As an AI researcher and developer, Hyperbolic’s mission is deeply personal to me. That’s why our $12M Series A with Variant and Polychain Capital is particularly significant. It gives us more resources to tackle long-standing issues my peers and I have faced firsthand: inaccessible compute, limited access to optimized open-source models, and inefficient development workflows.

During my time as an AI and computer science Ph.D. researcher and in my role at Microsoft Research, I struggled to make progress on AI research because GPU resources were difficult to access. I watched colleagues face the same limitation, forcing some projects to stall or be abandoned altogether.

Since I began working with my co-founder, Dr. Jasper Zhang, on Hyperbolic in 2022, I have had the opportunity to work with a like-minded partner on solving the AI compute problem. My background in AI systems and distributed infrastructure, together with his mathematical expertise in verification, came together around a shared goal: breaking down infrastructure barriers for the people pushing AI forward.

We share a vision of building open and accessible AI infrastructure that helps developers, researchers, and AI teams access the compute and models they need to build. Hyperbolic exists to make AI infrastructure more available, reliable, and cost-efficient, so more builders can experiment, deploy, and scale their work.

Open and Accessible AI

Hyperbolic’s infrastructure is designed to support collaboration, transparency, and accessibility across AI development.

By “open,” we mean:

Open GPU Access

Hyperbolic gives developers and researchers access to high-performance GPU compute without the long procurement cycles, high costs, or operational complexity that often slow AI teams down.

Open-Source Models

We believe open-source AI is critical to the future of the field. The rapid progress of open-source models has shown that they can rival, and in some cases surpass, closed alternatives. We are committed to supporting leading open-source models and making them easier for developers to use.

Open Community

Transparency and collaboration are central to how we build. By listening closely to developers, researchers, and AI teams, we aim to create infrastructure that solves real problems and helps users move faster.

How We Break Down Barriers to Progress

To support this vision, Hyperbolic provides two core products: GPU compute and inference services.

Hyperbolic GPU Compute

Accessible

Hyperbolic makes GPU compute easier to access for developers, researchers, startups, and AI teams. Users can rent compute when they need it, without waiting through lengthy procurement processes.

Reliable

Our infrastructure is built to support real AI workloads, with ongoing monitoring for uptime, availability, and performance.

Easy to Launch

Renters can connect to GPUs quickly, allowing them to start running workloads without unnecessary setup friction.

Built for AI Workloads

Developers can use Hyperbolic for experimentation, model development, fine-tuning, training, and other compute-intensive AI workloads.

Hyperbolic Inference Services: On-Demand Access to Open-Source Models

Scalable

Hyperbolic’s inference services are designed to support many users and workloads simultaneously, giving developers an easier way to access powerful open-source models.

Fast

We have achieved high throughput and low latency. For example, our proprietary optimization methods led to a 145% increase in output token throughput for the Llama-3-8B model on an AMD MI250 chip compared to vLLM, a widely adopted serving framework for LLMs.

Heterogeneous

Our inference stack supports Nvidia, AMD, Intel, and Apple GPUs, making it versatile across hardware and more accessible for developers.

Broad Model Coverage

We support leading open-source models, including base models, and prioritize onboarding new models based on user needs.

Where the Series A Will Take Us

This Series A enables us to keep building the infrastructure developers and researchers need to push AI forward. With these resources, we will expand our engineering team, improve the performance and reliability of our platform, and continue scaling the products that power GPU compute and inference on Hyperbolic.

At the same time, we are growing our go-to-market and business development teams so we can reach more AI builders and support broader adoption across the developer and research communities.

Looking ahead, we plan to use this round to continue improving our AI cloud, expand support for open-source models, and make high-performance compute easier to access. As we move forward, we remain committed to transparency, collaboration, and building infrastructure that helps more people participate in AI development.

We are not just building technology. We are building the foundation that helps developers, researchers, startups, and AI teams bring ambitious AI ideas to life.

Take Your Ideas Hyperbolic

Visit app.hyperbolic.ai to access the infrastructure you need to bring your AI ideas to life.

Want to Learn More About Our Raise?

Check out our press release announcing the raise.

About Hyperbolic

Hyperbolic is the Open-Access AI Cloud, giving researchers, startups, developers, and AI-native companies fast, flexible access to high-performance GPU capacity. The platform helps teams start on demand, scale programmatically, and grow into reserved infrastructure without long waitlists, rigid contracts, or complex procurement cycles.

Founded by award-winning math and AI researchers from UC Berkeley and the University of Washington, Hyperbolic is committed to making advanced AI infrastructure more accessible to builders around the world.


Website | X | LinkedIn | YouTube | GitHub | Documentation