CUDO Compute

CUDO Compute is a GPU cloud provider headquartered in London. Third-party sources say it was founded in 2017 and started out as a crypto miner. It is an NVIDIA Cloud Partner. Its current site sells NVIDIA Hopper, Blackwell and Vera Rubin systems, including H100, H200, B200, B300, GB200 NVL72 and GB300. These come as dedicated multi-node clusters, full AI factory builds with colocation, and managed services such as 24x7 monitoring, L3 support and lifecycle management. Independent listings also describe on-demand GPU VMs in 1x, 2x, 4x and 8x sizes, 8-GPU bare-metal servers and attachable persistent volumes, all managed from a project-based web console. Facilities are in the UK, Europe, North America, APAC and the Middle East. CUDO cites ISO 27001 and SOC 2 Type II, GDPR-aligned operations and in-region data residency. On networking and storage, it supports InfiniBand and Ethernet fabrics and integrates with VAST, WEKA, DDN and NetApp. All GPU prices on the vendor's pricing page are quote on request. Shadeform reports discounts for commitments as short as one month. CUDO suits AI startups, research groups and enterprises that need GPU capacity in a particular region, especially where data sovereignty matters. Its main limitation comes from SemiAnalysis's ClusterMAX 2.0 review, which rated it Bronze. That review found no managed Kubernetes or Slurm and no shared file storage. It also found outdated base images, a shared root login with no role-based access, and interconnected GPU nodes in only one location.