Inside Asturias' First Net-Zero AI Infrastructure

Empathy AI operates its self-hosted GPU infrastructure from a bioclimatic building in the Gijón Science and Technology Park, Asturias, Spain. The first net-zero energy office building in the region, this facility was designed to generate more energy than a conventional office consumes. With GPU infrastructure running inside, on-site solar generation covers between 60% and 80% of total energy consumption, with the remainder supplied through direct renewable energy agreements.

This is not just an office. It is the physical foundation of our commitment to sustainable AI, and the proof that high-performance AI infrastructure does not require environmental sacrifice.

Why Sustainable AI Infrastructure Matters

The International Energy Agency projects that global data center electricity consumption could exceed 1,000 TWh by 2026, roughly equivalent to Japan's entire electricity demand. Most AI companies rely on massive data centers powered by fossil fuels, consuming enormous amounts of water for cooling. A single large language model training run can emit as much carbon as five cars over their entire lifetimes, according to research from the University of Massachusetts Amherst.

We chose a different path.

How Our Bioclimatic Building Works

Our building uses architectural and engineering principles that minimize energy consumption before any renewable generation even begins:

• Active Slab Systems circulate fluid through the building's concrete floor slabs, providing efficient heating and cooling with minimal energy demand.
• Natural ventilation leverages Asturias' Atlantic climate for server cooling, eliminating water-intensive cooling towers.
• 60kWp photovoltaic pergola on the south facade doubles as solar shading and on-site energy generation, covering 60-80% of the facility's total energy consumption.
• Thermal mass construction stabilizes internal temperatures, reducing peak energy loads.
• Direct renewable energy agreements supply the remaining 20-40% of consumption, ensuring a net-zero energy balance annually.

The result: a facility that delivers the computational power enterprises need while maintaining a net-zero energy footprint.

Self-Hosted, Dedicated GPU Infrastructure

Inside this building sits Empathy AI's dedicated GPU infrastructure: L40, H100, and H200 GPU servers alongside NVIDIA DGX Spark supercomputers. Every model we train, every query we process, runs on hardware we own and operate. No AWS. No Google Cloud. No third-party dependencies.

Each DGX Spark can host up to 128 isolated workspaces for teams to create, iterate, and experiment, turning what would exceed €130,000 per year in cloud AI costs into a one-time hardware investment that powers innovation instead of dependency.

This means complete data sovereignty for our clients. When you use Empathy AI, your data never leaves our controlled environment. Your queries are processed on open-source LLMs running on GPUs we maintain, physically secured in a building designed for both performance and sustainability.

When AI is hosted externally, reliability depends on the stability of global networks and third-party vendors. When AI is hosted internally, reliability becomes an architectural advantage. If your AI lives inside your own walls, no outage thousands of kilometers away can interrupt it.

Sustainability as Architecture, Not Marketing

Many technology companies claim sustainability through carbon offset programs or renewable energy credits purchased from distant wind farms. Our approach is different: sustainability is embedded in the physical architecture of our infrastructure.

Our on-site 60kWp photovoltaic pergola covers 60-80% of the energy our GPU clusters consume. The remaining demand is met through direct renewable supply agreements, not carbon credits or distant offset schemes. There is no greenwashing. The building itself is the sustainability strategy.

This is what powers the #BigTechRebellion, infrastructure that proves you can take back control from big tech without inheriting their environmental footprint.