Revolutionizing ZK Tech with Hardware
Interest in zero-knowledge (ZK) technology has surged in recent years, with a particular focus on addressing scalability challenges in blockchain and unlocking new applications in identity, machine learning, dePIN, and various industries. Notably, the combined market capitalization of leading ZK Layer1s and Layer2s, including Polygon, Aztec, Scroll*, zkSync*, Starkware, Aleo, and others, has surpassed $10 billion [1], attracting over $1 billion in venture capital investments [2, 3, 4, 5].Despite the tremendous potential of ZK technologies, their efficacy is hindered by the substantial computational overhead required for zero-knowledge proof generation, which is typically 10^4 to 10^6 times more computationally intensive than simply running the computation itself [6].
To address this pressing bottleneck, we at Mirana have been looking toward hardware-based acceleration approaches such as GPUs, FPGAs, and ASICs. However, the development of effective chips is not only expensive but also requires both deep hardware expertise and sharp insights into the relevant bottlenecks. Choosing the right team(s) to tackle this daunting and multi-faceted problem must be done carefully.
Recognizing the complexity and cost associated with developing effective chips, Ingonyama is meticulously assembling a team with deep hardware expertise and acute insights into relevant bottlenecks, emphasizing the critical importance of choosing the right team for this multifaceted task. After getting to know the Ingonyama team and their capabilities over the course of a year, we felt comfortable investing in their seed round and joining them in their journey to tackle this difficult problem.
The Nascent Stage of ZK
TechnologyZK technology, while promising, is still in its early stages. The dominant proving stacks and their associated bottlenecks remain largely uncharted territory. This unexplored landscape presents a significant opportunity but also poses risks. Chip manufacture is expensive, and spending years and millions of dollars creating a chip that doesn't properly address market demands can easily bankrupt a seed-stage company and leave their competitors with an unsurmountable lead.Ingonyama stands out with its empirical and multi-faceted approach, leveraging strong cryptographic and hardware expertise to navigate and innovate in this evolving domain. Ingonyama's strategy revolves around the calculated development of the ICICLE framework, an abstraction layer that streamlines proof generation across GPU, FPGA, and ASIC backends with minimal modifications**.
ASIC Development: A Calculated Future Move
A critical part of Ingonyama’s strategy is the development of the ICICLE framework. ICICLE abstracts away the hardware backend used for proof generation behind a uniform API. Developers can easily port their code between GPU, FPGA, and ASIC backends with minimal modifications**, freeing their time to focus on application logic implementation. By developing this free and easy-to-use middleware, Ingonyama plans to attract a broad initial user base. This strategy not only provides valuable developer feedback, but also ensures that these developers become reliant on Ingonyama's technology stack.
Additionally, Ingonyama is working on a zkContainer product designed specifically for cloud datacenters. This represents a strategic move to gain insights and foster relationships with major players in the ZK space. By targeting these large companies and their cloud service needs, Ingonyama positions itself as a key partner and also continues to gain market insight into the requirements and bottlenecks of large ZK proof producers.
In the short term, Ingonyama plans to concentrate on GPUs and FPGAs for both ICICLE and zkContainer. The approach gathers market feedback and grows market share while minimizing risk. This groundwork is crucial for their long-term goal of addressing the market’s evolving needs and paving the way for eventual ASIC tapeout.
The Verifiable Internet of the Future
The ZK proof production market among dominant L2s is expected to exceed $1 billion in revenue within the next 3-5 years***. However, Ingonyama's chips may have applications beyond zk-rollups as well, extending beyond blockchain to areas such as deepfake prevention, data provenance, oracles, DID, verifiable machine learning inference, and more. Ingonyama's commitment to ZK hardware acceleration positions it as a catalyst for reshaping online interactions, moving towards trustless verifiability.
Mirana Ventures firmly believes in Ingonyama's vision and technical capabilities, anticipating its pivotal role in advancing zero-knowledge technologies as a whole.
*Part of the Mirana Portfolio
**FPGA and ASIC are future modifications. For now, ICICLE works only with a GPU backend
***Based on internal discussions with major rollup providers regarding their cloud compute costs
Reference:
[1] https://www.coingecko.com/en/categories/zero-knowledge-zk
[4] https://aleo.org/post/aleo-raises-series-b/
[6] https://www.coinlive.com/news/the-next-battleground-for-military-strategists-generating-a-zk-proof