Privacy Made Practical: Introducing Confidential Proving Delegation
Today, privacy is being reborn.
Until now, we could have only one of the two: privacy or seamless UX. This is changing for the first time with Fermah’s new product – Confidential Proving Delegation!
The task of generating a zero-knowledge proof (ZKP) is resource-intensive. This gives generating proofs on light local devices a terrible UX – imagine running out of memory and having the app you’re using crashing every time you generate a proof.
Until now, projects faced an impossible choice: compromise user experience with local proof generation or sacrifice privacy by exposing sensitive data.
Confidential Proving Delegation enables proving to be delegated to powerful machines while keeping private data completely confidential – finally delivering the UX of centralized systems that users expect with the privacy guarantees of zero-knowledge proofs.
At Fermah, we’re laser-focused on making ZK the substrate of all computation. CPD is a crucial technological breakthrough that brings us one step closer to realizing our mission.
Check out the video we put together to introduce the world to CPD!
The Problem
There are two main categories of applications using zero-knowledge proof technology:
- Scalability applications: that leverage the succinct verifiability of ZKPs. ZK rollups like ZKsync and Scroll are great examples of this.
- Privacy applications: that utilize ZKPs’ privacy, or rather the zero-knowledge, property. Examples here include projects that enable private transactions, like Labyrinth, Aztec, or Aleo; decentralized identity projects, like World and Privado ID (formerly Poylgon ID); zkTLS projects, like Reclaim; ZKML projects, like EZKL; and application-specific use cases, like ZK Email.
Generating proofs is a resource-intensive task. For scalability applications, it is straightforward to delegate proving to a proof market like Fermah for cheap and performant proofs. Two of the largest ZK rollups, ZKsync and Scroll, are already doing this.
However, when you are proving sensitive data, delegation is not that straightforward. This is why many zero-knowledge applications are forced to have user devices generate proofs. User devices are not built to carry out the heavy computations needed to generate a ZKP. This means that running such computations on user devices can cause the apps to crash or run out of memory, leading to a horrible UX.
This limitation has been a bottleneck for the widespread adoption of privacy applications, particularly in consumer-facing scenarios where device capabilities are limited and computational demands are too high.
Vitalik Buterin corroborated this in a recent post. To paraphrase, user devices are typically incapable of proving any sophisticated computations, and this is the reason why tools like zkVMs haven’t reached mass adoption to enable privacy applications.
Without a solution to this core problem, ZK technology remains locked away from the very users it was designed to protect.
Until now.
From Impossible to Inevitable: Introducing CPD
Confidential Proving Delegation (CPD) resolves this fundamental tension.
It enables projects to delegate proving to the powerful hardware on the Fermah Network while maintaining complete privacy of the underlying data. CPD ensures that no single party can access the sensitive information being proved by incorporating Multiparty Computation (MPC) across prover nodes. Any hypothetical collusion can be mitigated by using Trusted Execution Environments (TEEs) with remote attestation.
This transforms what was previously impossible into something inevitable: ZK applications that can finally achieve both privacy and performance, without compromising on security.
By eliminating the traditional trade-off between security and UX, CPD opens up entirely new possibilities for mainstream applications that require both confidentiality and real-world performance.
What CPD Offers
With CPD, you get two primary benefits:
Confidentiality: CPD introduces another cryptographic tool to protect privacy: Multiparty Computation (MPC). MPC helps perform computations without revealing sensitive information to any individual party. In CPD, the ZKP is generated by running the MPC protocol between multiple prover nodes. The security property of the MPC protocol helps ensure that no single party can uncover sensitive information, on top of the strong security already provided by TEEs.
UX on par with centralized systems: The work of a client ends as soon as it delegates the proving task. By combining beefy GPUs with powerful orchestration mechanisms, Fermah enables fast proving, leading to a great UX.
The implications of this technology reach far beyond existing applications though.
CPD enables entirely new categories of applications that were previously impractical. From private DeFi that protects user trading data to scalable identity solutions that keep credentials confidential, CPD makes privacy practical at scale.
Inside CPD: The Architecture of Practical Privacy
There are two components to the architecture that enable combining great UX with privacy. While one is the CPD protocol itself, the other is the Fermah Network.
The Fermah Network’s Architecture
Fermah is a universal proof market.
Specifically, Fermah is a two-sided marketplace. The supply side consists of machines such as GPUs and FPGAs. It is bootstrapped with EigenLayer Operators. On the other side, the demand involves anyone submitting a proving job.
The demand and supply sides are aligned with the Fermah Matchmaker. The Matchmaker ensures efficient allocation of resources while maintaining competitive pricing. It can also handle any workflow, regardless of complexity.
Intelligent orchestration optimizes the utilization rate of machines, ensuring fast proving. The network’s lean mechanism design minimizes the cost of generating proofs.
The sum of these components enables universal proving; Fermah can generate proofs for any proof system, chain, or VM.
The CPD Architecture
CPD builds on this foundation of intelligent workflow orchestration with a protocol for proving delegation while protecting the confidentiality of witness data. Specifically, CPD is a protocol that achieves confidentiality of witness data by information-theoretically secure secret shares across multiple prover nodes.
This is achieved using an MPC protocol. The MPC protocol ensures that no secret share reveals any information about witness data. Then, the protocol involves collaborative computation of the proof by these prover nodes so that no single party can uncover any information about the witness data.
The steps in this process are as follows:
- The Matchmaker requests MPC Nodes for the Client
- The Client sends its input or witness shares to these MPC Nodes over secure channels
- Those MPC nodes compute shares of the proof and send them to the Aggregator Node
- The Aggregator Node constructs the proof and sends it to the Matchmaker
- And finally, the Matchmaker handles the payment and returns the proof to the client
Security is maintained through careful protocol design ensuring that no single party can reconstruct the private data. The system establishes secure channels and even in scenarios where multiple parties attempt to collude and somehow manage to bypass the TEEs, the system can maintain its privacy guarantees as long as at least one party remains honest.
Acknowledgements
CPD is utilizing two MPC sources. One is from TACEO, and the other is based on Arkworks.
Catalyzing the Privacy Revolution
CPD represents a fundamental advancement in making privacy practical for mainstream applications.
It removes the last major barrier to widespread adoption of ZK in applications, enabling a new generation of products that can offer both privacy and performance.
For builders, this means the ability to implement privacy features without compromising user experience.
For users, it means access to applications that protect their privacy while maintaining the performance that users expect from modern software.
With CPD, Privacy becomes the default.
Join the Revolution
Integrating with CPD is straightforward.
Fill out our form, and our team will reach out to set you up. We'll create a dedicated Telegram chat to ensure you have direct access to our support team whenever you need it.
Come join us to help bring moon math to the masses.
