The replay test data is out. 929,731,274 transactions were pushed through the proposed gas repricing framework. The result: 2,687,652 transactions face potential breakage under EIP-8037. A further 3,036,537 under EIP-8038. Hashes don't lie. Wallets do. And the wallets say the cost of creating a new account is about to increase 7.3x.
This isn't a hypothetical. This is the Glamsterdam upgrade, scheduled for Q4 2026, and it is engineering a fundamental shift in how Ethereum prices state. While the marketing focuses on the promise of tripling network speed, the forensic analysis reveals a different story—one of forced migration, broken infrastructure, and a potential crisis for immutable contracts.
Context: The State-Gas Paradigm
Ethereum's core developers have long grappled with the tension between execution throughput and state bloat. The network's global state—the collection of all account balances, contract storage, and nonces—grows with every block. Unchecked, this growth increases the hardware requirements for running a full node, slowly eroding decentralization.
The solution proposed in Glamsterdam is a two-pronged attack: EIP-8037 and EIP-8038. The former introduces a separate state-gas dimension, distinct from execution gas. Every new byte of state created will cost a uniform 1,530 gas. The latter adjusts the costs of accessing existing state based on benchmark tests against the March 2026 mainnet snapshot. The goal is to cap annual state growth at 120 GiB, down from the current trajectory that threatens node operator viability.
The logic is sound. It aligns gas fees with actual resource consumption. But the execution details are where the silent breakage lives.
Core: The On-Chain Evidence Chain
Let's follow the liquidity of these specific operations. The numbers are stark:
- Creating a new account: Current cost 25,000 gas. Proposed state-gas cost: 183,600 gas. That's a 7.3x increase.
- Creating a new storage slot: Current cost 20,000 gas. Proposed cost: 97,920 gas. A 4.9x multiplier.
- Deploying a 24 KiB contract with a new account: Current cost ~4.95M gas. Proposed cost: ~37.78M gas. A 7.6x increase.
These aren't marginal adjustments. These are structural changes to the economics of contract deployment and account abstraction.
My audit experience from the 2020 DeFi Summer, where I mapped the yield fragmentation on Uniswap v2, taught me to look for the causal links hidden in volume spikes. Here, the causal link is clear. The Ethereum Foundation's own replay testing, conducted across a period from December 2024 to June 2026, identified the primary casualties. The public outreach report explicitly names eth-infinitism's ERC-4337 EntryPoint and the broader smart account stack as high-priority risk. They also flagged repeated failures in Across, Socket/Bungee, CoW Protocol, and 0x.
The counterfactual failures—transactions that work today but would fail under the new rules—are concentrated in specific DeFi operations. These protocols rely on assumptions about gas costs embedded in their logic. When the price of a state operation jumps 5-7x, the gasleft() checks and call-data validations break.
The infrastructure layer faces a systemic adaptation squeeze. Wallets, RPC providers, indexers, and node tools must all be updated to understand the new gas dimension. The announcement for the Platåberget testnet even warns that old tools might not function correctly. This is not a simple upgrade; it is a coordinated migration across the entire ecosystem.
Contrarian: Correlation is Not Causation—and Neither is the Replay
The immediate reaction is to scream "network upgrade, mass failure." But the contrarian angle requires a closer look at the methodology. The replay test assumes a "worst-case" 10x gas limit increase. This is an incentive-based extrapolation, not a measured outcome. The 387 GiB state growth projection is based on a non-linear response to the gas limit cap, and it's an estimate, not a guarantee.
More importantly, the counterfactual failures are just that—counterfactual. Under current rules, these transactions succeed. Under new rules, they might fail. But the test assumes no code changes by the affected protocols. The Ethereum Foundation has already issued public outreach. Teams like CoW Protocol and 0x are aware. The question is whether the long tail of smaller, less-maintained protocols will adapt in time.
There is also a subtle narrative trap here. Tripling throughput is an engineering support target, not a guaranteed capacity increase. The market may interpret "3x speed" as "3x cheaper and faster transactions immediately." That's not what this is. This is a resource pricing correction. It's a move to ensure the network's survival, not to make it a Solana competitor. Follow the liquidity, not the narrative. The liquidity is being repriced, and the narrative is just noise.
Takeaway: The Migration Market
The risk is real, but the opportunity is clearer. The upgrade will force a wave of contract migrations. Immutable contracts—those with no upgrade mechanism—cannot be patched. They will require entirely new implementations, new EntryPoints, and new user migration paths. This creates a demand for migration tooling that didn't exist before. The teams that build those tools will capture value.
But the immediate signal is defensive. If you are a developer with contracts relying on gasleft() or specific state-access costs, your code is already broken. If you are a user of a smart account or a bridge, watch for upgrade announcements. The Platåberget testnet is the canary in the coal mine. Watch its stability. Watch the public dashboard for the replay test numbers to drop as protocols adapt.
The real question isn't whether Glamsterdam ships. It's whether the ecosystem can adapt faster than the upgrade breaks it. The data suggests we're in for a rough transition. On-chain truth > Twitter narrative. The truth is that the cost of state is going up, and the cost of ignoring this data will be much higher.