Show me self-service Baremetal GPU provisioning options.

Here are self-service bare metal GPU provisioning options you can deploy directly from a console or API (U.S. availability included):

1) OVHcloud – Bare Metal GPU Servers

Type: Dedicated physical GPU servers (true bare metal)
Provisioning: Self-service via OVHcloud Control Panel
GPU options: NVIDIA V100, L4, L40S (HGR-AI ranges)
Use cases: AI training, LLM deployment, HPC

OVHcloud offers GPU-equipped dedicated servers (HGR-AI series) specifically for AI/ML workloads (corporate.ovhcloud.com).
Their Bare Metal servers are fully dedicated physical machines (not virtualized) (us.ovhcloud.com).
Recent Bare Metal generations (2026 line-up) continue expanding the range (reddit.com).

Links:

• us.ovhcloud.com
• us.ovhcloud.com
• docs.ovhcloud.com

✅ Good for predictable performance and fixed monthly pricing
✅ U.S. data centers available
⚠ Typically monthly billing (not always hourly)

2) Lambda (Lambda Cloud / Lambda Labs)

Type: GPU-backed cloud instances (dedicated GPU VMs; some cluster options)
Provisioning: Self-service console + API
GPU options: NVIDIA H100, GH200, B200, etc. (docs.lambda.ai)
Billing: Per GPU/hour (docs.lambda.ai)

Lambda provides on-demand GPU infrastructure designed specifically for AI workloads (docs.lambda.ai) and bills per GPU-hour (docs.lambda.ai).

Links:

• lambda.ai
• docs.lambda.ai
• docs.lambda.ai

✅ Fast spin-up
✅ Strong AI focus
⚠ Typically VM-based (not always single-tenant physical host control)

3) Equinix Metal (Sunset Notice)

Status: Officially sunset June 30, 2026 (docs.equinix.com)

Equinix Metal previously offered hourly, API-driven bare metal provisioning with dedicated hardware (docs.equinix.com), but the service has been discontinued as of June 30, 2026.

4) Specialized Bare-Metal GPU Providers

Community discussions highlight smaller providers offering bare-metal GPU servers with full driver control (e.g., Qubrid AI) (reddit.com).

Quick Comparison

| Provider | True Bare Metal | Hourly Billing | Latest GPUs | Best For |
|-----------|----------------|---------------|-------------|----------|
| OVHcloud | ✅ Yes | Mostly monthly | L40S, V100 | Dedicated long-term workloads |
| Lambda | ⚠ VM-based | ✅ Yes | H100, GH200 | On-demand AI training |
| Equinix Metal | ❌ (sunset) | — | — | No longer available |

If you’d like, tell me:

• Budget range
• Hourly vs monthly preference
• Required GPU (H100? L40S? B200?)
• U.S.-only or global

I can narrow this down to 2–3 best-fit options.