Swiss AI startup Giotto.ai is looking to raise funds at a valuation exceeding $1 billion, positioning itself as one of Europe’s emerging players in the race toward artificial general intelligence (AGI), according to sources.
Based in Lausanne, the company has engaged Lazard to manage a funding round of more than $200 million, the sources said on condition of anonymity.
Giotto.ai has told potential backers that the fresh capital will go toward advancing its AI research, developing initial commercial prototypes for enterprise and government clients, and releasing parts of its core technology as open source, the people added.
The funding round will gauge whether investors are willing to back a newcomer from outside Silicon Valley in an increasingly competitive race to build frontier AI models. U.S.-based labs like OpenAI and Anthropic have already secured billions in capital.
Meanwhile, AI investment in Europe has been accelerating as the region works to nurture its own champions, seeking to establish digital sovereignty and carve out a position beyond the U.S.–China rivalry.
Giotto.ai, founded in 2017 by CEO Aldo Podesta, has raised around CHF 15 million ($19 million) to date. In 2022, it sold its medical device compliance platform to RQM+, and has since shifted its focus to fundamental research on reasoning models.
Before starting Giotto, Podesta worked in sales strategy at Philip Morris, according to his LinkedIn profile.
The Swiss AI startup highlights its research strength through its top ranking on the Kaggle ARC-AGI-2 leaderboard, where it achieved a 25% score. Giotto also claims to operate at a significantly lower cost per task compared to larger AI labs.
The ARC-AGI-2 benchmark measures how well models can deduce rules from limited examples—an ability regarded as an early indicator of general reasoning. Higher performance suggests stronger generalisation to new problems, which is considered a proxy for progress toward more capable and dependable AI systems.
Giotto participates in Kaggle’s fixed-resource track, where all competitors face identical constraints: no internet access, a 12-hour runtime cap, and a standardised hardware budget. This setup contrasts with the unconstrained ARC Prize platform, used by major players like OpenAI and xAI, where teams can deploy larger models and tap into significantly greater compute resources to maximise accuracy.
