The Numbers Nobody Is Talking About
The Ethereum Foundation has a problem. It's called state growth, and it's been eating the network alive since the Merge.
On June 15, 2026, the foundation published something unusual: a public replay test dashboard. Not a blog post. Not a tweet. A live, queryable dashboard showing exactly what happens when you replay 929,731,274 historical transactions under new gas pricing rules. That's roughly 18 months of mainnet traffic—from December 2024 through June 2026—run through the proposed EIP-8037 and EIP-8038 pricing models.
The results should terrify every DeFi developer reading this.
Under EIP-8037 alone, 174 million transactions would need repairs. 2,687,652 transactions would be potentially broken beyond simple fixes. Under EIP-8038, the numbers get worse: 84.7 million repairable, 3,036,537 potentially destroyed.
We're not talking about some obscure NFT mint contract nobody uses. The foundation's own outreach report flags repeated failures in ERC-4337 EntryPoint contracts, Across Protocol, Socket/Bungee, CoW Protocol, and 0x. The infrastructure layer—wallets, RPC providers, indexers, gas estimators—all need to adapt.
The upgrade is codenamed Glamsterdam. It's scheduled for Q4 2026. And it's coming whether the ecosystem is ready or not.
The Mechanics: What's Actually Changing
Let me be precise about what EIP-8037 and EIP-8038 actually do, because the mainstream coverage has been sloppy.
EIP-8037 introduces a separate state-gas dimension, entirely distinct from execution gas. Every new byte of state created on-chain gets charged a flat 1,530 gas. The target is to cap annual state growth at 120 GiB under the 150 million reference block limit. Worst case—if block gas limits hit 200 million—that's 160 GiB per year.
EIP-8038 rebalances the cost of accessing existing state, based on client benchmarks taken from the March 2026 mainnet state snapshot.
The specific cost increases are where the pain lives:
- Creating a new account: 25,000 gas → 183,600 gas. That's a 7.3x multiplier
- Creating a new storage slot: 20,000 gas → 97,920 gas. 4.9x
- Deploying a 24 KiB contract with a new account: 4,947,200 gas → 37,784,880 gas. 7.6x
These aren't theoretical numbers. They're derived from actual client benchmarks against the current mainnet state. The foundation ran the tests. The data is public.
Why This Matters Now
Here's what I've learned from auditing code in this industry since 2017: volume spikes lie; liquidity flows tell the truth. And right now, the flow is telling us something uncomfortable.
The Ethereum Foundation knows these numbers. They published the replay test dashboard precisely because they understand the blast radius. But the market hasn't priced this in. The narrative is still "Ethereum triples throughput," not "Ethereum's state pricing model is about to break millions of transactions."
The three-times throughput target is an engineering support goal, not a guarantee. That distinction matters. When the foundation says "the protocol should support 3x capacity," they're saying the resource pricing model won't be the bottleneck. They are not saying mainnet will immediately process three times the transactions.
The chart doesn't lie, but the marketing does.
The Replay Test: What 929 Million Transactions Reveal
The replay test methodology is sound. Take historical transactions, re-execute them under the proposed gas pricing rules, and see what breaks. It's the closest thing we have to time travel for smart contract compatibility.
The scale is unprecedented: 929,731,274 transactions spanning December 2024 to June 2026. The foundation ran this through both EIP-8037 and EIP-8038 pricing models, plus a combined scenario with a 10x gas limit multiplier.
Here's the breakdown:
- EIP-8037: 174M transactions repairable, 2.7M potentially broken
- EIP-8038: 84.7M repairable, 3.0M potentially broken
- Combined: The overlap creates additional complexity for bundlers and infrastructure providers
But here's the critical nuance most analysts are missing: repeated activity dominates the count. A single contract with millions of transactions inflates the numbers. The actual number of unique affected contracts is likely far smaller than the transaction count suggests. That doesn't mean the problem is trivial—it means the problem is concentrated in specific, high-usage contracts.
And those contracts are exactly the ones the foundation flagged: ERC-4337 EntryPoint, Across, Socket/Bungee, CoW Protocol, 0x.
The Smart Account Stack Is Ground Zero
The foundation's outreach report is unusually candid. They prioritized eth-infinitism's ERC-4337 EntryPoint and the broader smart account stack. Why? Because these contracts are immutable in ways that make adaptation expensive.
Here's the technical reality: when you deploy a smart account with a specific EntryPoint version, you're locked into that version's gas handling logic. If EIP-8037 changes how state creation is priced, the EntryPoint's internal accounting—designed for the old pricing model—can fail in ways that are difficult to predict.
