A brand new report, The European Open-Source AI Landscape, has simply been launched by the StepUp Startups venture. The report affords a snapshot of the place Europe at present stands within the world AI race, and explores how sooner adoption of open-source AI might enhance competitiveness whereas strengthening the EU’s digital sovereignty.
Open-source AI refers to fashions, instruments and datasets whose parts (akin to supply code, mannequin weights and documentation) are brazenly out there. This transparency lowers the barrier for universities, public establishments, startups and established corporations alike, permitting them to develop and deploy AI with out being tied to proprietary methods.
In accordance with the report, open-source is already embedded throughout the AI tech stack: greater than half of builders depend on open fashions, datasets or instruments regularly. It additionally helps safer AI, giving researchers the visibility they should take a look at and consider fashions in depth.
A quick-moving world race
Since 2022, the variety of publicly launched AI fashions has greater than doubled, and extra of them now include open weights. This offers builders far higher perception and management. Open fashions are additionally quickly catching up with proprietary methods and, on some benchmarks, even matching expert-level efficiency.
Europe’s strengths: Collaboration and belief
Europe involves this problem with actual benefits. The area boasts top-tier analysis, shut collaboration between academia and trade, and main contributions to instruments like scikit-learn, spaCy and PyTorch, which underpin hundreds of business AI methods worldwide.
Many European AI startups are additionally releasing open-source fashions beneath open-source licences, offering constructing blocks that others can use and scale.
But adoption in European companies stays comparatively low: in 2024, simply 14% of EU companies have been utilizing AI. To sort out this, the EU launched its Apply AI technique in October 2025, centered on rising uptake throughout European corporations. The report highlights that open-source AI can play a key function in closing this ‘use hole’, providing clear, reusable, and cost-effective instruments aligned with EU values.
Scaling open-source AI: Europe’s alternative
Entry to computing stays a serious problem for European innovators. The 19 EU-funded AI Factories and expanded EuroHPC supercomputers purpose to alter that, giving startups and SMEs free entry to the GPU capability they should develop European fashions, together with open-weight methods.
This public infrastructure is already bearing fruit. In September 2025, Latvian SME Tilde launched TildeOen LLM, a 30-billion-parameter open-source language mannequin, educated utilizing 2 million GPU hours on the EuroHPC LUMI supercomputer.
A path in the direction of sovereign and aggressive AI
With inference prices falling by greater than 99% in simply two years, AI is quickly changing into low-cost and broadly accessible. Mixed with Europe’s analysis strengths, multilingual expertise and controlled strategy, open-source affords an actual alternative for the EU to carve out a number one place.
Slightly than focusing solely on constructing the biggest frontier fashions, Europe can take the lead by creating open, trusted, multilingual and sector-specific AI aligned with its industrial and societal priorities. Doing so would require funding, simpler entry to compute, and focused assist to assist organisations – significantly SMEs – undertake and deploy open-source AI successfully.
Download the report to discover the complete findings.
