Ethereum Token Supply in 2026: The “Ultrasound Money” Story Got Complicated

Originally published October 2022 · Fully updated April 7, 2026

In October 2022, right after the Merge, we wrote that Ethereum’s token supply was declining toward pre-Merge levels. On-chain data at the time showed ETH supply was roughly 1,700 tokens away from matching the 120,520,000 circulating at the moment of the Merge itself. The narrative was exciting: EIP-1559 burns were outpacing new issuance, Ethereum was going deflationary, and the “ultrasound money” thesis was playing out in real time.

Three and a half years later, the honest update is more complicated.

ETH supply has not continued declining. It has slightly increased since the Merge — now sitting at approximately 120.7–121.5 million tokens as of April 2026 — meaning more ETH is in circulation today than existed on September 15, 2022. The network is technically experiencing mild annual inflation of around 0.23%.

The reason for this reversal is not a failure of the burn mechanism. It’s something more interesting: the upgrades that made Ethereum dramatically more efficient also made it less deflationary.

What the 2022 Article Got Right

The fundamental mechanics described in 2022 remain accurate. Understanding them is necessary before explaining what changed.

EIP-1559 (August 2021) changed how Ethereum transaction fees work. Before this upgrade, the entire fee went to miners. After EIP-1559, fees are split: a “base fee” set algorithmically by the network is permanently burned (destroyed), and a small “priority fee” (tip) goes to validators. This means every transaction on Ethereum reduces the total supply of ETH in proportion to how much network activity is happening.

The Merge (September 15, 2022) switched Ethereum from Proof-of-Work to Proof-of-Stake. Before the Merge, the network issued approximately 13,000 ETH per day to miners. After, issuance dropped to roughly 1,700 ETH per day paid to stakers — a reduction of approximately 88%. Combined with the EIP-1559 burn, this created the conditions for Ethereum to be “net deflationary”: burning more ETH each day than it was creating.

The 2022 observation was correct. In the months immediately following the Merge, on days when average gas prices exceeded roughly 16 gwei, the daily burn exceeded daily issuance. Supply declined. The “ultrasound money” thesis was — briefly — empirically supported.

What Changed: The Dencun Effect

March 2024 introduced the Dencun upgrade, and specifically EIP-4844, known as proto-danksharding.

Before Dencun, Layer-2 networks like Arbitrum, Optimism, and Base had to post their transaction data to Ethereum’s main chain using regular calldata — competing for the same block space as every other transaction. This was expensive: L2 data storage alone accounted for roughly 95% of gas fees on some rollup transactions.

Dencun introduced “blobs” — a separate, cheaper data storage mechanism specifically for L2 rollups. Blobs are stored temporarily (around 18 days), just long enough for fraud proofs to be verified, then discarded. They don’t compete with regular transactions for block space.

The result: L2 transaction fees fell by 90–98% practically overnight. Millions of users who were previously transacting on Ethereum’s main chain migrated to L2s where transactions cost fractions of a cent.

This was exactly what Ethereum’s roadmap intended. But it had an unintended consequence for the supply story: if most users are transacting on L2s at near-zero fees, far less ETH is being burned per unit of economic activity on the network. The daily burn rate collapsed from thousands of ETH per day to as low as 50–70 ETH per day in Q1 2025 — historic lows not seen since before EIP-1559 launched.

With ~1,700 ETH still being issued to stakers daily and only 50–70 being burned, Ethereum’s net issuance turned positive. The “ultrasound money” narrative hit its first real stress test.

The Supply Numbers Today (April 2026)

Sources: Ultrasound.money, Glassnode, Ethereum.org, Everstake

The Staking Side of the Equation

While the “ultrasound money” narrative has been complicated by lower burn rates, the staking picture presents a different kind of supply compression.

As of April 2026, approximately 36–37 million ETH — more than 30% of total supply — is locked in staking contracts. These tokens are not liquid. They’re not on exchanges. They’re not available for day-to-day trading. The validators who operate them have made a deliberate commitment to secure the network in exchange for ~3.5–4.5% annual yield.

