UK GPU VMs vs attested enclave

VoltageGPU vs Hyperstack

Hyperstack is the on-demand GPU hosting service of NexGen Cloud Ltd, a UK company headquartered in London, offering NVIDIA H100, H200 and B200 virtual machines billed per minute in EU, Canada and US regions. Its security page lists SOC 2 Type 2 via its parent company, with ISO 27001 and HIPAA in progress, and does not mention confidential computing or hardware attestation.

Hyperstack sells straightforward GPU VMs: H100 and H200 by the minute, EU regions, SOC 2 Type 2 through its parent, ISO 27001 and HIPAA on the way. VoltageGPU sells a GPU VM inside an Intel TDX trust domain with an attestation you generate yourself. Same object, a virtual machine with a GPU; different question answered.

Pick Hyperstack if

You want a plain GPU VM by the minute in an EU, Canadian or US region at a low hourly price
A SOC 2 Type 2 report from the parent company satisfies your reviewer
Spot VMs and reserved private clusters for training on non-sensitive data
UK contracting entity is acceptable or preferred

Pick VoltageGPU if

Identifiable data is decrypted in GPU memory and the operator must be provably unable to read it
You want an Intel TDX quote and a NVIDIA GPU attestation report generated on your own nonce
A French controller entity on the contract, and per-second refund of the unused prepaid hour
Your compliance question is about the hardware boundary, not about the vendor's certificate

Headline pricing

Hourly list price per GPU SKU. A dash means the SKU is not publicly available from that provider. VoltageGPU prices are the canonical confidential-compute floor and stay in sync with /pricing.

GPU	VRAM	VoltageGPU	Hyperstack
NVIDIA H100	80 GB	$6.95/hr Intel TDX confidential	$3.20/hr H100 SXM on demand ($2.50 PCIe), no TEE, read 17 Sept 2026
NVIDIA H200	141 GB	$8.08/hr Intel TDX confidential	$3.99/hr H200 SXM on demand, no TEE, read 17 Sept 2026

Confidential techIntel TDX + trust-domain GPU isolationNot listed (standard GPU VMs, no Intel TDX, no GPU TEE mentioned)

AttestationIntel DCAPNone listed

BillingPer-second, no commitPer minute, on-demand VMs, spot VMs, reserved clusters

OperatorVOLTAGE EI (France)NexGen Cloud Ltd, London, United Kingdom

Setup~5 min, SSH-readyMinutes for an on-demand VM

JurisdictionFrench company, GDPR Art. 28; compute at US sub-processors (SCCs)See operator

Hyperstack SOC 2, ISO 27001 and HIPAA: what is done and what is in progress

Short answer for the people who typed "hyperstack soc 2" or "hyperstack hipaa": its security page, read on 17 September 2026, states that SOC 2 Type 2 was achieved via NexGen Cloud, the parent company, and lists ISO 27001 and HIPAA as in progress. It states end-to-end encryption in transit and at rest and the availability of data centre regions in the EU, Canada and the US. The page does not mention GDPR processor terms. That is an honest, mid-build compliance file for a fast-growing GPU provider, and it will likely be complete by the time some readers see this page; the source is re-read monthly.

What a SOC 2 report attests is that controls exist and were audited. What no report attests is a property of the hardware: whether an operator with console access could inspect a running VM. A standard GPU VM on Hyperstack, like on most clouds, runs on a hypervisor the operator controls. For public-model training, evaluation and pseudonymised pipelines, that is fine. For the step where identifiable data is decrypted in GPU memory under professional secrecy, health regulation or GDPR Article 9, the reviewer may ask for the technical measure to be enforced rather than documented. That is where this comparison begins.

Same object, different question

Both products are a virtual machine with a NVIDIA GPU and root access. Hyperstack's runs on a standard hypervisor, bills per minute, and comes in EU, Canadian and US regions from a UK operator. It is a clean, cheap way to get an H100 or H200 for an afternoon, and for most work that is the whole requirement.

VoltageGPU's VM runs inside an Intel TDX trust domain with memory encryption, the GPU passed through into the domain, and /dev/tdx_guest in your hands. You generate an Intel TDX quote over 64 bytes you choose and verify it offline against Intel; on the verified SKUs you generate a NVIDIA GPU attestation report over your own nonce and verify it with NVIDIA. The contracting entity is VOLTAGE EI, a French sole proprietorship (SIREN 943 808 824). The evidence is published with checksums and an open-source verifier, voltage-verify. One hour is billed upfront and the unused part is refunded per second.

