StarkWare launches new consumer-grade hardware ZK prover

This is a segment from the 0xResearch newsletter. To read full editions, subscribe.

---

More than a year after its initial announcement, StarkWare is today launching S-two (STARK Two) in public alpha.

S-two is a next-generation zero-knowledge STARK prover fast enough to run natively on consumer hardware. Developers will be able to start building with S-two today.

According to a press statement, S-two will be able to generate STARK proofs “directly on phones, laptops and browsers…enabling smooth mobile experiences, fast browser-based apps and new classes of client-side privacy and AI applications.”

A full Starknet integration is slated for later this year.

Various teams such as Kakarot, Nexus, ZAN, Giza, Herodotus and more have begun building with S-two.

Based on the team’s own benchmark tests, S-two claims to generate proofs up to 39x faster than Succinct’s SP1 and 28x faster than Risc Zero’s R0VM on specific workloads.

The history of zk proving

In the context of blockchains, a prover is the entity in a zero-knowledge system that produces the mathematical proof that transactions are executed correctly and in compliance with protocol rules.

Historically, proof generation was an expensive bottleneck. Rollups had to batch thousands of transactions into a single proof, accepting a delay of several minutes or hours before publishing the result to Ethereum.

To users, this meant limitations of cross-chain composability (a.k.a. synchronous composability) and a withdrawal time delay capped by proof-generation times. 

Recent breakthroughs in zk tech enable what Ethereum devs typically refer to as “real-time proving,” or the ability to compress proof-generation latency to mere seconds. Specifically, below 12 seconds — Ethereum’s block interval.

With real-time proving, rollups can finalize blocks at near-L1 speeds while still enjoying the quick throughput and cheaper fees of an L2.

S-two’s client-side proving will also mean users/rollups no longer have to rely on trusted and centralized prover clusters typically hosted on AWS or Google Cloud. 

With S-two, proofs will be generated locally on the client side and submitted directly to Ethereum, potentially making every browser a self-sovereign proving node and reducing liveness risks from single-hosted clusters.

A step towards the zk endgame

StarkWare is not alone in chasing single-digit-second proofs. 

Last week, Succinct unveiled SP1 Hypercube, a multilinear-polynomial zkVM that claims to be able to prove 93% of Ethereum mainnet blocks in just 10.3 seconds.

The announcement drew praise (and caution) from Vitalik Buterin, who tweeted that the milestone represented real-time average cases, but that “real-time worst case for safe L1 use” was still a challenge.

Where Succinct’s new zkVM demonstrated that real-time block proofs are feasible, StarkWare’s new S-two prover shows the consumer-grade hardware feasibility of real-time proving.

Just a few years ago, it was thought that real-time proving wouldn’t be ready anytime soon. 

“Run a zero-knowledge proof on your phone” was a mere punchline.

Yet, zero-knowledge technology continues to accelerate with continued breakthroughs.

Users will never notice the cryptography under the hood. What they will notice, however, is when bridging takes mere seconds instead of minutes and privacy comes baked into the wallet — all without compromising on trustlessness.

---

Get the news in your inbox. Explore Blockworks newsletters:

• The Breakdown: Decoding crypto and the markets. Daily.
• Empire: Crypto news and analysis to start your day.
• Forward Guidance: The intersection of crypto, macro and policy.
• 0xResearch: Alpha directly in your inbox.
• Lightspeed: All things Solana.
• The Drop: Apps, games, memes and more.
• Supply Shock: Bitcoin, bitcoin, bitcoin.