Vitalik Buterin has assigned a 60% probability that SNARKs, fully homomorphic encryption, and indistinguishability obfuscation will reach sub-10x overhead, a computational efficiency threshold he frames as the point where these cryptographic tools become practical for everyday blockchain use. The estimate, posted to X in 2026, reflects the Ethereum co-founder's assessment of how quickly the field's most powerful privacy and scaling technologies can shed the computational burden that has confined them to research environments and niche deployments.

What Does "Sub-10x Overhead" Actually Mean for Crypto?

The overhead ratio measures the computational cost of running a calculation under cryptographic protection compared to running it in plaintext. If a computation takes one second normally but 10 seconds under cryptographic protection, that is a 10x overhead. Buterin's 60% estimate targets a future where this ratio falls below 10x, making these powerful cryptographic primitives economically viable for production systems.

The significance lies in what becomes possible at that threshold. Zero-knowledge proofs (ZK proofs) are already deployed in production rollups, which are scaling solutions that batch thousands of transactions offchain and settle them to Ethereum with a single validity proof. Fully homomorphic encryption (FHE) remains far from practical, with current implementations running 100 to 1,000 times slower than plaintext operations. Indistinguishability obfuscation (iO), the most theoretically powerful but least mature of the three, remains largely confined to academic research.

Buterin has emphasized that researchers should report overhead as a ratio rather than absolute time, because ratios are "more hardware-independent" and produce "very informative numbers." This methodological preference normalizes progress across different hardware platforms, making it possible to compare a SNARK prover running on a 2026 server with an FHE scheme running on specialized silicon.

How Close Are We to Reaching This Milestone?

The distance to sub-10x overhead varies dramatically across the three cryptographic primitives. SNARKs are the most mature and already approaching the target for certain workloads. Research on block-wide signature aggregation indicates that the additional virtual machine layer used by Ethereum adds about a 10x proving performance overhead compared to a bare circuit, suggesting SNARKs are already at or near the sub-10x threshold for efficiently designed applications.

Fully homomorphic encryption sits roughly two orders of magnitude further out. Duality Technologies reports that FHE performance in 2026 is 1,000x to 10,000x faster than implementations from five years ago, but the absolute overhead remains severe. Hypersphere's analysis puts operations on FHE-encrypted data at 100 to 1,000 times slower than plaintext, meaning FHE still has substantial ground to cover.

Indistinguishability obfuscation represents the most aggressive part of Buterin's claim. Current iO schemes are, by the admission of researchers and Buterin himself, nowhere near practical. The overhead is not measured in multiples of plaintext execution but in orders of magnitude that make even toy programs prohibitively expensive. The gap between iO's current state and sub-10x overhead is the largest of the three primitives.

A 2026 arXiv paper on ZK-Tracer, a specialized hardware accelerator for zero-knowledge SNARKs, frames the remaining challenge as the "prohibitive computational overhead" of proof generation. The existence of hardware accelerators designed specifically to cut SNARK overhead indicates that the field is attacking the problem from both algorithmic and hardware directions simultaneously.

Which Applications Would Benefit First From Lower Overhead?

The practical implications of reaching sub-10x overhead extend across multiple layers of blockchain infrastructure and application design. Several use cases stand to benefit immediately once these efficiency targets are met:

• ZK Rollups: These scaling solutions are already in production on networks like zkSync, Polygon zkEVM, Scroll, and Starknet. Lower SNARK overhead translates directly into cheaper proof generation, which reduces transaction fees for users and accelerates finality for rollup batches.
• Confidential Smart Contracts: Fhenix has proposed an architecture for scaling confidential smart contracts on Ethereum using FHE. Sub-10x FHE overhead would make contracts that compute on encrypted data without revealing inputs economically viable for the first time. Current overhead makes such contracts a research curiosity; sub-10x makes them a product.
• Private Voting and Sealed-Bid Auctions: Buterin's endorsement of Interfold, a privacy protocol combining zero-knowledge SNARKs with FHE, points to a third application class. The protocol's viability depends on both primitives reaching overhead levels that make the combined cost acceptable for on-chain governance and auction mechanisms.

Why the 60% Probability Matters Without a Deadline

Buterin's 60% estimate is notably not tied to a specific calendar deadline. The framing suggests a directional bet on the trajectory of research rather than a commitment to a shipping date. This ambiguity matters for anyone trying to convert the estimate into a roadmap. A 60% probability without a time horizon is a statement about the likelihood that the overhead ratio will eventually fall below 10x, not a prediction of when that will occur.

The estimate applies jointly to all three primitives, SNARKs, FHE, and iO, rather than to each individually. That joint framing is significant because SNARKs are already deployed in production rollups, FHE is approaching enterprise viability, and iO remains largely theoretical. Bundling them into a single probability implies Buterin sees a common trajectory in the underlying mathematics and engineering, even if the starting points differ by orders of magnitude.

Buterin has published a technical series on program obfuscation, calling indistinguishability obfuscation "the most powerful idea in cryptography" while simultaneously warning that it remains "nowhere near ready" for production use. The 60% overhead estimate functions as a bridge between that caution and his optimism: the primitives are not ready, but the distance to readiness is measurable and, in his view, more likely than not to close.

The FHE Benchmarking Suite, launched to provide standardized workloads and metrics, is a signal that the field is maturing toward measurable progress. Without standardized benchmarks, overhead claims are difficult to compare across schemes and hardware. The suite's existence makes the sub-10x target falsifiable, which is a prerequisite for treating Buterin's 60% as a meaningful probability rather than a speculative guess.