STELIA

Our Story

Stelia is Without Boundaries

What began in 2020 as deep research, built from the infrastructure up, is today one of Europe’s leading frontier AI companies.

We build alongside the most advanced institutions on Earth, and we point that power where it counts: at discoveries that cross borders, at science that reaches further than it ever has, at outcomes measured in lives, in harvests, in habitable years.

Our capabilities are a single, living ecosystem. Each is a product with a clear and separable purpose, and each composes into platforms built for the real world. Take the few you need, or take the whole stack. Assembled, they reach what no part could alone.

Our people are engineers, researchers and philosophers, bound by one obsession: the real-world consequence of advanced reasoning. We build the entire depth of it, from the physical substrate and the foundations to the applications and the outcomes, hand in hand with the most advanced companies in the world.

Six years, from the fabric up.

Stelia began in 2020 as a research and development business, looking at the widespread distribution of AI from the infrastructure and the technology upward. We started with the hardest part, the networking, and built our way to the whole stack. This is the road we took, and the team that took it.

2020

The beginning

Stelia is founded as a research and development business, on one conviction: that we had to build the widespread distribution of AI from the infrastructure and technology upward. We began with the deepest layer and worked toward the surface.

2021 – 22

Networking first

We build Hyperband, our petabit-scale AI networking fabric, the distribution technology that moves models and workloads across GPU clusters and continents. Distribution first: the hardest problem, solved before the rest of the stack existed.

2022 – 24

The stack we own

Layer by layer we build the substrate outright: Meridian, one cloud operating model across any infrastructure; Foundry, the elastic compute beneath it; Bastion and Proximity for sovereign, single-tenant isolation. Owned, rather than rented.

A world first · groundbreaking
September 2024

Training across the world

Stelia becomes the first company in the world to successfully complete LLM training across two geographically distributed clusters, proving that frontier models can be trained across sites and continents, not only inside a single data centre. A genuinely groundbreaking result.

2024 – 25

The launch of Lyra

We launch Lyra, full-stack orchestration, distribution and agentic management. Lyra runs models, agents and workloads across the entire stack as one coordinated system, from the silicon to the application.

2025 – 26

Contextual awareness

With Continuum, the platform gains contextual awareness: a living, governed shared memory every person, model and agent reads from and writes to in real time, so a company’s intelligence compounds instead of resetting each day.

August 2026

Stelia Workspace

It all comes together in Stelia Workspace, one team-native surface over the whole stack. A team grown from a handful of researchers into one of Europe’s leading frontier AI companies, holding everything from silicon to workspace.

Legacy

Groundbreaking from the outset.

Research and development has been at the core of Stelia since the first day. We were working in artificial intelligence long before it was fashionable, building the deep infrastructure of distributed intelligence while the rest of the world was still deciding whether it mattered.

That head start is why we hold the whole stack today, and why we build from first principles rather than assembling someone else’s parts. The frontier is not a place we arrived at. It is where we began.

Without Boundaries

Everything we have built points one way: at the direct, real-world consequence of advanced reasoning, placed in the hands of the people doing the hardest work in the world. Six years in, we are only beginning.

Explore the platform  →