacularly well. Rollup fees fell by somewhere between 90% and 98% almost overnight, as the Bitget supply review documents. Users and applications that had been transacting on expensive mainnet moved to cheap L2s like Base, Arbitrum, and Optimism. That was the entire point of the roadmap: make Ethereum usable for ordinary amounts of money. It also quietly pulled the rug out from under the burn. By CoinLedger’s accounting, Ethereum’s inflation rate climbed to about 0.74% by September 2024, and the network has been mildly inflationary on balance ever since.

The L2 paradox: scaling worked, and it broke the burn

To see why a scaling success killed the deflation, you have to separate Ethereum’s two fee markets. The execution layer, where ordinary transactions and smart-contract calls happen, charges a base fee that gets burned. That is the ultrasound engine. The blob market, where rollups post their data, is a separate auction with its own pricing, and for most of the time since Dencun it has cleared at almost nothing because blob supply has exceeded blob demand. When a trade or a swap happens on an L2, the economic activity is real, but the fee it pays flows through the cheap blob lane, not the expensive execution lane. Very little ETH gets burned.

So Ethereum ended up in an awkward spot. The network is arguably more used than ever once you count L2 activity, yet the mainnet burn has fallen to roughly 50 to 70 ETH a day, per Bitget’s review, a small fraction of what issuance adds. The more successful the rollup-centric roadmap becomes, the less ETH the base layer burns. That is the paradox at the center of every honest supply analysis in 2026: Ethereum got what it wanted on scaling, and the price was the monetary property it had been advertising. Cheaper transactions are good for users and for the networks built on top, including the growing class of consumer applications that now live almost entirely on Layer 2. None of that activity does much to burn ETH on the base layer.

Fusaka put a floor under the burn, not a ceiling

The developers saw the burn collapse coming and built a partial answer into Fusaka, the upgrade that went live on 3 December 2025. Its most relevant piece for supply is EIP-7918, which sets a minimum price floor for blob transactions so that rollups always pay at least a baseline fee, even when demand is slack. Fidelity’s research team modeled the effect and found it would have added meaningful burn across the large majority of days since Dencun. Fusaka also shipped PeerDAS, which cuts the data load on home validators, and a series of blob-capacity increases that let the network carry far more rollup data than before.

Here is the catch, and it is the part that gets lost in bullish summaries. EIP-7918 is a floor, not a restoration. It stops the burn from falling to literally zero on quiet days; it does not bring back the thousands-of-ETH-a-day burn of the deflationary age. And the same Fusaka upgrade that added the floor also expanded blob capacity, which pushes more activity down the cheap lane. The net effect on supply is modest. Fusaka made the burn less fragile. It did not make Ethereum ultrasound again. For scarcity to come back automatically, mainnet base fees would need to sit far above where they have been for most of 2026, and the roadmap is explicitly designed to keep them low.

Issuance is the only lever left, and it points down

If the burn cannot be relied on, the supply story reduces to issuance, the new ETH paid to validators. And issuance is tightly linked to how much ETH is staked. Ethereum’s reward schedule pays each validator less as the total staked pool grows, because the consensus yield scales roughly with the inverse square root of the amount staked. More stakers means a thinner slice for each. With about 43 million ETH staked, close to 36% of supply and an all-time high, the base staking yield has compressed to around 2.6%, a three-year low, as Coinpedia reports. Add priority fees and MEV and a typical validator captures something in the 3.1% to 3.3% range.

The striking thing is that staking keeps rising even as the reward falls, because the marginal staker is no longer a yield-chasing individual but an institution that wants ETH exposure plus any yield at all. The validator exit queue has repeatedly cleared to zero this year while the entry queue holds well over a million ETH waiting to get in, which Arkham reads as a sign of unusually strong conviction. Jerome de Tychey, a co-founder of the Ethereum Community Conference and one of the people pushing the issuance debate, has warned that on the current trajectory there could be more than 70 million ETH staked by January 2028 if nothing changes. More staking means more total issuance even as each validator earns less, and that is the lever the reformers want to pull. If you want a fuller treatment of how those rewards are funded and what they buy, our piece on validator economics walks through the mechanics chain by chain.

