2CRSi has secured a binding order worth more than €60 million from Exaion for 83 NVIDIA HGX B300 AI servers, a contract that underlines how the scramble to build sovereign AI infrastructure in Europe is turning into tangible spending.
2CRSi Wins €60 Million Exaion AI Server Order
The order matters because it converts demand for AI capacity into a large, near-term revenue stream for a relatively small French server maker while reinforcing a broader industrial theme: governments and enterprises want AI systems they can control, host and secure within local jurisdictions. For 2CRSi, the deal offers validation that its high-performance, energy-efficient hardware and immersion-cooling technology can win business in one of the fastest-growing parts of the data-center market. For Exaion, the EDF-backed infrastructure operator, it helps move its offering from model training toward private on-premises inference, a segment that is drawing interest from companies wary of sending strategic data to foreign cloud platforms.
The servers will be delivered by the end of 2026, with 17 due this week, 33 by the end of November and the remaining 33 before year-end, making the contract unusually immediate in an industry where order announcements often stretch over multiple quarters. That timing should help 2CRSi with revenue recognition and visibility, while also supporting supply-chain relationships as it scales output of NVIDIA-based systems designed and assembled in France.
The deal also fits a wider investment pattern around AI infrastructure. NVIDIA has said it has expanded supply and capacity commitments sharply to meet future demand, and the company has introduced a business model aimed at select AI cloud partners and sovereign customers. That makes orders such as Exaion’s strategically important: they show that demand is not limited to hyperscale cloud giants but is filtering down into regional, sovereign and specialist operators building local compute capacity.
Investors will read the announcement as a positive for 2CRSi’s order book and as evidence that Europe’s sovereign AI theme is moving beyond policy rhetoric into procurement. The bull case is that more public and private operators follow Exaion’s lead, creating repeat business for niche infrastructure suppliers with technical differentiation. The bear case is that execution risk remains high: 2CRSi must deliver complex, capital-intensive systems on time, while the market for AI servers remains exposed to pricing pressure, component availability and shifting customer budgets.
The bigger implication is that AI infrastructure spending is broadening. As training clusters remain concentrated among a handful of global players, inference capacity is starting to decentralize, and that could benefit hardware vendors, datacenter operators and European suppliers positioned around data sovereignty and low-carbon power. The next test for 2CRSi is whether this contract becomes a one-off headline or the start of a more durable pipeline of sovereign AI orders.
| Entity | Gains | Losses |
|---|---|---|
| 2CRSi | ▲€60m+ order visibility | ▼Execution and delivery risk |
| Exaion | ▲Faster sovereign AI rollout | ▼Capital spending burden |
| NVIDIA | ▲More HGX B300 demand | ▼Dependence on capex cycle |
| Hyperscale cloud rivals | ▲— | ▼Share of sovereign AI demand |




