Zero Computing is the first dedicated compute platform for generating zero-knowledge proofs.
Introduction
Today, there remains a lack of cloud services and computational infrastructure specifically designed for zero-knowledge proofs. The prevailing practice for generating proofs involves consuming general-purpose instances on the cloud which, while scalable, fails to address the niche demands of generating zero-knowledge proofs. There’s already a vast network of latent hardware available in the market waiting to be consumed. It’s just that nobody knows how to do it - yet. Which platforms today are providing tailored cloud infrastructure to megawatts of dedicated zero-knowledge compute resources across a distributed network of data centers?
Optimizing the Proof Supply Chain with Zero Computing
Zero Computing is a platform for orchestrating vast amounts of compute to make zero-knowledge proof generation fast and efficient. We specialize in provisioning highly-customized instances tailored down to the specificities of individual proof requests, leading to the most competitive proof costs on the market. By combining this custom virtualization layer with a global scheduler to maximize the cross-network efficiency of resources, we provide an on-demand service that reduces costs for teams while being able to meet any shape of demand. Some of our accomplishments to date include:
Scroll zkEVM: 13 minute proving time (network average is 20 minutes)
RISC Zero zkVM: Proving time of 46 seconds per 4 million clock cycles (20% faster than the fastest reported benchmarks)
Cloud Services: 70% lower costs for equivalent performance
Zero Computing serves as a bridge between proof demand and supply. We act as an abstraction layer, receiving proof requests from applications, protocols, and proof markets. These requests are orchestrated by Zero Computing to target specialized proving hardware in order to meet specified constraints such as latency, cost, and decentralization. Proof requests can be submitted through nodes in proving networks and marketplaces, or directly from protocols and applications.
How It Works
Proof Requests: Customers provide proof specifications, requirements, and inputs. These inputs are combined into proof requests and sent to Zero Computing.
Data Center Network Info: Zero Computing gathers data center network information, including IP addresses, storage, GPU, and CPU capacities.
Global Scheduler: Zero Computing provisions dedicated resources and coordinates the downstream targets for where the proof request should be dispatched to.
File Uploader: Proof input files are uploaded to a shared storage accessible by multiple data centers for high-availability and fault-tolerance.
Deployment Manifest: A workload detailing the instructions for proof generation, resource requirements, and volume mounts is created and sent to the target data centers.
Proof Generation and Delivery: Data centers use the deployment manifest to physically generate the proof. Once the proof is generated, the workflow concludes with the delivery of the proof to the client.
Our Vision
This is our first step in building an orchestration platform for complex computational tasks that enable new possibilities for privacy, security, and economic possibilities. Our vision is to empower the next wave of computational technologies by providing the infrastructure and tools necessary to harness the full potential of emerging and latent hardware resources. We have already started expanding our horizons across other technologies such as fully homomorphic encryption (FHE), multi-party computation (MPC), AI inference and AI agent orchestration, and have only scratched the surface of what is possible. We are also building native support for quantum-safe SNARKs and STARKs, inline with Ethereum’s roadmap. We believe the future holds a myriad of complex computational requirements that will need to be orchestrated cost-effectively across various types of hardware and architectures - and we aim to do just that.
Let's Get In Touch!
We built Zero Computing to help teams focus on what they do best. Build. By leveraging our orchestrated network of computational supply, we aim to make proofs readily accessible and highly economic. If your team is looking for a tailored solution where general-purpose cloud services fall short, come join the largest network of zero-knowledge data centers in the world.
Check out our website for more here. And join the conversation and our community on Twitter, we’d love to get in touch!
Special thanks to the teams and individuals who have provided us with their help and feedback along the way. We are grateful to the team at Scroll for their discussions and direction regarding prover dynamics, RISC Zero for their benchmarking help and guidance over proving costs, Holonym for providing us with our first circuits, Ingonyama for their help in integrating ICICLE-enabled GPU acceleration, RockX for their contributions to our economic modelling, and the team at Strobe for their invaluable strategic input.