The foundation's report identifies the need for "new EntryPoint, account, factory, or validator implementations." That's developer-speak for: your existing smart accounts may need to be migrated to entirely new contracts.
For immutable contracts, this is catastrophic. There's no "patch" for a contract that can't be modified. The only solution is a new deployment and a migration path for users. And migration paths for smart accounts are notoriously complex—they involve social recovery, session keys, and all the other features that made account abstraction attractive in the first place.
The Counterfactual Failure Problem
Here's a term you need to understand: counterfactual failure.
A counterfactual failure is a transaction that succeeds under current rules but fails under the proposed new rules. It's not a bug in the transaction. It's a bug in the assumptions the contract was built on.
The replay tests are full of these. A DEX aggregator that assumes storage slots cost 20,000 gas will behave differently when they cost 97,920 gas. The transaction might still execute, but the gas accounting inside the contract—the gasleft() checks, the refund calculations, the internal price oracles—can all produce incorrect results.
Speed is safety when the exploit is already live. But this isn't an exploit. It's a deliberate protocol-level change that happens to break a significant chunk of the ecosystem's assumptions.
The Infrastructure Layer: Silent but Critical
While everyone's focused on DeFi protocols, the infrastructure layer is quietly facing the biggest adaptation burden.
Wallets need to update their gas estimation logic. RPC providers need to handle the new state-gas dimension. Indexers need to account for the new pricing model. Node operators need to verify their clients handle the new rules correctly. Gas estimators need to distinguish between execution gas and state gas.
The foundation announced the Platåberget testnet specifically for Glamsterdam testing, with a warning that "old tooling may not work as expected." That warning is doing a lot of heavy lifting. In practice, it means every wallet, every indexer, every gas estimation library, every monitoring tool needs to be updated and tested against the new rules.
We don't have to guess what happens when infrastructure fails during a network upgrade. We've seen it before.
The Contrarian Angle: What the Foundation Isn't Saying
Here's what I find most interesting about this upgrade, and it's the angle almost nobody is covering.
The state-gas mechanism isn't just about controlling state growth. It's about changing the economics of state ownership on Ethereum. By making state creation dramatically more expensive, the foundation is implicitly encouraging a specific kind of usage pattern: fewer, larger, more efficient contracts rather than many small, wasteful ones.
This has profound implications for the L2 ecosystem. If L1 state becomes more expensive, the cost of posting calldata or blobs to L1 changes. The foundation's own analysis suggests L2s might benefit from the L1 throughput increase, but the state-gas mechanism could also push more state-heavy activity onto L2s, where storage is cheaper.
There's also a longer-term signal here. The 160 GiB/year worst-case number suggests the foundation has already done the math on a 200 million gas ceiling. That's not speculation—that's planning. The foundation knows where this is going, and the state-gas mechanism is the first step toward a more deliberate state growth policy.
And here's the question that keeps me up at night: if state growth is controlled effectively, does Ethereum delay statelessness or Verkle trees? The entire rationale for those technologies was state growth being unsustainable. If EIP-8037 solves that problem through pricing, the urgency for statelessness diminishes.
The Governance Question
Let's talk about governance, because it matters here.
EIP-8037 and EIP-8038 are both in "Review" status. Not "Last Call." Not "Final." Review. That means the specific parameters—the 1,530 gas per byte, the 120 GiB annual target, the specific fee multipliers—can still change.
The foundation has been transparent about the process. The replay test dashboard is public. The outreach report is public. The testnet is announced. But transparency doesn't equal consensus.
Here's the tension: Ethereum's governance is deliberately conservative. The EIP process requires rough consensus across a diverse group of stakeholders—core developers, application developers, miners (well, validators now), and users. Each group has different incentives.
Validators want predictable gas prices. Application developers want stability. Users want low fees. The foundation wants sustainability. These aren't always aligned.
The fact that the replay tests show 2.7-3.0 million potentially broken transactions means the foundation knows this upgrade will break things. The question is whether the ecosystem accepts that cost as necessary for long-term sustainability, or whether enough stakeholders push back to delay the upgrade.
The chart doesn't lie, but the politics do.
What the Market Is Missing
The market hasn't priced this in. Not really.
ETH's price hasn't moved significantly on the Glamsterdam news. The narrative is still "Ethereum is scaling," not "Ethereum is about to break millions of transactions."
That's an opportunity for people who understand the technical details.
The short-term risk is FUD. If a prominent protocol fails during the upgrade—if a major DEX's aggregator starts producing wrong quotes, if a cross-chain bridge's counterfactual failure traps user funds—the market will react. Not because the fundamentals changed, but because the narrative shifted from "scaling" to "broken."
