The brief demonstrated real-time zero-knowledge proofs for Ethereum blocks, generating encrypted proofs in less than 12 seconds.
The brief SP1 “Hypercube” ZKVM generated evidence of Ethereum block 22309250 (143 transactions, 32 million gas) in 10.8 seconds, with internal benchmarks showing that 93% of the average 10,000 mainnet blocks can be proven in less than 12 seconds.
This achievement illustrates the technical leap of zero-knowledge space. A completely new proof system based on many polynomial polynomials, an optimized CUDA kernel for operations such as log-up GKR and sum-check, and a drop-down cloud architecture that spans hundreds of GPUs. Co-founder Uma Roy has characterized the milestones as a result of advances in encryption, hardware acceleration and engineering across distributed systems.
Roy commented,
“This is a ZK moon moment. The real-time Ethereum proof has landed.
If I had asked someone from ZK a year ago if this was possible, they would have said it was a literal moonshot. ”
Performance milestones narrow the latency gap between proof generation and block time, but Ethereum co-founder Vitalik Buterin outlined the remaining limits.
The current results represent the performance of the average case rather than the worst case. Buterin emphasized that the worst block proofs must remain within block time to prove in real time that they are suitable for Layer 1 security.
Furthermore, the SP1 proof has not been formal validated, with energy requirements hovering close to 100 kW per proof. He also pointed out that by expanding the layer-1 gas limit on Ethereum by an order of magnitude, it relies on further proof efficiency.
Community debate has emerged centering on decentralization and proof of capacity. As discussed in the Ethereum Magicians Forum, researcher Dankrad Odendaal argued to temporarily relax Prover’s hardware decentralization goals.
Odendaal noted that overhead proofs could be reduced by several orders of magnitude, allowing for further benefits through improved architecture or specialized hardware.
He proposed that proofs are reversible, unlike other areas of protocol scaling that carry a permanent burden. If scalability pushes prover capacity, the network could return to low gas limits without long-term state growth penalties.
Odendaal also said that proof generation is less susceptible to centralization risk than complete stateful nodes in parallel between distributed infrastructures. Even if single digits of overhead is not proven, spread across many machines allows delay targets to be achieved while maintaining a few sincere assumptions for system integrity.
Deploying SP1 in a real-time context is the culmination of both encryption innovation and infrastructure coordination.
The concise implementation extends to bare metal deployment and performance tuning distribution workloads, but it shows that energy and formal validation constraints remain on the trajectory, not on the endpoint.
Further reductions in latency and power, together with protocol-level integration, form the ability of Ethereum to lock the maximum trusted execution directly into the base layer.
