Hyperstack
5.8/10GPU cloud from UK-based NexGen Cloud that rents NVIDIA GPU virtual machines, Kubernetes clusters and storage to AI and machine learning teams, billed by the minute.
Data last updated November 2026 · Our methodology
Overview
Hyperstack is the GPU-as-a-service platform of NexGen Cloud, a London company. It launched in September 2023. It rents virtual machines with NVIDIA GPUs, from the low-cost A4000 up to H100, H200, B200 and B300. Larger reservations of GB200 and GB300 NVL72 clusters are also offered. Around the VMs it sells CPU-only VMs, block volumes, snapshots, firewalls, S3-compatible object storage, an On-Demand Kubernetes service with a free master node, and a dedicated inference service. An AI Studio for fine-tuning models is in development. Capacity sits in three regions: CANADA-1, NORWAY-1 and US-1. Pricing is usage-based. On-demand VMs are paid from a prepaid credit-card balance and billed per minute, at listed rates from $0.15 per hour for an A4000 to $7.40 per hour for a B300. Spot instances cost less but can be reclaimed. Reserved capacity is cheaper again and invoiced monthly. Ingress and egress traffic are free, which is unusual and helps teams that move large datasets. Developers can automate the platform through a REST API, Python, Go and JavaScript SDKs, and a Terraform provider that is still in alpha. The parent company holds a SOC 2 Type II attestation. ISO 27001 is still in progress. It suits startups and AI teams that want simple self-serve GPU VMs without the overhead of a hyperscaler. The main limitation is reliability beyond single VMs. SemiAnalysis rated it Bronze after Kubernetes clusters failed to provision. User reviews also report scarce GPU stock, slow support and billing surprises when the prepaid balance runs out.
Our Rating
AI Software6/10
Hosting & Infrastructure5.7/10
Who Hyperstack is best for
Best software for DevelopersGood fit
There is a REST API, Python, Go and JavaScript SDKs, and a Terraform provider, and reviewers praise the fast portal for spinning VMs up and down.
Self-serve sign-up, per-minute prepaid billing and low entry GPU prices suit AI startups. Hyperstack also mentions NVIDIA Inception discounts.
Less suited
It works for inference and training back ends, but there are only three regions and no managed databases or CDN.
Reserved clusters and SOC 2 Type II help, but analyst findings on Kubernetes reliability and mixed reports on support are concerns.
It is a specialised GPU cloud, not general web hosting, so it fits only small businesses that run AI workloads.
Pros & Cons
Pros
- Free ingress and egress traffic, so teams moving large datasets avoid the transfer fees most clouds charge.
- Per-minute billing on a wide range of GPUs, from a $0.15/hr A4000 up to B200 and B300.
- SemiAnalysis found the web console fast and easy for launching and tearing down GPU VMs.
- Automation through a REST API, Python, Go and JavaScript SDKs, and a Terraform provider.
- Free Kubernetes master node, plus spot and reserved options that cut costs.
Cons
- SemiAnalysis ClusterMAX rated it Bronze after two Kubernetes clusters failed to provision, one after hours stuck in Creating.
- Trustpilot score is about 2.4 out of 5 from only 9 reviews, with complaints about slow support, GPU stock and refunds.
- Billing is prepaid. Volumes keep charging after the balance runs out, and resources can be deleted about a week after the balance is used up.
- Object storage is only in CANADA-1, and the Terraform provider is still in alpha.
- No managed databases, CDN or serverless functions, so general application hosting needs another provider.
- ISO 27001 is not yet achieved, and the SLA page shows conflicting uptime figures (99.5% and 100%).
Key Features
Hyperstack Pricing
On-Demand
- Billed per minute
- Free ingress and egress
- Email alerts as the balance is used up
- CPU VMs from $0.35/hr (4 cores, 4 GB RAM)
- GPU VMs from $0.15/hr (A4000) to $7.40/hr (B300)
- Prepaid balance paid by credit card through Stripe
- H100 PCIe $2.50/hr, H100 SXM $3.20/hr, H200 SXM $3.99/hr, B200 $6.00/hr
Reserved
- Pricing through sales
- Contracted capacity, invoiced monthly
- GB200 and GB300 NVL72 clusters by reservation
- Starting rates: H100 $1.75/hr, H100 SXM $2.72/hr, H200 SXM $2.79/hr, B200 $5.10/hr
Spot
- L40 $0.80/hr
- A100 $1.08/hr
- A6000 $0.40/hr
- H100 PCIe $2.00/hr
- RTX Pro 6000 SE $1.48/hr
- Surplus capacity that can be reclaimed
Frequently Asked Questions
On-demand use is paid from a prepaid balance by credit card through Stripe and billed per minute. Reservations are invoiced monthly. Email alerts go out at 50, 70, 80, 90 and 100 percent of the balance used. Resources not recharged within about a week may be deleted.
No. The pricing page lists ingress and egress traffic as free. Public IPs cost about $0.0067 per hour.
Three: CANADA-1 in Quebec, NORWAY-1 in Norway and US-1 in Texas. Object storage is currently only in CANADA-1.
The pricing page shows no free tier or trial credits. The only free items listed are data transfer and the Kubernetes master node.
Its parent, NexGen Cloud, announced a SOC 2 Type 2 attestation in 2026. ISO 27001 is listed as in progress, and Hyperstack says its data centres are Tier 3.
Approved claims earn pro-rata account credits, capped at the fees for the affected service in that billing cycle. Claims must be emailed within 5 days after the billing cycle ends. Customers may terminate without penalty if uptime is missed three months in a row. The page shows both 99.5% and 100% as the uptime figure, so confirm which applies.
Not yet. Clusters can be scaled manually through the console or API, and Hyperstack says a cluster autoscaler is coming soon.
Similar Tools
See all alternatives
Amazon's public cloud platform, with over 200 pay-as-you-go services for compute, storage, databases, networking and AI, used by startups through large enterprises and governments.
DigitalOcean is a developer-focused cloud platform offering Droplet VMs, managed Kubernetes, databases, object storage, App Platform and GPU compute for startups and small businesses.
French cloud provider offering shared web hosting, VPS, dedicated servers and an OpenStack-based public cloud for businesses and developers.
Ranked near Hyperstack in AI Software
See Hyperstack in the listHyperstack is #280 in AI Software by our score.
