On April 4, 2026, Spain’s Teleprensa reported that Gijón‑based Empathy Holdings has built its own emissions‑free private AI cloud to serve more than 500 companies, avoiding hyperscale providers. The firm pitches its platform as a “sovereign” European alternative that reduces dependency on U.S. and Chinese cloud and AI vendors.
This article aggregates reporting from 1 news source. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Empathy’s decision to run its own AI‑capable private cloud instead of leaning on U.S. hyperscalers is a microcosm of Europe’s sovereignty push in the AI era. The company says it started as one of AWS’s biggest customers in northern Spain, then walked away after repeated concerns over privacy, unexpected cost spikes and IP control. Now it touts an in‑house, low‑emissions stack that it claims already serves over 500 customers, explicitly framed as a way to “break Europe’s dependency” on American and Chinese providers. ([teleprensa.com](https://www.teleprensa.com/articulo/nacional-3/compania-asturiana-empathy-lanza-inteligencia-artificial-margen-grandes-proveedores/202604041047242383924.html))
From a race‑to‑AGI perspective, Empathy isn’t competing to build frontier models; it’s competing to own the last‑mile infrastructure and trust layer for European businesses that want strong AI without surrendering data and sovereignty. If this model scales, it could erode the assumption that only hyperscalers can economically host AI workloads, especially for mid‑market enterprises. It also hints at a more federated AI future in which frontier models run on regional, sovereign clouds via licensing or on‑prem deployments, rather than everything centralizing in a few U.S. data‑center clusters. That fragmentation could slow down fully centralized AGI development at the margin, but it may also deepen real‑world adoption by making enterprises and regulators more comfortable deploying powerful models in sensitive domains.
