Hyperstack is active here, founded in 2023, headquarters London. Billing is per hour. Payment methods on file are credit card and invoice. The published catalogue includes H200 SXM, H100, H100 PCIe, H100 NVLink, A100, A100 NVLink, B200, L40, A6000, A4000, and RTX Pro 6000.
Why Hyperstack stands out
Our data records persistent storage, multi-GPU, Kubernetes, an API, Jupyter, Docker, reserved instances, and SOC 2. Spot is not set. The tagline matches that: NVIDIA GPUs with on-demand and reserved instances.
Pros
- Published H200, H100, A100, and B200 rows
- Per-hour billing, credit card and invoice
- Kubernetes, Jupyter, Docker, an API, persistent storage, and reserved instances recorded
- Marked SOC 2 compliant
Cons
- No spot instances in our data
- Founded in 2023, so the company is younger than the hyperscalers
- Pricing-competitiveness score is neutral. Compare the hourly rows
Getting started
- Open Hyperstack and match an H100 or A100 row to their instance.
- Pay by credit card or invoice.
- Use Kubernetes, Jupyter, or reserved capacity. Spot is not listed.
Best for: Hourly or reserved H100, H200, or A100 capacity from a London cloud with Kubernetes and a SOC 2 flag.