Spain’s Sherpa.ai Raises $18M for AI That Never Touches Your Data
Sherpa.ai, a Spanish startup building federated-learning AI for banks, hospitals, and governments, has raised $18 million led by US cyber investor Forgepoint. This funding comes as demand grows for AI that can be trained without ever sharing the underlying data.
A Solution to Data Sovereignty Concerns
July 6, 2026 – In a world where everyone craves AI power but few want to hand over sensitive data, a Basque startup offers a potential solution. Sherpa.ai has developed AI capable of learning from data without ever accessing it directly. The company caters primarily to banks, hospitals, and government entities most concerned about data privacy.
Federated Learning: Training Without Sharing
At its core, Sherpa.ai leverages federated learning. Instead of centralizing data to train an AI model, the model travels to the data. Each organization trains the model locally and only shares the learned insights, not the raw data. The startup claims its research can reduce data transfer by up to 99%.
This technology addresses significant challenges in regulated industries where privacy rules restrict traditional AI projects. As stated by founder and CEO Xabi Uribe-Etxebarria, the goal is to enable firms to " harness the full potential of AI without giving up control, privacy, and sovereignty over their data."
A European Success with American Backing
The round was supported by existing investors Mundi Ventures, Ekarpen, Allegra Holdings, and SETT. Surprisingly, Forgepoint Capital, a Silicon Valley cyber and AI investor, also participated. This investment highlights the growing interest in European startups focused on keeping AI within geographical borders.
Sherpa.ai’s client list underscores this trend, featuring clients like Spain’s Indra, banks Caja Laboral and Unicaja, security firm Prosegur, genomics company Centogene, and even the US National Institutes of Health (NIH). A privacy-first European firm serving a US federal agency serves as compelling evidence of its offering.
Research and Real-World Applications
While "sovereign AI" is a crowded label and federated learning isn’t new, Sherpa.ai’s research gives it an edge. The company has published peer-reviewed work on training large language models across private datasets and collaborated with the NIH and University College London on using federated learning for rare-disease diagnosis.
A Growing Market Opportunity
With governments implementing stricter data rules, the demand for AI that respects data borders is rising. As such, Sherpa.ai bets $18 million that its story will resonate in a market eager for privacy-preserving AI solutions.