Buterin: iO is cryptography’s hardest unsolved problem

On June 29, 2026, Ethereum co-founder Vitalik Buterin published a technical post that named indistinguishability obfuscation, or iO, as “the final boss of cryptography” and estimated that existing iO constructions would require runtimes longer than the age of the universe.

Buterin explained iO in plain terms: obfuscation hides a program’s internal logic so that two programs that do the same thing appear indistinguishable after obfuscation. The obfuscated program still runs and produces outputs, but it should reveal nothing about how it works, even to someone observing execution.

He linked iO to blockchain use cases. Because obfuscated programs can be copied and cannot by themselves enforce persistent state such as balances or transaction records, combining obfuscation with a blockchain could let the chain maintain state while obfuscated code enforces rules. He described a possible outcome as a “trustless trusted third party” and offered a secure, private voting system as an example that would not require a multisignature committee, a trusted setup or assumptions about some participants remaining honest.

Buterin reviewed the research history. Researchers have pursued provable obfuscation for roughly 20 years. A 2001 result showed a perfect general obfuscator is mathematically impossible, which shifted focus to iO as a more attainable goal. In recent years, cryptographers have produced constructions that achieve iO under specific security assumptions.

He identified practicality as the primary barrier. Current iO schemes are polynomial in theory but impose very large overhead in practice. The schemes stack multiple layers of advanced cryptographic tools, producing overhead factors Buterin cited above 10^10 and ciphertext sizes thousands of times larger than standard constructions. He estimated that expected runtimes on today’s schemes exceed the age of the universe, making them unsuitable for production.

Buterin outlined three research paths that could change that outcome: optimize the existing tower of constructions through engineering and tools such as artificial intelligence; adopt simpler, stronger cryptographic assumptions that yield faster schemes; or find a new approach to obfuscation that departs from the lattice-based methods common today. He wrote, “If we succeed in either path, the reward is high: there is a real sense in which we will have ‘solved cryptography’…” He also noted that current constructions remain far too slow and large for practical protocols, while saying recent theoretical progress offers a clearer research roadmap.