Like with my last post, I have a theory and while this might sound cliche, I think a solution would just be making ZK proofs easy fast and cheap for any project to use.
Sometime I really wonder how expensive ZK proof is.
I've seen many project about ZK proof right now, so many of them requires huge GPU running like 5090 and above.
My concern is how much energy consumed to prove a billion cycle and so on, so we can judge the effectiveness. I also want to see whether it will be sustaining.
PROVE is the token used to run ZK on the Succinct Prover Network. They use SP1 HyperCube and launched their journey on August 5th. They claim to have proven 93% of Ethereum blocks in under 12 seconds.
You can learn more here:
https://blog.succinct.xyz/sp1-hypercube/SP1 Hypercube introduces end-to-end improvements across the entire prover stack, from efficient execution of individual RISC-V instructions to low-latency recursion. These optimizations enable real-time proving on Ethereum while requiring significantly fewer GPUs than SP1 Turbo (roughly 2x). From our benchmarks, a cluster capable of real-time proving > 90% of mainnet blocks with SP1 Hypercube requires ~160 4090 GPUs and can be built for ~$300-400k. With more cost-efficient hardware, we estimate that the cost to set up a cluster could reach ~$100k, and with our open-source prover and cluster implementation (to be released after the SP1 Hypercube audit finishes), we expect this to make it feasible for anyone to run their own real-time Ethereum prover.