Ethereum co-founder Vitalik Buterin has assigned a 60% probability that three of cryptography’s most ambitious technologies — SNARKs, fully homomorphic encryption and indistinguishability obfuscation — can eventually operate at less than 10 times the cost of ordinary computation. Buterin outlined the forecast on September 6, describing it as an optimistic and “very-non-consensus” view of cryptography’s long-term development. His calculation defines overhead using total energy expenditure plus amortized computing costs rather than simply comparing execution time.Buterin went further with a second forecast: he sees a 33% probability that all three technologies can eventually approach 1+ε overhead for average real-world computation. At sufficient scale, that would mean the additional cost created by applying the cryptographic technology becomes arbitrarily small relative to performing the underlying computation normally.
SNARKs Could Reach Single-Digit Overhead This Decade
The three technologies solve different problems. SNARKs allow one party to produce a compact cryptographic proof demonstrating that a computation was performed correctly without requiring everyone else to repeat the entire calculation. They already play an important role in Ethereum’s scaling ecosystem, particularly in zero-knowledge rollups and emerging zkVM infrastructure. Fully homomorphic encryption, or FHE, tackles privacy differently. It allows computations to be performed directly on encrypted information without first decrypting the underlying data.Indistinguishability obfuscation, known as iO, is substantially more experimental. It aims to transform programs so their internal implementation cannot meaningfully be distinguished while preserving their functionality. Buterin has previously described iO as cryptography’s “final boss” because of both its theoretical power and extreme computational requirements. SNARKs are considerably further along. Buterin said there is a good chance at least one of the three technologies reaches single-digit overhead by the end of the decade, with SNARKs the most likely candidate. He pointed out that specialized hash functions and some large-language-model inference workloads have already reached comparable efficiency levels.
Cheaper Cryptography Could Reshape Ethereum
Reducing SNARK overhead has direct implications for Ethereum. Ethereum’s longer-term architecture increasingly relies on cryptographic proofs to verify computation more efficiently. Under a proof-based model, validators could verify a succinct proof showing that an Ethereum block was executed correctly rather than independently repeating every computation contained inside it. That becomes increasingly valuable as Ethereum raises its gas limit and processes more transactions. If proof generation remains computationally expensive, higher throughput can simply transfer the hardware burden from validators toward specialized proving infrastructure. Bringing overhead below 10x would make that trade-off substantially less severe.FHE could have similarly important consequences for blockchain privacy. Smart contracts currently operate primarily on publicly visible information. Practical FHE could allow applications to process encrypted balances, bids, votes or other sensitive data without exposing the underlying information during computation. iO represents an even more ambitious possibility, potentially allowing developers to conceal program logic itself while preserving verifiable functionality. However, Buterin’s probabilities should not be interpreted as an Ethereum roadmap or engineering commitment. The 60% estimate concerns what he believes may ultimately be technologically possible, not a scheduled protocol upgrade. Nor does his statement predict that SNARKs, FHE and iO will all reach sub-10x overhead by 2030.Only the narrower SNARK milestone is associated with the end of the decade. The distinction matters because FHE remains considerably more computationally intensive, while practical iO is much further from deployment. Buterin’s forecast is therefore less a prediction about Ethereum’s next upgrade than a bet on the trajectory of cryptography itself. If the 60% scenario materializes, techniques currently reserved for specialized blockchain, privacy and security applications could become cheap enough to sit underneath ordinary computing. And if his more aggressive 33% scenario proves correct, the long-term cost of making computation cryptographically private or verifiable could eventually become surprisingly close to the cost of performing the computation without those protections at all.