Axelera AI, based in the Netherlands, launched Europa on 15 September 2026. The company told Reuters it had secured chip-supply contracts for AI factories. Chief executive Fabrizio Del Maffeo withheld precise details. Its $1.5 billion sales pipeline represents prospective business, not booked orders or recognised revenue.
Europa arrives in validated server systems
The launch announcement says shipments have begun. Europa targets enterprise inference: running existing models rather than training them. The supplier offers several hardware formats:
- A processor for customers designing their own boards.
- The compact Edge 232p PCIe accelerator card.
- The larger Server 250p PCIe card.
Axelera identifies Dell XE5 and Supermicro 111AD as validated systems for Edge 232p. Its Voyager software supports model deployment across the company's hardware range.
An inference partition within IT4LIA
Earlier procurement documents provide more specific project context. EuroHPC's 22 April announcement describes IT4LIA in Italy, hosted by CINECA at Bologna's DAMA Tecnopolo. E4 Computer Engineering and Dell are supplying the system. A dedicated inference partition combines Axelera accelerators with SiPearl processors, alongside the wider NVIDIA-based infrastructure.
The €290 million budget covers acquisition, delivery, installation and maintenance of the complete supercomputer; EuroHPC funds half. This is a project budget, not the value of Axelera's chip orders. The intended users include startups, smaller businesses and researchers.
Installation timetable for MeluXina-AI
In Luxembourg, EuroHPC announced a separate E4 procurement contract on 22 July. Dell participates as subcontractor, with LuxProvide operating MeluXina-AI. Installation at Bissen and Bettembourg was expected to start in autumn 2026. The acquisition and associated services carry an €80 million budget, also half financed by EuroHPC.
That announcement describes an NVIDIA-based system without specifying a Europa allocation. It therefore does not establish how many Axelera chips the project will receive, or that their deployment is complete.