GPU cloud, bare metal and dedicated servers compared on what one H100 really costs per hour on demand and committed, whether a small team can sign up and get a GPU without a sales call or a quota ticket, who owns the hardware and where it sits, who owns the company, and the tooling (Kubernetes, Slurm, an API, Terraform and an MCP server).
H100 instances and dedicated servers from the largest European hoster, priced in pounds with free egress, for teams that want EU hardware and a fixed bill.
H100 per GPU-hour: £2.41/h (NVIDIA H100 80 GB (h100-380))Committed H100: NoneCheapest GPU: £0.34/h (NVIDIA RTX 5000 (rtx5000-28))
RunPod rents single NVIDIA GPUs by the second in two tiers, vetted partner data centres and a marketplace of third-party hosts, to teams that sign up with a card.
H100 per GPU-hour: $3.49/h (NVIDIA H100 SXM 80 GB (Secure Cloud))Committed H100: Sales onlyCheapest GPU: $0.16/h (NVIDIA RTX A5000 24 GB (Community Cloud))
A London company selling a Kubernetes-first cloud with NVIDIA GPUs, including H100 in its London region, to teams that sign up and pay by card.
H100 per GPU-hour: $2.99/h (NVIDIA H100 SXM 80 GB)Committed H100: $2.49/h (6 mo), 17% offCheapest GPU: $1.09/h (NVIDIA A100 40 GB)
Hourly NVIDIA and AMD GPU servers you can start with a card, for UK and EU teams training, fine-tuning or serving models without a sales call.
H100 per GPU-hour: $4.41/h (NVIDIA H100 80 GB)Committed H100: $3.26/h (12 mo), 26% offCheapest GPU: $0.76/h (NVIDIA RTX 4000 Ada)
Fractional and whole NVIDIA and AMD GPUs plus single-tenant bare metal, from a US cloud running 33 regions including London and Manchester.
H100 per GPU-hour: $2.99/h (NVIDIA H100 80 GB (eight-GPU node; preemptible-only self-serve))Committed H100: Sales onlyCheapest GPU: $0.06/h (NVIDIA A16 fraction (2 GB vGPU, 2 vCPU / 8 GB))
On-demand NVIDIA GPUs by the hour from a London-registered company, with published spot prices, managed Kubernetes and data centres in Norway, Canada and Texas.
H100 per GPU-hour: $3.20/h (NVIDIA H100 SXM 80 GB)Committed H100: $2.72/h, 15% offCheapest GPU: $0.15/h (NVIDIA RTX A4000)
German dedicated root servers and single-GPU Blackwell machines, billed by the hour with unlimited traffic, for teams that want their own hardware inside the EU.
H100 per GPU-hour: €1.92/h (NVIDIA RTX PRO 6000 Blackwell 96 GB (GEX131; no H100 sold))Committed H100: NoneCheapest GPU: €0.34/h (NVIDIA RTX PRO 4000 Blackwell SFF 24 GB (GEX45))
French GPU cloud and bare metal from Paris and Warsaw, billed by the minute, for teams that want H100s and dedicated servers kept inside the EU.
H100 per GPU-hour: €2.87/h (NVIDIA H100 PCIe 80 GB (H100-1-80G))Committed H100: NoneCheapest GPU: €0.79/h (NVIDIA L4 24 GB)
GPU cloud for AI training and inference, with H100 and B200 nodes, InfiniBand clusters and one European region, booked self-serve and billed by the minute.
H100 per GPU-hour: $4.29/h (NVIDIA H100 SXM 80 GB)Committed H100: Sales onlyCheapest GPU: $0.69/h (NVIDIA Quadro RTX 6000 24 GB)
CoreWeave is a Nasdaq-listed AI cloud renting whole eight-GPU NVIDIA nodes with InfiniBand, Kubernetes and Slurm to teams training and serving large models.
H100 per GPU-hour: $6.16/h (NVIDIA H100 80 GB SXM (sold only as an eight-GPU HGX node))Committed H100: Sales onlyCheapest GPU: $1.25/h (NVIDIA L40 (sold only as an eight-GPU node))
Amazon's compute service, offering NVIDIA H100, A100, L40S and L4 GPUs, HPC instances and bare metal servers across 39 regions including London.
H100 per GPU-hour: $8.94/h (NVIDIA H100 80 GB (p5.48xlarge, eight GPUs, London))Committed H100: $3.36/h (36 mo), 62% offCheapest GPU: $1.02/h (NVIDIA L4 (g6.xlarge, London))
An Amsterdam-headquartered, Nasdaq-listed AI cloud renting NVIDIA GPUs by the second, aimed at teams that want to train and serve models without a sales call.
H100 per GPU-hour: $3.85/h (NVIDIA H100 NVLink (SXM) 80 GB)Committed H100: Sales onlyCheapest GPU: $1.55/h (NVIDIA L40S PCIe)