Here's a number that should bother you: 0.335 gwei.
That's the base fee on Ethereum mainnet as of this writing. Not a quiet weekend anomaly. Not a holiday lull. The sustained resting rate of the most valuable general-purpose block space on the planet.
Now here's the number that should bother you more: 13.85 gwei. That's the base fee Ethereum needs to hold, at the current 60 million gas ceiling, for EIP-1559's burn mechanism to fully offset new issuance. The distance between those two figures โ roughly 41x โ is the entire story of Ethereum's supply problem in 2026, and almost nobody is pricing it correctly.
I spent the last two weeks rebuilding the issuance and burn ledger from raw block data, cross-checking ethsupply.fyi against my own tallies. The result isn't a scandal. It's worse than a scandal. It's a structural condition the market has been trained to ignore.
Let me lay out the machinery, because the machinery is where the misunderstanding lives.
EIP-1559, live since August 2021, split transaction fees into two streams: a base fee that gets burned, and a priority fee that goes to the block producer. The burn was supposed to be Ethereum's scarcity engine โ the mechanism that made ETH "ultrasound money," a supply that shrinks as usage grows.
For a while, it worked. During the 2021-2022 bull run, base fees routinely printed 50, 100, even 300 gwei. Daily burns ran into the tens of thousands of ETH. The narrative calcified into dogma: Ethereum is a fee-burning machine, and the more people use it, the scarcer ETH becomes.
Then EIP-4844 arrived with blobs โ a separate data-availability channel for Layer 2 rollups โ and the migration began in earnest. High-value execution moved off L1. Rollups absorbed the transaction flow. And the base fee, stripped of its most lucrative demand, collapsed.
The Glamsterdam upgrade is the next chapter: a conditional push toward a 200 million gas limit, roughly triple today's ceiling. The pitch is simple โ more block space, more capacity, more room for activity. What the pitch omits is the arithmetic that follows. Expanding the gas ceiling doesn't increase the burn; it dilutes the fee required per unit of gas to hit any given burn target. More gas per block means the same total fee revenue spreads across a wider surface, pushing the equilibrium base fee lower, not higher.
That's the trap. And it's a trap that a specific kind of investor โ the one who bought the ultrasonic-money thesis in 2021 and never updated the model โ is walking into blind.
Let me walk through the numbers.
Over the 282-day window ending October 9, 2026, Ethereum issued 796,623.377 ETH to validators. Against that, the network burned 16,524.553 ETH through execution-layer base fees and blob fees combined. Slashing โ the penalty mechanism โ removed another 1,685.919 ETH, plus a rounding-error 0.059 ETH from other burn events.
Net that out: 778,412.846 ETH added to supply. Annualized, that's roughly 1,007,000 ETH of net new coins per year, against a supply snapshot of about 122,116,000 ETH.
Run the percentages. The fee burn covered 2.07% of issuance. If you generously fold slashing into the "burn" column โ which you shouldn't, because slashing isn't a fee and doesn't reflect user demand โ you get 2.29%. Either way, the headline is brutal: Ethereum's fee burn is now a rounding error against its own money printer.
This is not a temporary dip. It's the equilibrium state of a network whose execution demand has been structurally exported to L2s while its issuance continues on a schedule tied to staking participation.
And that last point deserves its own paragraph, because it's the part the ultrasonic-money crowd never internalized. Proof-of-stake issuance is not fixed. It scales with the amount of ETH staked. More validators, more issuance. The protocol pays for its security budget in newly minted coins, and the security budget grows as the validator set grows.
As of this writing, 43,754,000 ETH is staked across 853,325 active validators โ roughly 35.8% of total supply. Run issuance against that stake and you get a gross staking yield around 2.36%. Subtract node costs, VPS bills, the operational drag of running validator infrastructure, and the net sits somewhere between 1.5% and 2.0%.
That's the incentive structure. Staking rewards are not a share of network revenue. They're a transfer of new coins from all holders to validators. When the fee burn was strong, that transfer was partly offset. In 2026, it isn't offset at all.
I remember making this exact argument in a Warsaw fintech meeting back in 2022, when everyone wanted to talk about Terra's algorithmic stablecoin as a "tech failure." I argued then that it was a liquidity crisis wearing a technical costume โ and the contagion into Celsius and Three Arrows proved the point. The lesson holds here. A supply mechanism is only as strong as the demand that feeds it, and Ethereum's demand has been rerouted to a layer that pays it almost nothing.
Now watch what happens when you try to fix the burn through the gas limit.
At the current 60 million gas ceiling, the break-even base fee โ the rate at which burn equals issuance โ is 13.85 gwei. At the proposed 200 million ceiling, it drops to about 4.16 gwei. Both numbers assume the same total gas consumption. Raising the ceiling lowers the break-even rate, because you're asking each unit of gas to carry less of the burn burden.
So the Glamsterdam "expansion" is, from a supply perspective, a downgrade. It makes the burn target easier to hit in per-gas terms but harder to sustain in aggregate, because the extra block space has to be filled to generate the same fee revenue โ and there's no evidence the demand exists to fill it.
