European AI cloud provider NexGen Cloud has secured a $106 million senior secured loan to finance a new GPU deployment at a Tier III data centre in Finland. The facility, led by GPU lender USD.AI with Silver Point Capital and Victory Park Capital, is the latest sign that lenders are treating AI hardware and its contracted revenue as a financeable asset class.
What happened
USD.AI announced on 29 October that it had led and closed the four-year term loan, which amortises over its life. The money will pay for 216 Nvidia HGX B300 servers, a total of 1,728 GPUs. The capacity is fully contracted under a take-or-pay agreement with what the companies describe as a leading AI inference platform, which they did not name. The Finnish data centre operator was also not disclosed.
Funding sits in escrow and will be released as the servers ship, which is expected in the fourth quarter of 2026. NexGen Cloud, an Nvidia Cloud Partner that runs the Hyperstack platform, hosts infrastructure in data centres across Europe and North America.
The key numbers
- $106 million senior secured term loan, four-year tenor
- 216 Nvidia HGX B300 servers, 1,728 GPUs
- $140 million in total USD.AI financing for NexGen Cloud, including a $34 million facility for a B200 deployment in Sweden announced in June
- More than $500 million of GPU financing originated by USD.AI this year, over $200 million of it through its GPU Loan Exchange
How the financing works
The deal ran through USD.AI’s GPU Loan Exchange (GLX), launched on 1 October. USD.AI originates and structures loans secured on GPUs and the contracted cash flows they generate, then places them with institutional credit investors. Its first GLX transaction was a $98 million deal for Duos Edge AI.
USD.AI co-founder and chief operating officer Conor Moore pitched the platform at what he called the “largely underserved Compute Middle Market”. NexGen Cloud chief executive Chris Starkey said the route gives the company speed and certainty in turning customer demand into deployed capacity.
Why it matters for modular and edge in Europe
Most European AI deployments below hyperscale size are not new campuses but GPU clusters placed in existing colocation halls or in factory-built modules. For these projects, the hardware often costs more than the building, and access to equipment finance can decide whether a deployment goes ahead.
Loans secured against servers and backed by take-or-pay contracts give mid-sized operators a way to fund capacity without large equity raises. The same structure could suit containerised and modular AI deployments, where the IT load and the enclosure are delivered as a single package and can be relocated. Lenders will still focus on the quality of the offtake contract and on how quickly GPU values fall. NexGen’s choice of Finland and Sweden also underlines continued interest in Nordic sites with available power, even as grid debates intensify across the region.
Leave a Reply