From application to compute in four steps.
The programme runs on capacity beyond our signed customer contracts and follows them in priority. It is an investment, not a discount: we take a stake because we expect the company to grow.
- 01 · Apply
Tell us about the company
The team, the models you train or serve, and the compute you need over the next twelve months.
- 02 · Due diligence
We look at team, technology and plan
The same review an investor would run, because that is what we become.
- 03 · Compute for equity
Capacity against a stake
Dedicated NVIDIA Blackwell capacity on Sapience infrastructure in the EU, in exchange for equity agreed case by case.
- 04 · Ecosystem
Introductions that move a company forward
Angels, VCs, investment banks, law firms and auditors from our network, introduced to the companies we back.
Europe's emerging AI companies.
Start-ups and scale-ups in Central and Eastern Europe and across the EU that train or serve their own models and need more compute than they can fund today. Europe's public programmes are too slow for a fast-moving sector; this one is built to move at the pace of the companies in it.
Model builders
Teams training or fine-tuning their own models who are compute-constrained before they are capital-constrained.
Inference-heavy products
Products whose unit economics depend on dedicated European GPUs rather than metered hyperscaler APIs.
Regulated builders
Companies whose customers require EU data residency from day one.
Pre-Series A
Companies that need compute and introductions before their next equity round.
A European home for models.
A neutral repository and distribution layer for open-source and proprietary machine learning: upload, discover, share and run models, datasets and applications, with EU residency. A long-term platform investment that sits on top of the infrastructure backbone.
Building in Europe and short of compute?
Send us a short note on the company, the models and the capacity you need. We reply with next steps.