Which brings me to the structural problem no upgrade addresses: the priority fee doesn't burn. It goes to the block producer. In a low-base-fee environment, validators still earn from priority fees and MEV, but those flows bypass the burn entirely. The value capture that ETH holders were promised โ "hold the asset, share the network's income" โ has been quietly rerouted to the validator set and the MEV supply chain.
Let me be precise about magnitudes, because I don't want to overstate this. Blob fees, the new revenue channel from EIP-4844, are negligible. In one sampled epoch, blob burn totaled 0.000442 ETH. That's not a channel. That's a rounding artifact.
And the competitive picture sharpens the point. Solana's fee revenue has now surpassed Ethereum's L1. The chain with the highest institutional acceptance, the longest security track record, the deepest liquidity โ earns less in transaction fees than the throughput-focused challenger. That's not a death knell. But it's a signal that Ethereum's economic gravity has shifted from "where activity happens" to "where activity settles," and the settlement layer is not capturing the value.
Here's where my own experience colors the read. I spent three months in 2020 reverse-engineering Curve and Uniswap V2 pool mechanics, and the lesson that stuck wasn't about arbitrage. It was that fee flows follow the path of least friction, and protocol designers consistently underestimate how fast capital routes around expensive infrastructure. I watched the same pattern in 2017, when I built a Python script to track gas fees and token distribution across 50+ ICOs and found that 80% of them failed on vesting structure, not technology. The market kept buying the pitch. The ledger kept saying otherwise.
Ethereum in 2026 isn't Terra. There's no death spiral here. But there is a slow-motion divergence between the narrative ("scarce, deflationary, ultrasonic") and the ledger (0.83% annualized net issuance, 2% burn coverage). That divergence is where the market is mispricing risk.
Let me put the supply trajectory in plain terms. Ethereum today looks less like Bitcoin after the 2024 halving โ where issuance is capped and shrinking โ and more like a productive asset with a modest, persistent inflation rate. If you're holding ETH as a "digital gold" hedge, you're holding an asset that prints roughly 0.8% more of itself every year, with issuance trending up if staking participation keeps growing.
Now let me argue against myself, because the comfortable version of this thesis is too easy.
The bearish read says Ethereum's fee burn is dying and ETH is a stealth inflation asset. The counter-read is that burn rate was never the right metric, and everyone chasing it got the causal story backwards.
Think about what the burn actually represents: it's a measure of L1 congestion, not network value. A chain that burns enormous fees is a chain that's too expensive to use โ that was Ethereum's 2021 problem, not its achievement. The migration to L2s, which crushed the burn, is the same migration that made Ethereum usable. You can't celebrate cheap rollup transactions and mourn the lost burn revenue. They're the same phenomenon.
The deeper point is about where value accrues. Ethereum's strategic bet is that the L1 becomes a settlement and security layer โ high-value, low-frequency, maximally secure โ while execution scales out to L2s. In that world, L1 fee revenue was always going to be small relative to the ecosystem's total economic activity. Judging Ethereum by its L1 burn is like judging Visa by the revenue from its fraud-detection department. It's a real line item. It's not the business.
But โ and this is where the contrarian read folds back on itself โ that defense only works if the value eventually flows back to ETH holders through some channel. And right now, it doesn't. The L2s pay near-zero for settlement and data availability. Blob fees are negligible. Priority fees bypass holders. The staking yield is a dilution transfer, not a revenue share.
So the bull case requires a leap of faith: that a settlement layer with minimal direct revenue can still justify a premium valuation because of its security and network effects. That's not an absurd claim. It's just not a claim the current supply data supports. The market is paying for a revenue stream that has been structurally capped, and calling it scarcity.
There's also a coordination risk nobody's modeling. The 200 million gas target is conditional โ it depends on validators actually configuring their nodes to support it. The upgrade doesn't set the ceiling; the validator set does. If validators, chasing priority fees, raise the limit while demand stays flat, the base fee falls further and the burn-issuance gap widens. The "expansion" becomes self-defeating. And if they don't raise it, the upgrade's headline benefit never materializes. Either way, the mechanism is governed by a decentralized set of profit-maximizing operators, not a protocol-level guarantee.
That's the blind spot: the market treats "Glamsterdam" as a deflationary catalyst. The arithmetic says it's more likely to be a dilution catalyst. Liquidity doesn't move because a roadmap promises it will. It moves when the incentive math actually changes โ and on this ledger, it hasn't.
Another rug? No, just a liquidity trap. The kind that looks like progress from thirty thousand feet and looks like a supply overhang from ground level.
So where does this leave the cycle?
Ethereum in late 2026 is a network in transition, priced as if the transition is already won. The supply data says the transition has a cost: persistent mild inflation, weak value capture at L1, and a staking economy that runs on issuance rather than revenue.
Watch two things. First, the validator gas-limit adoption curve after Glamsterdam โ if the ceiling rises while the base fee keeps falling, the burn thesis is dead for this cycle. Second, the staking participation trend โ if the ~2.36% gross yield stops attracting new validators, issuance decelerates, and the inflation picture improves for reasons nobody will credit to the protocol.
The uncomfortable question isn't whether Ethereum is scarce. It's whether anyone still believes the scarcity story enough to keep buying it while the ledger says otherwise. Because the coins keep printing, the burn keeps rounding to zero, and the gap keeps sitting there at 41x โ patient, quiet, and completely unpriced.