EIP-8363: the plan to burn issuance to zero

The proposal at the heart of the 2026 supply debate was introduced in early August, authored by a group that includes Justin Drake and submitted through de Tychey. It was first self-numbered EIP-8361, but the editors found that number already taken and reassigned it as EIP-8363, so both labels float through the press coverage of what is a single idea. The mechanic is called the Tapered Issuance Burn. Instead of paying validators their full consensus reward, the protocol would burn a growing fraction of that reward as the staked share of supply rises. At low staking levels the burn is small. As staking approaches roughly 60.25 million ETH, about half of all supply, the burn reaches 100% and net protocol issuance to stakers falls to zero, according to Crowdfund Insider. The change would phase in over roughly 18 months and touch only the consensus layer.

The philosophy behind it is what Ethereum researchers call minimum viable issuance: pay validators only as much as security genuinely requires, and no more. The argument is that beyond a certain point, extra staking does not buy meaningfully more security, it just dilutes holders and pulls capital into staking that might otherwise circulate. EIP-8363 is a draft. It has no approval, no schedule, and no place in any upcoming upgrade. It did not make the cut for the next fork, Hegota, and its pull request remains open for debate. But it has already reframed the entire supply conversation, because it makes explicit what the broken burn implied: from here, ETH’s scarcity is something humans choose.

The builders’ revolt

The reaction from the application layer was fast and hostile, and that reaction is the real news. Stani Kulechov, the founder of lending giant Aave, argued the proposal was self-defeating, saying bluntly that Ethereum should not be punished for its growth. His worry is practical: if staking ETH eventually yields nothing net, holders who want yield may rotate into other assets, and a zero-yield ETH becomes harder to use as productive collateral across DeFi. Mike Silagadze, the chief executive of liquid-staking provider ether.fi, went further, calling the proposal so disappointing on every level and bad for both decentralization and adoption.

The counterargument, from de Tychey and the other authors, is that the window to act is closing precisely because staking is climbing so fast, and that waiting until 70 million ETH is staked would mean trying to cut rewards from a far larger and more entrenched constituency. This is not a narrow technical disagreement. It is a fight over who Ethereum’s monetary policy is supposed to serve: holders who benefit from lower issuance, or the staking and DeFi businesses whose models assume a yield. The contested resource this time is the issuance schedule itself. For the first time since the Merge, ETH’s inflation rate is on the table as a live political question, and the outcome will set the supply trajectory for years.

The vanishing float, and who controls it

While engineers argue about issuance, the market is quietly answering a different question: how much ETH is actually available to trade. The answer is less and less. The amount of ETH sitting on centralized exchanges, the most liquid pool of sell-side supply, fell to about 14.88 million ETH on 8 September, a multi-year low, according to CryptoQuant data cited by Finbold. More than 6 million ETH has left exchanges since mid-2025. That ETH did not vanish; it moved into staking contracts, into corporate treasuries, and into the custody accounts behind exchange-traded funds, all places it is far less likely to be sold on a whim.

The treasury bid is the loudest part. A group of 34 public companies now holds close to 7.9 million ETH, around 6.5% of supply, led by BitMine Immersion, which has accumulated roughly 5.93 million ETH, about 4.9% of supply, and stakes the bulk of it. US spot ETH funds hold around $15.6 billion in assets after roughly $2.3 billion of net inflows since July. Silagadze, despite opposing EIP-8363, captured the float dynamic in one line when he argued that people who stake ETH do not sell it. The result is a thin tradable float sitting on top of a slowly growing total supply, which is why the same week can bring both an inflation complaint and a supply-squeeze rally. A full accounting of who is absorbing that float appeared in our earlier supply piece; the point to carry forward here is that issuance policy and float dynamics are now two separate stories that happen to push on the same price.

