Nexus Exclusive Interview | Jens Groth Reveals How to Build the Future of Globally Verifiable Computation?

Respondent: Jens Groth, Chief Scientist of Nexus

Interview and article by: Alex, OpenBuild Content Team

In the current era where AI, big data, and Web3 are accelerating their integration, how to verify the authenticity of large-scale computations has become a core technical challenge facing the digital world. Nexus is promoting a cutting-edge exploration - building a globally collaborative verifiable computing infrastructure that integrates the zero-knowledge virtual machine zkVM[1] with a distributed prover network[2], creating a trusted "supercomputer."

In this exclusive interview, OpenBuild has an in-depth conversation with Professor Jens Groth[3], Chief Scientist of Nexus and an authority on cryptography, and discusses key topics such as the technical evolution of zkVM, the design of distributed computing architecture, and the trusted execution path in AI scenarios. He also shared how Nexus balances performance and scalability while maintaining privacy, providing the underlying support for building a verifiable digital civilization.

As an important pioneer in the field of zero-knowledge proofs, Professor Jens Groth proposed the widely used Groth16[4] system in 2016, and his research laid a solid foundation for the development of zk computation today.

The construction logic of a verifiable world

OpenBuild: Can you summarize the vision and mission of Nexus in one sentence?

Jens Groth: Nexus is a platform for building the underlying infrastructure for a "verifiable world."

We believe that all important digital activities in the future—whether it is data generation, AI decision-making, or on-chain transactions—will require verifiability assurance, and Nexus is building a technological foundation for this.

The technological evolution of zkVM 3.0

OpenBuild: What are the important technical iterations of zkVM 3.0 compared to the previous two versions?

Jens Groth: This is a comprehensive rewrite from the underlying architecture to execution performance.

Nexus zkVM 3.0[5] targets the RISC-V[6] instruction set, featuring powerful modularity, scalability, and supported by formal specifications to ensure safety and standardization.

In addition, we have introduced a redesigned guest runtime, an efficient memory checking mechanism, and provided backend proof capabilities through StarkWare[7]'s Stwo prover[8]. The end result is a speed increase of 1000 times compared to versions 1.0 and 2.0, while being easier to use.

A distributed computing power network that anyone can participate in.

OpenBuild: How does the distributed prover network of Nexus operate? Can ordinary users participate?

Jens Groth: Absolutely. We have lowered the computing power threshold, allowing users to connect to the network with just one click using a laptop or even a mobile phone, contributing their idle computing resources. Each device that connects enhances the overall computing power of the network.

We are different from the traditional blockchain "miner model"; instead, we activate the potential of global devices. Users can contribute computing power to earn the incentive points NEX Points[9] provided by Nexus.

Developer-friendly verifiable computing toolchain

OpenBuild: How does Nexus support developers in building verifiable applications based on zkVM?

Jens Groth: We provide a comprehensive SDK, CLI toolchain, and detailed documentation [10] to help developers easily get started in building Verifiable programs.

Our API design incorporates misuse prevention mechanisms and balances performance, verifiability, and security, aiming to make Verifiable Execution[11] a standard capability for developers.

The integration and implementation with AI scenarios

OpenBuild: How does Nexus serve AI scenarios? What are the typical applications?

Jens Groth: The trustworthiness of AI [12] becoming a key challenge. The verifiable computing power of the Nexus can be used in a number of scenarios:

• AI agent verification: For example, verifying whether the AI has completed flight bookings within the authorized and budget limits.

• Crypto Trading Assistant: Ensures that AI tools do not generate "hallucination trades."

• Verifiable media: Such as images captured by cameras that automatically sign geographic coordinates and time, preventing deepfakes.

These are all common scenarios, and Nexus is committed to becoming the underlying standard for the trustworthiness of these AI applications [13].

The design philosophy of privacy vs performance

OpenBuild: How does Nexus achieve a technical balance between privacy protection and performance optimization?

Jens Groth: The core cost of verifiable computation is generating zk proofs, a process that is still relatively expensive. Nexus significantly reduces latency by parallelizing computation through the prover network.

Running Verifiable Computation locally can almost provide privacy protection for free, as zk proofs are both compact and do not expose confidential data. Distributed processing may pose privacy risks, so we are actively researching how to maintain both privacy and efficiency in a distributed architecture.

Roadmap and Future Plans

OpenBuild: What are the key focuses for Nexus in the next 1-2 years?

Jens Groth: We plan to launch another testnet this year in preparation for the mainnet launch. At the same time, we've just launched the Verifiable AI Lab[14] to explore the deep integration of AI and verifiable computing.

When it comes to market expansion, we have ambitious goals: we want the entire digital world to be verifiable in the future. We have established cooperation with more than 50 projects and adhere to the product philosophy of "fast delivery, continuous iteration".

Conclusion

Through this interview, we can see that Nexus is not just creating another "chain", but is redefining the way "trustworthy computing" works in the digital world. The zkVM and distributed prover network it has built will become a new paradigm for trustworthy collaboration in AI, Web3, and even the traditional internet.

Reference link

[1]zkVM:

[2]prover Network:

[3]Jens Groth:

[4]Groth16:

[5]Nexus zkVM 3.0:

[6]RISC-V:

[7]StarkWare:

[8]Stwo prover:

[9]NEX Points:

[10] Document:

[11]Verifiable Executio:

[12]AI's credibility issues:

[13]AI Application Trustworthiness:

[14]Verifiable AI Lab: