Ethereum just changed the cost of a basic send. Now wallets have to stop trusting 21,000.

Glamsterdam splits Ethereum's basic transfer into two cases.

According to EIP-2780, a value transfer to an existing account still costs 21,000 regular gas. A value transfer that creates a new account still has that 21,000 regular gas, plus 183,600 units of state gas. Same user action. Different resource bill.

That sounds small until you remember how much wallet software treats 21,000 as sacred. For years, 21,000 was not only the minimum for a plain ETH send. It was often treated as the maximum. Fee calculators, gas limit defaults, service checks, and backend alerts all learned that habit.

Glamsterdam makes that habit wrong. The code does what the announcement says, plus one important thing: it charges the permanent database growth separately. If the recipient already exists, no new account leaf gets added. If the recipient does not exist, Ethereum has to carry that account forever.

EIP-8037 gives the math. It prices state at 1,530 gas per byte. It counts a new account as 120 state bytes. Multiply those and you get 183,600 state gas for one new account. That is not a vibes number. It is a pricing rule for durable data.

At the contract level, this is a state-growth problem. The old 21,000 rule mixed two paths that were never the same. Sending ETH to an existing wallet touches and updates known state. Sending ETH to a fresh wallet creates new state. The protocol is now saying those should not pay the same bill.

If you're building a wallet, this is where your bag meets infrastructure. A send screen that assumes 21,000 can still look fine in the common case. Then it fails on the edge case users care about: sending ETH to a brand-new address. And that's the catch.

The tradeoff is clear. Ethereum gets more honest pricing for permanent state growth. Wallets lose a simple constant they could hard-code everywhere. This is like replacing a global default with a branch that checks whether a record already exists. Annoying, but more correct.

The planned upgrade also sits inside a bigger scaling target. Ethereum.org says the state creation gas increase is meant to keep growth near a safe and predictable 120 GiB per year. The Ethereum Foundation's Soldogn recap also names a post-Glamsterdam gas limit floor target of 200M. Bigger blocks need better pricing, or the database becomes the quiet bottleneck.

There is one more implementation detail. EIP-2780 keeps transaction validity tied to state-independent intrinsic gas. State-dependent charges happen later, at runtime. That matters because block builders should not need to read the current state just to decide whether a transaction is valid.

So no, Glamsterdam does not break the basic ETH transfer. It breaks the shortcut around it. Read the PR diff, not the thread. The new rule is simple: 21,000 is still the existing-account path, for now. It is no longer the whole transfer model.