Locking supply away has a side effect that rarely shows up in a bullish supply thread: concentration. When more than a third of all ETH is staked and the float keeps thinning, the question of who controls the locked ETH becomes a genuine risk to the network, not just a talking point. Lido, the largest liquid-staking protocol, still commands the biggest single share of staked ETH, though that share has slipped to around 21% from a peak near 32% in late 2023, and it captured only a small fraction of this year’s staking growth, according to CryptoSlate. The slack was taken up by exchanges and treasuries, which concentrates stake in a different way, behind a handful of custodians and corporate balance sheets.

Vitalik Buterin has repeatedly flagged staking concentration as one of the biggest risks to the Ethereum L1, pointing to the two chokepoints that matter most, block construction and the provision of staking capital. A supply that is increasingly staked through a few large operators is a supply that could, in principle, be pressured to censor or reorganize. This is why the staking debate and the supply debate are the same debate. Proposals like EIP-8363 aim partly to slow the march toward ever-higher staking; critics counter that capping rewards could push solo stakers out and leave the field to exactly the large, professionalized operators everyone says they fear. The competitive dynamics among those operators, and how each is rebuilding for a more concentrated market, are the subject of our comparison of Lido, Rocket Pool, and Frax.

Scarcity by policy versus scarcity by protocol

The cleanest way to frame ETH’s 2026 supply question is to contrast it with Bitcoin’s. Bitcoin’s scarcity is scarcity by protocol. The 21-million cap and the halving schedule are effectively constitutional; changing them is close to unthinkable and would fracture the community. An investor buying Bitcoin for its supply properties is buying a number that almost certainly will not move. The cost is rigidity: Bitcoin cannot adjust issuance to fund security as fees evolve, a tension that surfaces in long-running debates about miner revenue.

Ethereum’s scarcity, after 2026, is scarcity by policy. It is produced by a combination of a burn that now depends on mainnet demand and an issuance schedule that the community can, and is actively debating whether to, change. The upside is flexibility; Ethereum can tune its monetary policy to match what the network actually needs. The downside is credibility. A supply that a governance process can alter is a supply that a governance process can alter in either direction, and a buyer who wants a hard monetary guarantee has to weigh the risk that the rules shift. That trade-off, adaptive but discretionary versus fixed but rigid, is the real distinction between the two assets now, and it is more honest than the old ultrasound-versus-sound slogan.

Where the SEC now sits on staked ETH

One reason the float is disappearing into staking is that the regulatory fog around it has lifted in the US. In a statement issued in August 2025, the SEC’s Division of Corporation Finance said certain liquid-staking activities do not involve the offer and sale of securities, treating a liquid-staking token as closer to a receipt for deposited goods than an investment contract, provided the provider stays in an administrative role. A joint SEC and CFTC interpretation in March 2026 then classed a group of major tokens, Ethereum among them, as digital commodities and treated staking rewards as outside the securities framework.

That clarity is why staking yield could be packaged into regulated products. Grayscale’s Ethereum fund began distributing staking rewards, and BlackRock launched a staked-ether trust in 2026, giving institutions a way to earn the consensus yield without running validators. The more that staked ETH becomes an institutional product, the more supply gets locked in long-horizon hands, and the more the EIP-8363 question matters to those institutions, because a zero-net-issuance future changes the yield their products can advertise. Enforcement has not disappeared, as our review of SEC crypto enforcement in 2026 makes clear, but the staking question specifically has moved from existential to operational.

What a hot CPI and Fed week mean for the supply story

Supply is a slow, structural variable. Over any few-week window, ETH trades on demand and macro, and this particular week is loud. On 11 September the Bureau of Labor Statistics reported that August consumer prices rose 0.4% on the month, putting annual inflation at 3.4%, in line with forecasts, while core inflation ran a touch hot at 0.3% monthly and 2.4% annually, according to CNBC. Gasoline did much of the monthly damage. Traders took it as confirmation that the Federal Reserve has room to tighten: futures moved to price roughly an 85% chance of a 25-basis-point rate hike at the 15-16 September meeting, up from about 71% a day earlier, Benzinga noted. The first dot plot under the new Fed chair lands the same week, and a CLARITY Act procedural vote in the Senate is scheduled for 15 September.

