Ethereum co-founder Vitalik Buterin has attached concrete numbers to a long-running question in cryptography: how close advanced privacy and verification tools are to ordinary computing costs. In a September 6 post, Buterin assigned a 60% probability that SNARKs, fully homomorphic encryption (FHE), and indistinguishability obfuscation (iO) will eventually operate below 10x computational overhead, measured through total energy use and amortized computing expenses. He also gave all three a 33% chance of approaching 1+ε overhead for average real-world computation.

Ethereum co-founder Vitalik Buterin said he believes there is a 60% chance that SNARKs, fully homomorphic encryption and indistinguishability obfuscation could eventually be implemented with single-digit… pic.twitter.com/kdtCcHwajq — Wu Blockchain (@WuBlockchain) September 6, 2026

That distinction matters because the timeline is narrower than the headline probability suggests. Buterin did not say all three technologies would cross the sub-10x threshold by 2030. Instead, he said at least one could reach single-digit overhead by the end of the decade, with SNARKs the most likely candidate. The framing builds on a longstanding problem in applied cryptography: all three techniques have been theoretically viable for years — FHE was first constructed by Craig Gentry in 2009 — but their computational cost has kept them confined mostly to research settings rather than production systems.

SNARKs Lead the Push Toward Sub-10x Computing Costs

SNARKs allow a system to prove that a computation was performed correctly without requiring every verifier to repeat the entire process. As a result, they have become central to Ethereum's zero-knowledge scaling model. Zero-knowledge rollups already use this approach, processing batches of transactions away from Ethereum's base layer and then submitting validity proofs to Mainnet, which allows the network to verify those transactions while preserving Ethereum's security guarantees.

Proof generation, however, remains a major constraint. Complex proofs still require substantial computing power, and some workloads depend on specialized hardware. Even so, Buterin said the efficiency gap is beginning to narrow. In August, he highlighted research showing that certain large language model inference workloads were approaching less than 10x proving overhead. He also pointed to specialized hash functions, where single-digit overhead has already been achieved.

As proving costs decline, the improvement could have direct implications for Ethereum's future architecture. The network's zkEVM roadmap envisions validators verifying proofs of entire blocks instead of independently replaying every transaction. Sufficiently efficient proof generation could therefore allow Ethereum to raise gas limits without requiring proportional increases in validator hardware, making lower-cost SNARKs relevant to both scalability and validator efficiency. The trend is also visible across the wider zero-knowledge ecosystem, where proving systems have seen successive optimization efforts aimed at reducing prover time and hardware dependence.

FHE and iO Face Higher Barriers

While SNARKs focus on verifying computation, FHE addresses a different challenge: privacy. It allows calculations to run directly on encrypted data without first revealing the underlying information. NIST describes FHE as a privacy-enhancing technology that can apply arbitrary functions to encrypted data without access to the secret decryption key. For blockchains, that capability could support private automated market makers, confidential lending markets, and sealed-bid auctions — applications identified directly in Ethereum's privacy roadmap. FHE, however, still carries significantly higher computational costs than ordinary plaintext processing, so reaching sub-10x overhead would represent a major step toward broader practical use.

Indistinguishability obfuscation presents an even tougher challenge. The technology aims to transform software while preserving its functionality, making equivalent obfuscated programs computationally indistinguishable. Recent research has strengthened iO's theoretical foundations, but its practical costs remain extremely high; Buterin has described traditional constructions as effectively "galactic" in computational expense. To reduce those costs, his recent work has explored approaches including diamond iO and local mixing. Diamond iO lowers the theoretical burden but remains impractical, while local mixing takes a different route whose security is still unproven.

Each technology nonetheless targets a distinct part of the broader cryptographic problem. Cheaper SNARKs could make verifiable computation more routine, FHE could expand private computation across shared data, and efficient iO could help protect the internal logic of executable software. For now, Buterin's 60% estimate remains a personal probability assessment rather than an Ethereum roadmap commitment.

Source: Blockonomi