The Pectra upgrade in May 2025 meaningfully changed staking mechanics. Before Pectra, each validator had a maximum effective balance of 32 ETH — meaning large operators needed to run thousands of separate validators for large positions. Pectra raised the maximum to 2,048 ETH, allowing consolidation. Despite 16,000 fewer active validators after the upgrade, total staked ETH actually rose to 36.8 million, as individual positions grew rather than proliferating.

On March 30, 2026, the Ethereum Foundation made its largest single staking event — depositing 22,517 ETH (worth approximately $46 million) into the Beacon Chain. This is both a commitment signal from the network’s primary development organisation and a direct reduction in circulating supply.

Public companies and “digital asset treasury” (DAT) firms — the Ethereum equivalent of MicroStrategy — now control over 5.5% of Ethereum’s total circulating supply. These entities tend to be very long-duration holders, further reducing the liquid float.

The practical supply picture: even with ~0.23% annual inflation from net issuance, roughly 30% of all ETH is locked in staking. The “tradeable float” — ETH actually available on exchanges — is significantly smaller than the headline circulating supply number.

The Upgrade Calendar: How Ethereum Tried to Fix the Burn Rate

After the Dencun upgrade produced the burn rate collapse, the Ethereum development community recognised the tension: the scaling success had undermined the deflationary narrative.

Fusaka (December 2025) addressed this directly with EIP-7918, the “Blob Base Fee Bound.” This establishes a minimum price floor for blob transactions — even during low L2 data demand, rollups now pay a minimum fee proportional to the execution base fee. The effect: a guaranteed minimum stream of ETH being burned even during quiet periods, preventing the burn rate from collapsing to near-zero again. The market is still watching to see whether this materialises into measurable improvement in 2026.

PeerDAS (EIP-7594, part of Fusaka) enables data availability sampling — validators can verify large data blobs by sampling small pieces rather than downloading entire datasets. This allows safe blob capacity expansion to approximately 8x. With more L2 throughput comes more blob fee volume, which under EIP-7918’s minimum pricing means more consistent burning.

Glamsterdam (2026 roadmap) is the next planned upgrade, focused on further gas accounting refinements and validator performance improvements.

The pattern: Ethereum’s roadmap is optimising for scalability (which reduces mainnet fees) while simultaneously trying to preserve the economic mechanisms that make ETH scarce. The tension between these goals is not fully resolved.

How DeFi Contributes to Ethereum Burns

Despite the reduced burn rate from L2 migration, significant on-chain DeFi activity continues burning meaningful amounts of ETH on the mainnet.

Uniswap has consistently led Ethereum’s burn leaderboard, contributing millions of dollars in burned ETH during high-activity weeks. The week of October 2024 alone saw $30.6 million burned by top DeFi protocols, with Uniswap, 1inch, and MetaMask leading the charge.

More recently, March 2026 data showed Gnosis, MetaMask, and Uniswap topping the weekly burn charts, burning approximately 2,277 ETH worth $3.63 million in a single week. These burns reflect real economic activity on Ethereum’s mainnet — primarily large-value DeFi transactions and institutional flows that prefer the security of L1 settlement over L2 speed.

A 163% surge in Ethereum’s burn rate in September 2024 preceded a significant ETH price rally — historical data suggests elevated burn rates signal increased network demand, which tends to correlate with price appreciation.

The weekly top-ten burner data tracked consistently by Phoenix Group shows that Uniswap, MetaMask, Gnosis, Aave, 1inch, Pendle, Kyber, 0x Protocol, and Paraswap collectively remove hundreds of ETH from supply each week. Over months and years, this adds up: over 5,022 ETH burned by top DeFi projects in a single 30-day period has been documented multiple times.

The “Ultrasound Money” Debate in 2026

The honest 2026 assessment requires separating the narrative from the mechanics.

