Artificial intelligence startup Giza has secured $3 million in pre-seed funding to build a platform that enables AI developers to generate zero-knowledge proofs for their machine learning models, marking a significant step toward making computationally intensive AI operations verifiable on blockchain networks. The funding round, announced on July 11, 2023, was led by CoinFund with participation from StarkWare, TA Ventures, and Arrington Capital.
The Synergy
The convergence of AI and blockchain technology has been one of the most discussed narratives in the crypto space throughout 2023, but practical implementation has remained elusive. Giza, founded in October 2022 by Cem Dagdelen, Fran Algaba, and Renç Korzay, is tackling one of the fundamental challenges at this intersection: how to run AI models on blockchains that are inherently too slow and computationally limited to handle them directly. The answer, according to Giza, lies in zero-knowledge proofs, a cryptographic technique that allows one party to prove to another that a computation was performed correctly without revealing the underlying data. By executing AI models on conventional high-performance computers and then generating cryptographic proofs of the computation’s accuracy, Giza enables blockchains to verify AI outputs without actually running the models themselves. This approach dramatically reduces the computational burden on-chain while maintaining trustlessness and verifiability.
AI Use Cases in Web3
Giza’s platform opens up numerous possibilities for AI integration within the Web3 ecosystem. Smart contracts could incorporate verified AI predictions for decentralized lending protocols, enabling automated risk assessment based on verifiable machine learning models. Decentralized exchanges could use AI-powered price feeds that are cryptographically proven to be accurate, reducing oracle manipulation risks that have plagued protocols like Rodeo Finance. The platform could also enable verifiable AI-driven NFT generation, where the creative process is provably performed by a specific model with specific parameters. Einar Braathen, an investor at CoinFund, noted that Giza is positioned to become a pioneer in making AI accessible to smart contracts, significantly expanding their design capabilities. AI-themed cryptocurrencies had outperformed Bitcoin earlier in 2023, though price gains have moderated as initial enthusiasm cooled. According to Crunchbase data, AI startups raised nearly $58 billion in the second quarter of 2023, compared to approximately $2.3 billion raised by crypto firms during the same period.
Data Privacy Implications
The use of zero-knowledge proofs for AI computations carries significant implications for data privacy. Organizations could deploy proprietary AI models on-chain without revealing the model’s architecture or training data, addressing a key concern that has prevented many enterprises from engaging with public blockchain networks. Giza’s approach means sensitive machine learning models could be used in decentralized applications without exposing intellectual property. The company’s CEO, Fran Algaba, brings substantial AI expertise to the project, having previously served as head of machine learning at Banco Bilbao Vizcaya Argentaria, a multinational Spanish bank, and later at sportswear giant Adidas. His transition from traditional AI roles to the blockchain space reflects the growing trend of AI professionals exploring decentralized computing paradigms.
The Innovation Frontier
Giza currently employs a team of 12 and plans to release the underlying infrastructure of its platform shortly after the funding announcement, with a full launch targeted for the end of 2023. The broader AI-crypto intersection is attracting increasing venture capital attention, even as overall crypto funding has declined. The challenge remains significant: implementing zero-knowledge proofs is complex, particularly for developers without prior experience in cryptographic techniques. Giza aims to abstract this complexity by providing tools that allow AI developers to generate ZK proofs without deep cryptographic knowledge. If successful, the platform could catalyze a new wave of AI-powered decentralized applications, from autonomous trading agents to decentralized prediction markets powered by verified machine learning models.
Concluding Thoughts
Giza’s $3 million pre-seed round represents a meaningful bet on the AI-blockchain convergence thesis. While the technical challenges are substantial, the potential to bring verifiable AI computations to smart contracts could unlock entirely new categories of decentralized applications. With Bitcoin trading at approximately $30,620 and Ethereum at $1,878 at the time of the announcement, the broader crypto market remained stable, suggesting that investor interest in AI-crypto projects is driven more by long-term technological potential than short-term market dynamics. As the AI sector continues to attract outsized venture capital investment, expect more projects like Giza to emerge at the intersection of these two transformative technologies.
Disclaimer: This article is for informational purposes only and does not constitute financial or investment advice. Always conduct your own research before making investment decisions.
proving ML computations off-chain with ZK avoids the gas cost problem entirely. right architectural call for the EVM
ZK proofs for ML model verification is genuinely interesting. running inference on-chain is a dead end, proving the computation off-chain is the right approach
CoinFund and StarkWare backing gives credibility but $3M pre-seed is basically ramen money for something this ambitious. execution risk is enormous
$3M is nothing for ZK infrastructure. StarkWare alone raised hundreds of millions. Giza needs way more capital to ship something production grade
starkware participating in a $3M round is basically a signaling play. they want first look at ZK ML infra before it gets expensive
verifiable ML on chain is the one AI narrative that actually needs crypto. the rest is just marketing
$3M pre seed from starkware and coinfund is a strong signal. but proving ML inference with ZK is computationally heavy, the gas costs will be brutal
gas costs for ZK verification on ETH mainnet are the bottleneck. unless they batch proofs efficiently the verification alone could cost more than the ML inference
batching proofs is the only way gas stays manageable. a single zkML verification on mainnet can cost 500k+ gas. recursive aggregation brings that down 10x
proving ML off-chain and verifying on-chain is the only architecture that scales. running inference in a smart contract was always a fantasy
verifiable ML on chain is the one AI x crypto use case that makes sense. you dont need the model on chain, just proof it ran correctly
Arrington Capital in a 3M pre-seed for ZK ML infra. they were early on Solana too. worth tracking where they deploy early checks
Cem Dagdelen founding this in Oct 2022 means they shipped to mainnet in under a year. most ZK projects take twice that for a testnet
shipping ZK infra in under a year is impressive but also terrifying. most ZK projects take 2+ years because the math actually matters