The companies powering Europe's next frontier labs.
The challenge kickstarts the 10 frontier labs. Suppliers give them the compute and software to build.
Frontier AI runs on two things before it runs on anything else: compute and the software stack around it. New teams need both from day one, at rates that let them experiment rather than ration. Next Frontier AI brings in the suppliers who provide that, offer the funded teams preferred terms, and grow with them as they scale.
Every one of the 10 labs is backed to raise up to €1 bn. The suppliers who back them early are the ones they buy from for years.
How the suppliers programme works
What suppliers do
There are two categories: Compute and Software. Each covers a part of the stack the labs depend on to train, run and ship. Suppliers offer the funded teams preferred rates. In return they get early access to those teams, a hand in helping them scale, and the long-term commercial relationships that come with being there from the start.
How suppliers are selected
Suppliers are selected by SPRIND's AI core team against eligibility criteria set for each category. Onboarding takes no longer than two weeks from first contact. Suppliers are featured on this page, get early sight of programme progress before it is announced publicly, and take part in engagement formats built around their category.
Interested in becoming a supplier?
If your product fits one of the two categories, reach out and apply:
- Compute: the providers of GPU capacity, HPC and data centre infrastructure that frontier training and inference run on.
- Software: the makers of the platforms, tooling and infrastructure software the teams build with.
Participation in the program is limited to a maximum of ten suppliers. SPRIND does not endorse individual offers or become a contractual party to any arrangement between a supplier and a funded team.
Compute
Compute
You run a cloud, an HPC facility or a data centre, or you supply the accelerators, systems and networking that large-scale AI training depends on. You know what it costs a new team to secure serious capacity, and you know how few can do it on terms that leave room to experiment. What you have is the one input no lab can start without.
Who qualifies
- Provider of GPU or accelerator cloud capacity, HPC, data centre infrastructure, or the hardware and networking that underpins large-scale AI training and inference
- Capacity to serve training and inference workloads at the scale frontier labs need
- Willingness to offer the funded teams preferred rates or reserved capacity
- No direct commercial conflict with SPRIND's AI challenge teams
What you bring
Compute is the largest line item a frontier lab carries and the hardest thing for a new team to lock in. Preferred rates and reserved capacity are the difference between teams that iterate freely and teams that ration every run. Get the labs onto your stack early, at terms they can build on, and you remove the single biggest constraint on how fast they move.
What you get
Early access to 10 teams that start small and grow into significant, long-term compute buyers. A direct line to their technical leads while they are still choosing where to train. The chance to help those teams scale, and to hold the account as they do it. Long-term contracts with labs backed to raise up to €1 bn.
Nscale is the global hyperscaler engineered for AI infrastructure — vertically integrated compute, data centres, and orchestration across Europe, North America, and beyond. Backed by $2bn in Series C funding (the largest in European history) at a $14.6bn valuation, and by investors including Aker ASA, Nokia, and NVIDIA.
- AI Infrastructure
- United Kingdom
European NVIDIA Cloud Partner building sovereign AI factories across Germany, with a high-density GPU data centre in Munich and further sites scaling towards 120 MW.
- Sovereign AI Cloud
- Germany
Deutsche Telekom's enterprise IT and cloud arm, providing sovereign cloud, data centre capacity and managed AI infrastructure to European industry and the public sector.
- Cloud & IT Services
- Germany
Berlin-based GPU cloud for European AI teams, running EU-sovereign NVIDIA Blackwell hardware with per-second billing, one-click deployment and no vendor lock-in.
- GPU Cloud
- Germany
Operator of Azure, one of the world's largest AI clouds, supplying GPU compute and the surrounding infrastructure for training and inference at global scale.
- Cloud Platform
- Global
Denmark's first AI supercomputer, built on an NVIDIA DGX SuperPOD with 1,528 NVIDIA H100 GPUs, operated by the Danish Centre for AI Innovation (DCAI) through a public private partnership between the Novo Nordisk Foundation and Denmark's Export and Investment Fund (EIFO).
- AI Infrastructure
- Denmark
European AI cloud provider offering GPU compute, storage, and networking infrastructure for AI training and inference, powered by NVIDIA hardware and an in-house AI research lab.
- AI Infrastructure
- Europe
AI cloud company listed on Nasdaq, building a full-stack platform spanning data, model training and tuning, and production deployment, with proprietary compute, storage, and software engineered specifically for AI workloads.
- AI Infrastructure
- Netherlands
Software
Coming Soon
You build what technical teams run on top of the hardware: data platforms, training and inference frameworks, orchestration, evaluation, observability, storage, security, developer tooling. The labs assemble their stack in their first months, and the tools they pick then are the tools they scale with. That is the moment you want to be in the room.
Who qualifies
- Provider of software that AI teams depend on in production: data tooling, ML platforms, orchestration, developer tools, evaluation, monitoring, security, storage, or adjacent categories
- A product ready for use by demanding technical teams, not a roadmap promise
- Willingness to offer the funded teams preferred rates or licences
- No direct commercial conflict with SPRIND's AI challenge teams
What you bring
A small team moves like a large one when its tooling is right. Preferred access to your product, at the point when the labs are deciding what to build with, takes friction out of their first year and puts your software at the core of how they work. The teams that adopt you early are the ones who never have to migrate away.
What you get
Early adoption by 10 well-funded teams choosing their stack from a blank slate. As the teams grow, your footprint inside them grows too. Long-term partnerships, real usage at scale, and reference accounts that carry weight with the rest of the European AI market.