What’s still true:

• Ethereum’s issuance is 90% lower than pre-Merge levels
• ~4.6 million ETH has been permanently destroyed since EIP-1559
• 30%+ of supply is locked in staking, reducing liquid float
• During high network activity, Ethereum can and does go net deflationary
• EIP-7918 (Fusaka) establishes a burn floor that prevents rate collapse to near-zero
• Ethereum’s annual inflation of ~0.23% compares favourably to Bitcoin’s current ~1.6% post-halving issuance rate and extremely favourably to pre-Merge Ethereum at 4–5%

What’s no longer straightforwardly true:

• Ethereum is not currently net deflationary
• The supply has grown, not shrunk, since the Merge
• The ultrasound money thesis is conditional: it requires mainnet gas activity to push average fees above approximately 16 gwei

What the data actually says: Ethereum is a “low-inflation asset” that can become deflationary during demand spikes. The Fusaka upgrade attempts to permanently lock in some deflationary pressure via minimum blob pricing. Whether this is sufficient depends on whether mainnet L1 activity — primarily large DeFi transactions, institutional settlements, RWA tokenization, and stablecoin flows — continues to grow enough to push base fees consistently above the deflation threshold.

Calling ETH “inflationary” because of 0.23% annual net issuance misses the context. Pre-Merge, Ethereum inflated at 4–5% annually. Pre-EIP-1559, there was no burn at all. The current situation — mild inflation with a powerful mechanism that flips to deflation during high activity, plus 30% of supply locked in yield-bearing staking — is genuinely different from every previous version of Ethereum’s monetary policy.

Real-World Asset Tokenization and Future Burn Dynamics

One development that may meaningfully change the burn rate picture in 2026–2027: institutional adoption of Ethereum for real-world asset (RWA) tokenization.

On April 6, 2026, a report identified major financial institutions actively transitioning parts of the $12.5 trillion global repo market onto Ethereum. BlackRock’s BUIDL fund has grown to over $1.8 billion on-chain. Franklin Templeton, JPMorgan, and UBS have run live asset settlement pilots on Ethereum infrastructure. Each of these institutional transactions generates L1 gas fees — high-value, low-frequency transactions that burn significant ETH per transaction even if they’re individually rare.

If institutional RWA settlement at the L1 level grows alongside L2 retail activity, the two together could push average mainnet gas above the 16 gwei deflation threshold more consistently. The bull case for Ethereum’s supply model in 2026–2030: L2s handle retail volume at near-zero fees (small per-transaction burn), while L1 handles institutional settlement at high value per transaction (large per-transaction burn). Together, the two activity types maintain consistent deflationary pressure.

The bear case: most institutional settlement also migrates to L2s as they mature, reducing L1 fee pressure, and the only remaining mechanism keeping ETH from higher inflation is the Fusaka blob fee floor.

Where Supply Stands Now vs. the Original Article’s Thesis

The October 2022 article correctly identified that:

1. The Merge dramatically reduced ETH issuance ✓
2. EIP-1559 burns were removing ETH from circulation ✓
3. Supply was declining toward pre-Merge levels ✓

What the 2022 article couldn’t know:

1. The Dencun upgrade (March 2024) would shift L2 activity off-chain, collapsing the burn rate
2. Supply would reverse direction and slightly exceed Merge levels by 2026
3. The Pectra upgrade would allow validator consolidation while staked ETH continued growing
4. Fusaka would add minimum blob pricing to restore a burn floor

The supply story has not disproven the long-term thesis — it has complicated it. Ethereum’s monetary policy is dynamic, not fixed. It responds to the network’s own success. When Ethereum’s scaling succeeds (users migrate to L2s), burns decrease. When institutional demand for L1 settlement grows, burns increase. The “correct” level of ETH issuance and burning is the one that balances network security (validators need yield), user affordability (fees must stay low), and store-of-value properties (supply should not inflate excessively).

As of April 2026, the balance sits at approximately 0.23% annual inflation — lower than almost every fiat currency, far lower than pre-Merge Ethereum, and with mechanisms in place to flip deflationary when network demand warrants it.