Prime Intellect, an open and decentralized AI platform and protocol democratizing access to computing and intelligence, announced a $5.5 million seed funding round co-led by Distributed Global and CoinFund with participation from Compound, Collab+Currency, and Filecoin Creator Juan Benet with Protocol Labs, and renowned angel investors including Hugging Face CEO Clem Delangue.
This funding round will build the computing platform, enabling decentralized training across instances and co-ownership of AI models by contributing computing, code, data, capital, or expertise.
Generative AI will become over a $1 trillion market within the next few years. However, AI progress is hampered by the category’s operations already leaning heavily toward a traditional and centralized control model. Startups are capped by prohibitive computational costs, inflexible contracts, proprietary models that stifle open collaboration, and insufficient incentives to pool resources and expertise. Prime Intellect is working to solve these problems through a decentralized AI platform powered by advances in distributed model training.
This platform aggregates global computing power to provide instant, affordable, and scalable on-demand computing. Using Prime Intellect, anyone can help train powerful AI models and share in their ownership.
At the heart of the company’s platform is the Prime Intellect (PI) protocol, which provides the critical infrastructure and cryptoeconomic building blocks to align incentives around decentralized AI development, including:
- Scalable and distributed compute marketplace to make AI training globally accessible and affordable
- Collective ownership of AI models and computing via the PI protocol to incentivize contributions of compute capital, code, and continuous model improvement
- On-chain AI protocol primitives for facilitating trust and seamless collaboration
- Decentralized governance frameworks for aligning AI progress with collective human values
The computing platform is now live in beta and aggregating supply from 10+ centralized and decentralized GPU cloud providers, offering the lowest-cost GPUs on demand. Prime Intellect will soon roll out multi-node compute on-demand (16 to 128+ GPUs) and decentralized training across instances with optimized orchestration for efficiency.
Prime Intellect cofounders Vincent Weisser and Johannes Hagemann have an unusual and deep combination of AI and web3 experience. Weisser helped start VitaDAO, the leading organization in decentralized science, before leading product and AI at Molecule (the platform for biopharma IP). And Hagemann built and scaled distributed training frameworks from scratch at Aleph Alpha, and received a Best Paper Award for his recent work at NeurIPS 2023.
KEY QUOTES:
“By enabling collective governance of AI models and compute, Prime Intellect enables anyone to have a stake in the future of AI. Decentralized AI is one of the highest impact opportunities in history. We must break down the barriers that prevent brilliant minds from contributing to our collective future. Diverse communities of researchers, engineers, and organizations will be able to come together using the Prime Intellect platform to create transformative AI capabilities that would be impossible in isolation.”
- Vincent Weisser, Co-Founder, Prime Intellect
“The future of publicly available AI will be about verifiability and decentralization across the stack. Prime Intellect will enable a critical component of this through access to compute, optimized training and tokenized ownership. Vincent and Johannes are an incredibly talented team, with a rare combination of experience and vision, and we’re thrilled to support them in the early stages of their journey.”
- Einar Braathen, Partner at CoinFund
“The Prime Intellect platform removes the barriers of prohibitive computational costs, long contracts and limited access to scalable compute. Their platform provides critical infrastructure to align incentives around decentralized AI development”
- Lennard Schmidt, Cofounder at Langdock, an AI platform for team productivity