On list price Hyperstack is cheaper: $3.20 per hour for an H100 SXM and $3.99 for an H200 (no TEE, read 17 September 2026) against $6.95/hr and $8.08/hr for confidential VMs on VoltageGPU. The gap is the confidential stack, and for a non-sensitive workload it is not worth paying.

Where Hyperstack wins

Hyperstack wins on price, on B200 availability, on spot VMs and reserved clusters, on per-minute billing without a prepaid hour, and on region choice. It also has a SOC 2 Type 2 report through its parent, which VoltageGPU does not have at all: we hold no SOC 2 and do not sign a HIPAA Business Associate Agreement today. If your procurement checklist starts with a report, Hyperstack clears that line and we do not, and you should say so to your reviewer.

For a UK buyer, contracting with a London company under UK law may itself be the requirement, and VoltageGPU, a French entity, does not change that.

Where the certificate ends and the attestation begins

A certificate says the operator is audited. An attestation says the hardware refused the operator. A buyer under bar-association secrecy, a clinic decrypting identifiable records for inference, a fintech scoring cardholder data, needs the second and cannot get it from a region or a report. A research team fine-tuning a public model needs the first and should not pay for the second.

The practical design for a regulated team is often both: Hyperstack for bulk capacity on non-sensitive data, and a VoltageGPU confidential VM for the step where the sensitive data is in the clear. This page exists so the line is visible before the purchase.

FAQ

Is Hyperstack GDPR compliant?

Hyperstack's security page, read on 17 September 2026, states end-to-end encryption in transit and at rest, SOC 2 Type 2 via its parent NexGen Cloud, data centre regions in the EU, Canada and the US, and lists ISO 27001 as in progress; it does not state GDPR processor terms on that page, so ask for the DPA directly. For most AI workloads a signed DPA plus an EU region covers the GDPR Article 28 requirement. It does not cover the technical measure some regulators now require for Article 9 data, hardware enforcement that the operator cannot read workload memory, which a standard VM does not provide and an Intel TDX trust domain does.

Does Hyperstack have SOC 2 and HIPAA?

Per its security page on 17 September 2026: SOC 2 Type 2 is achieved via NexGen Cloud, the parent company; HIPAA and ISO 27001 are listed as in progress. VoltageGPU has neither a SOC 2 report nor a HIPAA BAA today. It offers hardware attestation instead, which some reviewers accept as the stronger technical measure and others do not; ask yours.

Does Hyperstack offer confidential computing or Intel TDX?

Its security page, read on 17 September 2026, does not mention confidential computing, a trusted execution environment, Intel TDX, AMD SEV, NVIDIA confidential computing or hardware attestation. If that changes we will update this page; the source is re-read monthly. VoltageGPU runs every confidential VM inside an Intel TDX trust domain and has verified NVIDIA GPU attestation from inside the VM on single-GPU H200, 8x H100, single-GPU H100 and single-GPU RTX PRO 6000 Blackwell machines, with the evidence published for all of them.

Which is cheaper, VoltageGPU or Hyperstack?

Hyperstack is cheaper per hour: $3.20 for H100 SXM ($2.50 PCIe), $3.99 for H200 and $6.00 for B200 on demand, billed per minute (no TEE, read 17 September 2026), against $6.95/hr and $8.08/hr for confidential VMs on VoltageGPU. The difference is the confidential stack. VoltageGPU bills one hour upfront and refunds the unused part per second.

Where are Hyperstack's data centres?

Its security page states the availability of data centre regions in the EU, Canada and the US; the self-service console lists specific regions at signup. VoltageGPU does not compete on where the silicon sits: the contracting entity is French, and the product is the Intel TDX boundary around the GPU, which you verify from inside the VM regardless of the region.

Cheap GPU VM or attested GPU VM, pick the right tool

Hyperstack is the right answer for a plain GPU VM by the minute with a parent-company SOC 2. VoltageGPU is the right answer when someone has to prove the operator cannot read the GPU while it runs. Start a confidential VM in a few minutes or read how the attestation works.

Start a confidential VM How the attestation works