Solana co-founder Anatoly Yakovenko has warned that the most pressing risk around post-quantum cryptography may not be quantum computers themselves, but the possibility that AI could expose weaknesses in the signature schemes designed to defend against them. His comments add a sharper edge to Solana’s recent quantum-readiness push, which has centered on Falcon signatures, migration planning and wallet-level resilience.
Solana’s Post-Quantum Plan Gets New ScrutinyThe more consequential remark came shortly after, when Yakovenko framed the problem less as a simple migration from today’s cryptography to post-quantum signatures, and more as a security-design issue with unresolved unknowns.
“I think the biggest risk is that pqc signature schemes will get broken by ai,” Yakovenko wrote. “We don’t know all the implementation footguns even, let alone the math footguns. So we need to support 2/3 wallets for them. @fusewallet or ideally natively with PDAs in the tx processor.”
The point is notable because Solana’s official messaging on quantum readiness has been broadly confident. In an April 27 developer post, Solana said quantum computing remains “years away” and that, if the threat materializes, migration work is “well-researched, understood, and ready to deploy.” The post described a roadmap built around continued research, adoption of a post-quantum scheme for new wallets if needed, and migration of existing wallets to the selected scheme.
That distinction matters for builders. The quantum-readiness debate often treats post-quantum signatures as the endpoint: once a chain can verify Falcon or a similar scheme efficiently, the network has a path forward.
Yakovenko’s comments suggest the safer architecture may be one that avoids dependence on one scheme, even after migration. His preference for “2/3 different signature schemes” indicates a defense-in-depth model, where wallets or transaction processors could require threshold approval across multiple cryptographic primitives. Michael Egorov, founder of Curve Finance, asked whether “proper formal verification” might help address the concern. Yakovenko’s reply was cautious: “If we know exactly what to verify. I’d still like 2/3 different signature schemes.”
That response captures the unresolved part of the debate. Formal verification can reduce implementation risk when the target properties are precisely defined. Yakovenko’s concern is that the industry may not yet know all the relevant failure modes, especially if AI systems become better at finding edge cases, deployment flaws or deeper mathematical weaknesses in post-quantum constructions.
At press time, SOL traded at $84.03.
