Two years ago - I met Jasper Zhang and Yuchen Jin, the co-founders of Hyperbolic.
Instead of opening the call with normal introductions, their first question to me was: “What is a blockchain?”
In any other context, I’d be able to respond effortlessly. But since I had looked up their backgrounds, I felt nervous about getting this one right.
Jasper had won multiple gold medals in international mathematics competitions, graduated as the student of the year from China’s premier university Peking University for mathematics, and completed his Ph.D. in Mathematics at the University of California, Berkeley – one of the world’s most prestigious programs – in just two years. Yuchen had a Ph.D. in distributed systems and networking from the University of Washington, achieving multiple publications in ICLR, Sigcomm, NSDI, and SOSP. He’d also led an AI engineering team at OctoAI, an AI company offering production-scale optimization solutions for models and applications.
Thankfully, I made it through the blockchain definition, and we kept that first conversation moving.
For the next 30 minutes of the call, the team painted a vision of interoperable, generalized, decentralized infrastructure — a world grounded in reliability, where blockchain is leveraged to make software cheaper, faster, and more secure.
And now – I couldn’t be more excited to announce our participation in Hyperbolic’s Seed Round, led by Polychain Capital and Lightspeed Faction.
Hyperbolic is building at the intersection of blockchain and AI, providing open-access AI computing and high-performance inference services.
Hyperbolic’s mission in its open-access platform is to make AI more widely available without restrictive barriers, such as high costs or limited compute availability. The team wants anyone, anywhere in the world to access models and resources – from individual developers to research organizations to large-scale enterprises.
To achieve this – they have developed state-of-the-art architecture that sits under the hood of their platform. This is where their competitive advantage comes from.
First, Hyperbolic applies a high-performance inference engine on “heterogeneous compute” which uses a compiling stack that optimizes tensor operators across a number of hardware backends, including Nvidia and AMD GPUs. This method allows them to efficiently utilize different types of GPUs, which optimizes model performance and drives down costs. Secondly, they have developed a decentralized orchestration layer that efficiently aggregates global compute and a GPU marketplace that enables users to rent and supply GPU resources. The marketplace is designed to be intuitive and automated, making it easy for anyone to participate, whether they are providing or consuming computing resources. And lastly, they invented and applied a novel verification mechanism Proof of Sampling on blockchain to ensure that the network of nodes operating the GPUs is verifiable and secure.
The team’s first product is an AI inference service that allows users to run AI models and applications at a significantly reduced cost compared to current platforms such as AWS and Microsoft Azure.
They’ve developed everything with ease and simplicity in mind, allowing anyone to quickly integrate and start using the latest open-source models. Right now, the platform offers Llama3 and Qwen for text generation, Cambrian for visual language models, Stable Diffusion 2, SD XL 1.0, SD XL 1.0 Turbo, Segmind SD for image generation, and MeloTTS for audio generation.
If you’re a team looking for advanced AI capabilities and high-performance computing resources, check out their products:
Run inference for free with up to 60 API requests per minute at Hyperbolic Models
Rent or supply GPUs in their marketplace at Hyperbolic GPU Marketplace
Congrats to Jasper, Yuchen, and the Hyperbolic team on the successful raise and product launch!! We at Chapter One look forward to continuing to be a part of your incredible journey 🚀 🔥
I always find it nice to put faces to names, so below is a picture of Jasper and Yuchen!
Congrats Jamesin!