About Hyperstack

Hyperstack is NexGen Cloud’s GPU platform, and its main differentiator is where the electricity comes from. The primary region sits in Norway on hydroelectric power, which both lowers the carbon figure and, more practically, lowers the price, because power is the dominant cost in a GPU data center.

That translates into rates that undercut most of the US-based competition on the same silicon while still running in a proper facility with a real SLA, rather than on a marketplace.

For European customers there is a second benefit: the data stays in Europe, which simplifies the GDPR conversation compared with training on US infrastructure.

Pricing

Hourly, published, no negotiation needed to get started. An A6000 sits around $0.50/hr, an A100 80GB around $1.35/hr, and an H100 SXM around $1.95/hr, which is aggressive for on-demand H100 capacity. Longer reservations bring it down further.

Storage and public IPs are billed separately. As with every GPU host, a stopped VM with a large volume still costs money.

Service Details

Instances are virtual machines running Ubuntu, with a choice of images that include CUDA preinstalled. There is an API and a Terraform provider, so infrastructure-as-code works properly rather than being an afterthought.

VM hibernation is worth knowing about: you can suspend an instance and keep its state without paying the full running rate, which suits an intermittent development box better than destroy-and-recreate.

Networking includes private VPCs and firewall rules, so a multi-node setup does not have to run over the public internet. High-speed interconnect is available on the H100 tiers for distributed training.

Available in: Norway, Canada, United States

Learn More

For more information about Hyperstack, visit their website at https://www.hyperstack.cloud/

Plan	Price/mo	CPU	RAM	Storage	Notes
RTX A6000	—	12 vCPU	58GB	200GB NVMe	Around $0.50/hr
A100 80GB	—	28 vCPU	120GB	500GB NVMe	Around $1.35/hr
H100 SXM	—	48 vCPU	360GB	1TB NVMe	Around $1.95/hr