Leaseweb brings NVIDIA L4 GPUs to UK sovereign cloud
Leaseweb has launched NVIDIA L4 graphics processing units on its Public Cloud platform in the UK, adding GPU resources to its sovereign cloud infrastructure in the country for the first time.
The launch extends its AI-focused infrastructure beyond dedicated servers and gives customers a new option for running GPU workloads in UK and European data centres. The NVIDIA L4 supports AI inference, machine learning, video rendering, graphics and virtual desktop infrastructure.
Instance options
Customers can choose cloud instances with one to four L4 GPUs, paired with a range of CPU and disk configurations. The service uses a pay-per-use model, with pricing billed hourly or monthly.
The offering targets variable or short-term workloads that need GPU capacity without a long-term contract. It is also aimed at organisations with data sovereignty requirements that keep workloads within UK and European sites.
Leaseweb lists 99.99% availability for the service and says it has structured its L4 instances in line with commonly used public cloud instance types. That is intended to give customers a familiar reference point when comparing configurations.
Hyperscaler comparison
Leaseweb estimates its new L4 cloud instances are more than 30% more cost-effective than comparable offerings from hyperscale cloud providers. It says the estimate is based on benchmarks of standardised workloads, and that results will vary by application and workload profile.
The launch also changes how existing customers can buy GPU resources. Leaseweb has offered monthly subscriptions for some services, and says pay-per-use billing adds another purchasing option for GPU infrastructure.
The UK release reflects broader demand for GPUs in cloud environments as organisations test and deploy AI systems. Many start with shared cloud services for experimentation, then shift to more controlled environments for production. Data residency rules and procurement constraints have also pushed some buyers towards providers that offer regional infrastructure and clearer control over where data is stored and processed.
GPU availability has become a differentiator among infrastructure providers as demand for accelerated computing has tightened supply across the market. NVIDIA positions the L4 as a general-purpose accelerator for inference and graphics-focused workloads, often used where organisations need predictable performance within constrained power and space limits.
Leaseweb is an Infrastructure as a Service provider offering Public Cloud, Private Cloud, Dedicated Servers, colocation and content delivery. It says it serves 20,000 customers and runs more than 80,000 servers across 28 data centres in Europe, Asia, Australia and North America, with network capacity of more than 10 Tbps.
Leaseweb has made "sovereign cloud" central to its positioning in Europe. The term has gained traction as governments and regulated industries seek clearer control over location, access and operational oversight. Providers have responded with regional platforms and contractual commitments on data handling and operational processes.
Customers can use the new L4 instances without the vendor lock-in it associates with hyperscale providers, but it did not provide technical details on migration paths or management tooling for moving workloads between providers.
In a statement, UK Managing Director Terry Storrar described the move as part of an expanding GPU roadmap.
"This is a significant addition to our range of GPU options available in the cloud and extends our AI-ready infrastructure beyond dedicated servers," said Storrar,
"The NVIDIA L4 GPUs offer our customers highly cost-effective flexibility to scale data-intensive workloads to precisely meet their business needs. At the same time, they can be reassured that their operations fall within a sovereign European cloud. And this is only the beginning; we will continue to expand our GPU portfolio to support customers at every stage of their AI journey."