How Ingonyama Is Accelerating the Future of Zero-Knowledge Proofs in AI, Gaming, and Beyond

Ingonyama is a next-generation semiconductor company focusing on Zero-Knowledge Proof (ZKP or ZK) hardware acceleration. The company’s mission is to lower the barrier of entry to ZKPs by making them inexpensive, accessible and fast.

But what exactly is ZKP? It’s a foundational technology that can disrupt AI, gaming, healthcare, blockchain and more by enabling privacy, verifiability and scalability. The basic model involves two parties: a prover and a verifier. These two parties mutually agree on a specific function and set of inputs to be used in their interaction, and the prover shares the result of computing the function on the inputs while keeping some parts of its input confidential. But if the input is kept confidential, how could the verifier then be assured that the result from the prover is correct?

The magic of ZK is that it’s actually possible to do this: ZK enables the verifier to be certain that the results shared by the prover is the correct output of the function operating on the inputs, even though the verifier cannot see the confidential parts of the inputs. Therefore, it becomes possible to prove something about some underlying data without revealing any information about that data (hence the term “zero knowledge”). Here’s another way to say this: ZKPs are a way to prove you know something, own something, or have done something without revealing any information about that something.

In addition to privacy and verifiability, another key characteristic of ZK is their scalability. The prover can compress a very large computation into a short proof (this is called “succinctness”), which makes it possible for a large network of verifiers to validate the proof with low computational cost and energy.

Many would say ZK is the single most important technology for the future of blockchains and Web3. In fact, a significant part of R&D in blockchain over the past couple of years has focused on layer 2 scaling solutions, in which ZK plays a central role. Outside of Web3 use cases, however, ZK is in a much earlier stage of maturity. Ingonyama, for its part, has identified two areas which are poised to become early adopters for ZK: AI and online gaming.

For AI and, specifically, generative AI and AI as a service, the company sees a strong regulatory push for accountability when it comes to mission critical applications, while at the same time the clients for AI services desire confidentiality for their data and content. ZK can be used to provide the necessary transparency to solve this tension. For online gaming, ZK can be used to offer a more balanced compute model, off-loading some of the heavy-lifting from the servers and distributing it to clients. ZK can also enable in-game assets and histories to be managed independently and objectively by the community of gamers, to eliminate the risk of the game studios as single-points-of-failure.

Unlocking the potential of ZK

In order for ZK to fulfill its potential, it must be possible for run ZK provers on large numbers of devices everywhere. However, the algorithms behind ZKP are complex and can be expensive and slow to operate, and in the past large, costly servers were required to run ZKP. Ingonyama’s mission is make ZK easy to deploy (inexpensive, fast, and accessible) by accelerating ZK on all the leading types of hardware devices.

ZK made easy

Ingonyama’s most recent product is a GPU library called ICICLE, which allows developers to accelerate ZK provers on Nvidia GPUs, leveraging types of GPUs that are widely available and not expensive. They are powerful enough to deliver a nice performance boost over other available technologies.

Ingonyama’s API provides a way to accelerate ZK, even with zero experience in programming GPUs. So far, customers are mostly product companies who use ZK as part of their product’s tech stack. Using ICICLE, they are able to scale their product and reduce overall compute costs. The goal is to make ICICLE a standard part of any ZK tech stack, to enable developers who are not ZK-specialists or GPU-experts to accelerate provers on specialized hardware such as GPUs or other dedicated devices. 

In another recent launch, Ingonyama has expanded its offering to support other types of customers, such as ZK-Cloud companies. These companies operate, or provide access to, a large fleet of GPUs to support multiple users and multiple applications to share resources simultaneously. Ingonyama’s new ZKContainers offer ZK Clouds an easy way to deploy and manage ICICLE based applications. 

Getting the word out

The demand for ZK is growing larger by the day. Ingonyama is witnessing—and influencing—a virtuous cycle where cheaper/easier proofs drive new use cases and increased adoption, and increasing adoption drives the development of cheaper/easier proofs. Currently, Ingonyama’s product focus is mainly software running on existing hardware. But with growing demand, the need for specialized hardware will lead to new market opportunities for the company.

To promote awareness about ZK, Ingonyama recently began hosting ZK Accelerate, an initiative aimed at market education and community building. The goal is to celebrate engineering breakthroughs and showcase the technology to a wide range of businesses. The company also launched Ingopedia, a community-driven effort to compile a collection of resources and information related to Zero Knowledge Proofs.

Walden Catalyst is proud to support Ingonyama as it continues to resolve the computational bottlenecks in ZKP and forge the foundation on which applications requiring ZKP performance can be built at greater speed and scale.