TL;DR: Solana’s core developers just filed a proposal to permanently kill “durable nonces,” the transaction feature that attackers exploited to drain more than $270 million from the Drift protocol back in April. The draft, submitted November 5 by an engineer at Anza, won’t take effect before May 2027 at the earliest. That gives custodians and exchanges that depend on the feature at least six months to rebuild how they sign and send transactions, and it reopens an old wound that Solana thought it had already closed with a security overhaul back in the spring.

Durable nonces sound like obscure plumbing, and for most Solana users they are. A normal transaction on the network carries a timestamp called a blockhash that expires after roughly 30 seconds, now that Solana runs on 200-millisecond blocks. That works fine if you’re swapping tokens from a hot wallet. It falls apart if you’re a custodian signing transactions on an air-gapped machine that isn’t connected to the internet, where getting a signature out the door can take minutes, not seconds.
Durable nonces solved that problem by swapping the expiring blockhash for one that never goes stale. Sign a transaction today, and it can sit around and still execute next month. Exchanges and custodians built entire signing workflows around that convenience.
Then, on April 1, attackers used exactly that convenience against Drift. According to Coindesk’s reporting, the attackers got approvals signed in advance, let those transactions sit untouched for more than a week, and then fired them off to drain at least $270 million from the protocol. Drift later pinned the attack on a North Korean state-affiliated group, the same playbook that’s been showing up across crypto all year, from the record-setting Binance breach to the leadership shakeup that followed it.
Solana’s response in the immediate aftermath was a security overhaul announced just days later. It clearly wasn’t the end of the conversation. The new proposal, filed as SIMD-0723 on GitHub by Anza engineer Gabe Rodriguez, goes further than a patch. It would require every transaction on the network to carry a fresh, expiring blockhash after activation, eliminating the durable-nonce path entirely rather than just hardening it.
Why it took this long
The proposal’s own justification isn’t just the Drift hack. It points to a history of bugs tied to the feature, including a 2022 incident where a failed transaction got processed through both the blockhash and nonce paths at once and halted the entire Solana mainnet for about 4.5 hours. Durable nonces, in other words, have been a recurring headache well before they became a $270 million one.
What makes the fix politically tricky is who actually uses the feature. Data from researcher Max Resnick, cited in the proposal’s GitHub discussion, shows durable nonce transactions made up only about 3% of Solana’s traffic during one measured period in June, yet accounted for roughly 30% of all priority fees paid on the network. Just 555 accounts were responsible for 97.5% of that volume. That’s a small, concentrated, and almost certainly well-resourced group, the custodians and institutional players the feature was built for in the first place, now facing a forced migration they didn’t ask for.
The draft doesn’t pretend this is painless. It calls out custodians and exchanges that rely on air-gapped signing by name as the group that will need new tooling. Two replacement approaches are already on the table: Anza’s own Programmatic Signer and a competing option from Blueshift called Vector, both of which check an offline signature through an onchain program instead of a nonce that lives forever. The proposal also flags a subtler risk in the handoff itself, warning that a failed execution can leave an authorization still usable, so a failed submission isn’t the same thing as a revocation. Anyone migrating signing infrastructure under that proposal needs to understand that distinction before they ship it.
Nothing is final yet
This is still a draft, not a done deal. SIMD-0723 needs sign-off from at least one engineer at Anza and one at Firedancer, the rival validator client built by Jump Trading, before it can move toward activation. The Solana Foundation’s own deprecation tracker currently lists the cutoff as “no date set,” and the six-month migration clock the Foundation promised back in October only starts once a replacement is actually ready to use, which could easily push the real deadline past May 2027 rather than hitting it.
That ambiguity matters more than it might look. Solana has spent 2026 trying to convince institutions and custodians that it’s a serious, bank-grade settlement layer, not just a venue for memecoin speculation. A sprawling, repeatedly-bugged feature that institutional players specifically rely on, and that just got blamed for one of the year’s bigger exploits, is exactly the kind of thing that undermines that pitch if it isn’t handled carefully. Killing durable nonces outright is the clean, defensible answer from a security standpoint. Forcing hundreds of custodial accounts to rebuild signing infrastructure on a vague timeline is the kind of thing that makes institutional partners nervous in a different way.
It’s also a reminder that crypto’s biggest exploits rarely get fixed with a single patch. The industry’s response to this year’s wave of breaches, from Gnosis Pay’s card-data exposure to the Web3Auth flaw that exposed Keplr wallet keys, has mostly played out the same way: an initial fix, followed months later by a deeper structural change once someone finally traces the root cause all the way to the feature that made the exploit possible in the first place. Solana is now doing that second step, seven months after the first one.
The bottom line
Retiring durable nonces is the right call on security grounds alone, a feature that’s caused a mainnet-halting bug and a quarter-billion-dollar hack has no business sticking around just because a few hundred accounts find it convenient. But “no date set” and a six-month clock that hasn’t started yet isn’t a migration plan, it’s a placeholder. If Solana wants custodians and exchanges to trust it with serious money, it needs to turn this proposal into a real deadline with real tooling behind it, not another GitHub thread that quietly stalls once the news cycle moves on.



