Foundry

At Mithril, we envision a world where AI infrastructure is universally accessible and effortless to navigate, empowering innovators to focus purely on advancing intelligent technologies instead of managing complex compute logistics. Our mission is to dismantle the barriers of AI capacity planning by delivering a seamless omnicloud platform that unites distributed GPU resources across multiple cloud providers in one transparent ecosystem.

We are driven by a commitment to enhance the efficiency and affordability of AI compute, using innovative multi-cloud orchestration and market-based pricing mechanisms to transform how AI workloads are scaled and executed. Our platform enables developers and researchers to accelerate breakthroughs without the friction of fragmented cloud negotiations or infrastructure bottlenecks.

Grounded in expertise from DeepMind and Stanford, Mithril is constructing the future of AI infrastructure — where powerful, flexible, and reliable compute is available on demand, fueling a new wave of discovery and progress in AI.

When we first encountered Mithril, we were immediately struck by how they're tackling one of AI development's biggest headaches: managing compute resources across multiple cloud providers. It's like they've built a universal remote control for cloud GPU resources, and that's pretty exciting stuff.

A Fresh Take on Cloud Computing

What really caught our attention is Mithril's omnicloud approach. Instead of dealing with the hassle of juggling different cloud providers (and their varying pricing models), they've created a unified platform that handles all the heavy lifting. Think of it as having a smart assistant that always knows where to find the best GPU resources at the best prices.

Impressive Cost Savings

The numbers here are compelling. Their batch inference service is slashing costs by 2-5x for large AI workloads. We've seen plenty of companies promise cost savings, but Mithril's backing these claims with real results, especially in resource-intensive tasks like document analysis and content generation.

Built by AI Veterans

What gives us confidence in Mithril's approach is their team's pedigree. With alumni from Google DeepMind's Deep Learning team and Stanford's computer science program, they're not just theorizing about solutions – they've lived through these problems firsthand. Their rapid 550% platform usage growth in just one year suggests they're onto something big.

Where It Really Shines

We're particularly impressed by how Mithril serves both established players and up-and-coming AI startups. Their client roster, including LG AI Research and Arc Institute, shows they can handle enterprise-grade demands while remaining accessible to smaller teams. The platform's flexibility in resource provisioning – from hours to months – makes it practical for both experimental projects and long-term deployments.

With $80 million in seed funding and a clear vision for democratizing AI infrastructure, Mithril is positioning itself as a key player in making advanced AI development more accessible. While they're still relatively new to the scene, they're showing all the right signs of becoming a crucial piece of the AI development puzzle.