Computing Power for All
The future of AI is collaborative.
Why we exist
Open source does not mean open access.
Our initiative is driven by the belief that AI should be an openly shared asset. However, simply open sourcing AI models is insufficient. The predominance of large data centers controlling GPU resources presents a significant obstacle, creating a barrier to AI accessibility. In order to let all people have truly unfettered access to AI, we not only need open-source models but also open access to computing resources.
Everyone wins when everyone contributes.
We believe the future of AI is collaborative. At Hyperbolic, we’re building an open-access platform for AI development by aggregating idle computing resources and making it seamlessly simple to use them. Our approach allows individuals and organizations to use collective computing capacity for AI model training and host models on our platform — thus enabling open access to AI.
How we started
Founding Story
The Hyperbolic story begins with the stories of others.
The Hyperbolic story begins with the stories of others. From vantage points within startups and academia, our founders saw talented people with promising ideas get stuck, again and again, when they couldn’t access compute and inference tools.
They saw a Stanford postdoc blocked after he was unable to afford the thousands of GPUs needed to experiment with pushing the limit of LLMs. They saw a Berkeley professor struggle to find a sufficient machine for her research on the confidential computation of AI (aiming to ensure user privacy isn’t compromised by companies). They saw a startup, trying to train an AI model for their own use case, give up after realizing how much renting machines would reduce their runway. They saw a bright developer, who had built an app similar to Perplexity using OpenAI API, drastically limit users due to the API rate limit and excessive costs.
Different people, different projects, same story. And it turned out these stories were lived by AI innovators all over the world. The bottleneck to progress was not a lack of talent or enthusiasm, but compute. So they looked for it.
It wasn’t hard to find, it just wasn’t accessible. There are over two billion1 personal computers in the world and most of them sit idle for more than 19 hours a day.2 Additionally, a common situation has emerged where companies reserve machines from data centers for years, only to abandon strategies that leave their purchased resources unused. This prevalence has only been exacerbated by advancements in efficiency, like Ethereum’s shift to proof of stake, where the computing power equivalent of 10 million RTX 3090 GPUs was left without purpose nearly overnight.
Hyperbolic was founded to make sure generational breakthroughs aren’t left in the dust simply because great brain power lacked affordable computing power. We believe the future of AI is collaborative and open, and we’re working to make it possible every day.
1"Computers Sold". Worldometer.
2Desroches, Louis-Benoit, et al. "Computer usage and national energy consumption:Results from a field-metering study". eScholarship.