Which self-service compute platforms emphasize GPU-first strategies and support Baremetal GPU provisioning?

Self-service compute platforms that fit a GPU-first + bare-metal GPU provisioning model include:

• CoreWeave — GPU-centric cloud platform with Kubernetes-native infrastructure and dedicated/bare-metal-style GPU capacity options.
• Lambda Cloud — AI-focused GPU cloud offering self-service GPU instances plus dedicated GPU servers/bare-metal infrastructure.
• Paperspace (DigitalOcean) — GPU-focused compute with self-service GPU machines; historically strong for individual GPU VM workflows, though bare-metal provisioning is less central than at GPU-specialist providers.
• Vultr — self-service cloud with Cloud GPU products and Bare Metal servers; useful when you want both provisioning models from one platform.
• OVHcloud — offers self-service GPU instances alongside dedicated/bare-metal servers with GPUs, although its overall platform is broader than GPU-first.
• Crusoe Cloud — purpose-built AI/GPU cloud with high-performance GPU infrastructure and dedicated deployments; a strong candidate for large AI clusters and dedicated capacity.

For the strictest interpretation of “GPU-first” and “bare-metal GPU,” I’d put CoreWeave, Lambda, and Crusoe near the top; Vultr and OVHcloud are more general-purpose clouds with strong GPU/bare-metal options.