# Hyperstack

Official pricing: https://www.hyperstack.cloud/gpu-pricing  
Category: gpu-cloud · Isolation: vm

## Pricing regimes (raw)

- **H100 PCIe 80GB (n3-H100x1) on-demand** (resource): $0/vCPU-h, $0/GiB-h
- **A100 80GB PCIe (n3-A100x1) on-demand** (resource): $0/vCPU-h, $0/GiB-h
- **RTX Pro 6000 SE (n3-RTX-PRO6000-SEx1) on-demand** (resource): $0/vCPU-h, $0/GiB-h
- **L40 (n3-L40x1) on-demand** (resource): $0/vCPU-h, $0/GiB-h
- **RTX A6000 (n3-RTX-A6000x1) on-demand** (resource): $0/vCPU-h, $0/GiB-h
- **RTX A4000 (n3-RTX-A4000x1) on-demand** (resource): $0/vCPU-h, $0/GiB-h
- **H100 SXM5 (n3-H100-SXM5x8) 8-GPU node, on-demand** (resource): $0/vCPU-h, $0/GiB-h
- **H100 NVLink (n3-H100x8) 8-GPU node, on-demand** (resource): $0/vCPU-h, $0/GiB-h

## Features

Yes: snapshots, pause/resume, persistent disk, whole disk survives a stop, what survives a stop, volumes, ≥24 h sessions, custom image (Docker or snapshot), start from your own snapshot, full VM (own kernel), Docker inside, root, SSH, public IPv4, egress allowlist, open internet, static egress IP, SOC 2, GPU, Python, Node.js, EU data residency, MCP server, snapshots on demand, gVisor or VM (no shared kernel), inbound access rules, firewall inside (nftables), extra volumes, FUSE mounts, kernel features (eBPF, mounts, cgroups)

No: memory snapshots, fork/clone, idle auto-stop, your own Docker/OCI image, browser, desktop GUI, computer-use API, code interpreter, browser + desktop control, anti-bot stealth, CAPTCHA solving, residential IPs, preinstalled agents, HTTPS preview URLs, custom domains, raw TCP inbound, self-hosting, BYOC, open source, HIPAA, SSO, arm64, Windows, macOS, GPU desktop, HTTP method/path egress rules, wake on request, live resize, live fork (no pause), memory fork, usage per sandbox (API), automatic snapshots, agent harness API, hosted agent API (their own agent), secret proxy, secret proxy for any API, shared volumes

Unknown: everything else. Evidence (source + quote) per feature: https://battleships.dev/data/providers/hyperstack.json → feature_evidence

## Company

- Ownership: independent
- Stage: series-a [source](https://www.nexgencloud.com/news/nexgen-cloud-secures-45-million-in-series-a-funding-to-expand-sovereign-ai-infrastructure-in-europe)
- Accelerators: none found
- Total raised: $64.4M (estimated: Compworth aggregator: 'NexGen Cloud has completed 3 funding rounds, raising $64.4M in total.' No primary source states a cumulative equity total; the confirmed Series A alone is $45M.) [source](https://compworth.com/company/nexgen-cloud)
- Last round: series-a, $45M, 2025-04 (confirmed) [source](https://www.nexgencloud.com/news/nexgen-cloud-secures-45-million-in-series-a-funding-to-expand-sovereign-ai-infrastructure-in-europe)
- Investors: High-net-worth individuals and family trusts (unnamed)
- Revenue: $91.3M (confirmed: £72 million converted at ~1.27 USD/GBP (approximate); figure is the company's own stated revenue for FY2023-2024.) [source](https://finance.yahoo.com/news/nexgen-cloud-secures-45-million-134800660.html) annual revenue, 2023-2024

Hyperstack is the GPU-cloud brand of NexGen Cloud, a UK (London-based) company founded 2020, still independent/privately held (no acquisition or parent found). No accelerator/incubator affiliation found. In addition to the $45M Series A (Apr 2025), NexGen Cloud also announced a separate $34M non-recourse debt facility from USD.AI (June 2026) to finance GPU hardware deployment in Sweden — debt, not equity, so excluded from funding stage/round figures above but worth noting to a buyer as another capital source. No accelerator batch, no public IPO; total funding figure is a third-party estimate (Compworth) not corroborated by a primary source.

## Caveats

- Per-GPU vCPU/RAM are "max per GPU" figures from the pricing table; region availability varies by flavour.
- Volume storage $0.000096774/GB-h (~$0.07/GB-month); public IP $0.00672043/h; egress/ingress free.
- Reserved pricing (from $1.75/h H100 PCIe) not encoded (contract).
- Flavor docs list the 1x H100 PCIe in CANADA-1 only (region class "other"); other regions not verified.
- SHUTOFF (stopped) VMs are billed at the full flavor rate, GPU included; only HIBERNATED (root disk ~$0.07/GB-month + retained IP) or DELETED stop GPU billing. Restoring from hibernation needs the flavor in stock.
- Prepaid: to deploy, balance must cover all running resources plus the new VM for 1 hour. At $0 balance VMs are hibernated (or deleted if the flavor can't hibernate: spot, large-root-disk, K8s clusters) and billing continues into negative balance.
- API rate limit 500 requests/minute per source IP across all endpoints (raise via support). No published VM-count quota or start-rate limit.
- Object storage $0.000019397/GB-h (~$14.29/TB-month). Kubernetes: worker nodes at flavor rate + public IPs; master/LB/bastion not billed.

## How this provider charges


GPU cloud. Added by the missing-providers audit (single-H100 SKUs).

| Regime | When it applies | How billed | Numbers | Source |
|---|---|---|---|---|
| On-demand 1x | 1x flavours | per minute | H100 PCIe $2.50, A100 PCIe $1.35, L40 $1.00, A6000 $0.50 /GPU-h | https://www.hyperstack.cloud/gpu-pricing |
| On-demand 8x | SXM/NVLink/H200/B200 | per minute, 8-GPU flavours | H100 SXM $3.20, H100 NVLink $2.60, H200 $3.99, B200 $6.00 /GPU-h | https://www.hyperstack.cloud/gpu-pricing |
| Spot | spot VMs | per minute, interruptible | H100 PCIe $2.00 | https://www.hyperstack.cloud/gpu-pricing |
| Reserved | contract | monthly invoice | H100 PCIe from $1.75 | https://www.hyperstack.cloud/gpu-pricing |

## Gotchas
- SXM H100 only as 8x.
- Storage and IPv4 billed hourly; egress free.

## Worked example
1x H100 PCIe, 500 sessions x 30 min = 250 h x $2.50 = $625/month (+ volume storage).

Sources: https://www.hyperstack.cloud/gpu-pricing, https://docs.hyperstack.cloud/docs/hardware/flavors