The longer-term opportunity is in adaptation services. Every protocol that needs to migrate to new EntryPoint contracts needs developers who understand the migration path. Every infrastructure provider needs to update their tooling. Every wallet needs to update its gas estimation. That's a lot of work, and it's all needed before Q4 2026.
Speed is safety when the exploit is already live. But the exploit isn't live yet. The upgrade is.
The Immutable Contract Problem
Let me be specific about the immutable contract problem, because it's the highest-risk item on the board.
Immutable contracts can't be patched. If they break under new gas pricing rules, the only solution is a new deployment. That means:
- Deploying a new version of the contract
- Migrating user state from the old contract to the new one
- Updating any external references to the old contract
- Convincing users to interact with the new contract
For simple contracts, this is manageable. For complex DeFi protocols with billions in TVL, it's a nightmare. The migration path for a lending protocol, for example, involves moving positions, updating collateralization ratios, and ensuring no funds are trapped during the transition.
And here's the uncomfortable truth: some protocols won't migrate. If a contract has minimal usage, if the protocol team has moved on, if the migration cost exceeds the value at stake—the contract will simply break. Users with funds in those contracts will lose access.
The foundation's outreach report identifies the most critical contracts. But the long tail of smart contracts—the thousands of small, unmaintained contracts that still hold user funds—won't all get fixed. Some will become "zombie contracts," technically present on-chain but functionally dead.
The Opportunistic Angle
Every disruption creates opportunity. Here's where I see the upside.
First, migration services. The demand for moving immutable contracts to new implementations will create a market for specialized migration tooling and services. Teams that can do this safely and efficiently will have more work than they can handle.
Second, audit demand. Any protocol that modifies its contracts to accommodate the new gas pricing rules needs those changes audited. The upgrade cycle will trigger a wave of re-audits.
Third, infrastructure modernization. The upgrade forces every wallet, every RPC provider, every indexer to update their software. That's a forcing function for modernization. The teams that embrace it will emerge with better products.
Fourth, L2 expansion. If L1 state becomes more expensive, the relative economics of L2s improve. The upgrade could accelerate the migration of state-heavy applications to L2s, strengthening the entire rollup ecosystem.
The Timeline
Let's be clear about the timeline, because it's tighter than most people realize.
- Now (June 2026): EIPs in Review. Replay test data public. Outreach ongoing.
- Q3 2026: Expected move to Last Call. Final parameter adjustments. Testnet validation.
- Q4 2026: Glamsterdam upgrade on mainnet.
That gives the ecosystem roughly 4-6 months to adapt. For protocols with complex contracts, that's not a lot of time.
The foundation has announced the Platåberget testnet, which will be the staging ground for the upgrade. Every protocol team should be testing their contracts against the new rules on that testnet. Every infrastructure provider should be validating their tooling. Every wallet should be checking gas estimation logic.
If you're a DeFi developer and you haven't started looking at the replay test dashboard, you're already behind.
The Verdict
Glamsterdam is the most consequential Ethereum upgrade since EIP-1559. It changes the fundamental economics of state on Ethereum, and it does so in a way that will break real, working applications.
The foundation has been unusually transparent about the risks. The replay test data is public. The outreach report is candid. The testnet is announced. This is how a mature protocol handles a difficult transition.
But transparency doesn't eliminate the risk. 2.7 million potentially broken transactions under EIP-8037. 3.0 million under EIP-8038. Those are real numbers with real consequences.
The chart doesn't lie, and neither does the replay test data.
The question isn't whether Glamsterdam will break things. It will. The question is whether the ecosystem can adapt fast enough to minimize the damage.
The answer to that question will be written in the migration code, the testnet deployments, and the protocol upgrades between now and Q4 2026.
Speed is safety. And right now, the clock is ticking.
This analysis is based on public information available as of June 2026, including the Ethereum Foundation's replay test dashboard, EIP-8037 and EIP-8038 specifications, and the foundation's outreach report on ecosystem impact. This is not financial advice. Do your own research.
Key Watch Signals for Q3-Q4 2026:
- EIP Status Changes: Watch for EIP-8037/8038 moving from Review to Last Call. That's the signal that parameters are locked.
- Platåberget Testnet: Monitor whether major protocols successfully deploy and operate on the testnet. Failures there predict failures on mainnet.
- Protocol Adaptation Announcements: Across, CoW Protocol, 0x, and the ERC-4337 stack are the canaries. Their adaptation announcements will signal ecosystem readiness.
- Replay Test Dashboard Updates: The foundation is likely to update the dashboard as parameters change. A declining "potentially broken" count is a positive sign.
- Official Timeline: The fork date announcement will crystallize market expectations.
