Jiri Fiala

The AI industry has a dirty secret nobody talks about at conferences. Every foundation model company, every enterprise AI team, every ambitious startup building the future—they're all fighting over the same constrained, overpriced, slow-to-deploy compute infrastructure. AWS charges €98/hour. Google Cloud takes €88. Traditional data centers need 5 years and a billion dollars before generating a single euro of revenue. I decided to fix that. DCXPS deploys containerized, renewable-powered AI data centers in 120 days. 520 NVIDIA GPUs per module. €36/hour—70% cheaper than hyperscalers. The math isn't complicated. The execution is. Before DCXPS, I spent 11+ years inside Hewlett Packard Enterprise designing data center's. cloud and outsourcing Enterprise Architecture for Fortune 500 clients—Deutsche Bank, Volkswagen, T-Systems. Then 12 MW of crypto mining modular facilities. Then an 8 MW AI data center. Each project taught me the same lesson: the grid can't keep up with GPU clusters. The future isn't about waiting for infrastructure—it's about bringing power directly to the compute. DCXPS is the result of that conviction. My approach hasn't changed in 20 years: Technology → Processes → People → Money. In that order. Always. I've co-founded 11 AI startups across the US and Europe, helped architect the trajectory of 100+ companies—several to unicorn status—and now channel everything into building the infrastructure layer upon which superintelligence will emerge. Not because it's trendy. Because someone has to build the rails before the trains can run. If you're working on something that needs serious compute at a price that doesn't require selling your equity—let's talk.