For ETH supply that is a specific kind of test. A hawkish Fed raises the dollar cost of holding any non-yielding or low-yielding asset, and ETH’s base staking yield near 2.6% does not clear a short-term Treasury bill. So the structural story, shrinking float and a possible future of zero net issuance, collides with a macro story that makes the opportunity cost of holding ETH higher. The supply tailwind is real but slow; the macro headwind is immediate. Our analysis of how the inflation print and Fed week reshaped price targets is the companion read for the macro side of this. The supply analysis tells you which way the structural wind blows; it does not tell you the network will outrun a rate hike in September.

Three ways this resolves into 2027

Pulling the threads together, the supply trajectory into 2027 hinges on three variables: whether issuance policy changes, how high staking climbs, and whether future upgrades revive or further suppress the burn. The scenarios below are not forecasts with probabilities attached; they are the branches worth watching, with the signals that would tell you which one is playing out.

The wildcard running under all three is the next upgrade. Glamsterdam, in its final testing phase and slated for the back half of 2026 though hard-fork dates routinely slip, brings enshrined proposer-builder separation, block-level access lists, and a large gas-limit increase aimed at pushing Layer 1 throughput sharply higher, per The Block. More throughput can mean more base-fee burn if mainnet demand shows up, or continued suppression if activity keeps migrating to rollups. Hegota, the upgrade after it, is focused on censorship resistance rather than supply. None of this is priced with any confidence, which is why the market’s own read, laid out in our look at what the market already priced, leans on demand and macro rather than on a supply curve that has become a moving target.

The bottom line for anyone analyzing ETH supply in 2026 is to stop looking for a single number that says deflationary or inflationary. That framing died with the burn. What replaced it is a set of decisions: how much to issue, how much staking to encourage, and how aggressively to scale. ETH is still becoming scarcer in the way that counts for price, through a shrinking tradable float, but the clean, automatic scarcity of the ultrasound era is gone. In its place is a monetary policy that Ethereum now has to argue about in public, one pull request at a time.

Frequently Asked Questions

Is Ethereum still deflationary in 2026?

No. Ethereum has been mildly inflationary on a net basis since the March 2024 Dencun upgrade, with net issuance running roughly 0.2% to 0.8% a year depending on the window. It still has brief deflationary stretches during periods of heavy mainnet activity, but the sustained deflation of the ultrasound-money era is gone because the burn collapsed after transactions moved to Layer 2 networks.

How much ETH has been burned?

More than 4.6 million ETH has been permanently destroyed since the EIP-1559 burn mechanism went live in August 2021. But the pace has slowed dramatically: the daily burn has fallen to roughly 50 to 70 ETH, down from thousands per day at the peak, which is why issuance now outruns it.

What is EIP-8363, the tapered issuance burn?

EIP-8363, sometimes still called EIP-8361, is a draft proposal that would burn a rising share of validator rewards as the staked share of supply climbs, reaching zero net issuance once roughly half of all ETH, about 60.25 million, is staked. It is not scheduled for any upgrade and is opposed by major builders including Aave and ether.fi.

How much ETH is staked right now?

About 43 million ETH, close to 36% of the total supply, which is an all-time high. The base staking yield has compressed to around 2.6%, a three-year low, because rewards are shared across a larger validator set; priority fees and MEV push the all-in yield to roughly 3.1% to 3.3%.

Does a shrinking ETH supply mean the price will go up?

Not directly. A tighter supply and a shrinking tradable float are a structural tailwind, but near-term price is driven by demand and macro conditions, such as Federal Reserve policy and inflation data. In September 2026 the supply story is bullish on a slow structural basis while the macro backdrop, including a likely rate hike, works the other way.

By Priya Reddy, markets editor at HOGE Wire, covering Ethereum monetary policy, staking, and digital-asset supply dynamics.