---
title: "Civo review: pricing, pros and cons"
description: "A London company selling a Kubernetes-first cloud with NVIDIA GPUs, including H100 in its London region, to teams that sign up and pay by card."
canonical: "https://saaskly.com/category/gpu-compute/civo"
last-updated: "2026-09-22"
---

# Civo review: pricing, pros and cons

> Saaskly verdict: **3.9/5**. A London company selling a Kubernetes-first cloud with NVIDIA GPUs, including H100 in its London region, to teams that sign up and pay by card.

- Category: [GPU & High Performance Compute](https://saaskly.com/category/gpu-compute.md)
- Vendor website: https://www.civo.com
- Last updated: 2026-09-22

## Pros and cons

**Pros**

- H100 SXM at $2.99 an hour in the London region, the only one of Civo's five regions with GPUs, so the reference GPU and its data stay in the UK
- Data transfer is free and unlimited with no allowance and no overage rate, and the published SLA is 99.95% with a 100% credit of the monthly fee if uptime falls below 95%
- Committed rates are published rather than quoted: terms of 6, 12, 24 and 36 months, with the H100 at $2.49 an hour and the A100 40 GB at $0.69 an hour on 36 months

**Cons**

- A stopped instance is billed in full, in Civo's own words "exactly the same as an active instance", billing is hourly rounded up, and an unpaid invoice means suspension after 8 days and deletion of resources after two weeks
- Quota numbers are published nowhere: all accounts start on a basic level and Civo raises limits "after a period of proving that existing account quota is being handled correctly", so a new account may not be able to launch an eight-GPU node on day one
- Prices are US dollars excluding VAT from a UK company, SMS phone verification is mandatory before anything can be launched, some countries are excluded by that step, and virtual and prepaid cards are refused
- GPUs are in London only, there is no spot or preemptible option, no cluster fabric such as InfiniBand is mentioned anywhere, and the GPU instance specs for vCPU, RAM and local disk are not published

## Pricing

Month-to-month price first (what you pay without committing), then the annual-commit rate.

- **H100 SXM 80 GB, one GPU**: $2,183/mo month-to-month; $1,818/mo on an annual commitment (month-to-month costs 20% more)
  - Includes: One NVIDIA H100 SXM 80 GB at $2.99 an hour on demand, $2,183 a month at 730 hours; $2.49 an hour on a 36-month commitment, $1,818 a month, 17% lower. Terms of 6, 12, 24 and 36 months are offered but only the 36-month rate is published. H100 PCIe is $2.49 an hour on demand. London region only. USD, excluding VAT, read 22 Sep 2026.
- **A100 40 GB, one GPU**: $796/mo, same price on any term
  - Includes: One NVIDIA A100 40 GB at $1.09 an hour, $796 a month at 730 hours, falling to $0.69 an hour on a 36-month commitment. Volumes $0.11 per GB a month, object storage $0.010860 per GB a month, data transfer free and unlimited. A stopped instance is billed in full.

Minimum commitment: **$2,183** ($2,183/mo × 1 seat × 1 month (no commitment)).

## Specification

| Attribute | Value |
| --- | --- |
| H100 per GPU-hour* | $2.99/h (NVIDIA H100 SXM 80 GB) |
| Committed H100 | $2.49/h (6 mo), 17% off |
| Cheapest GPU | $1.09/h (NVIDIA A100 40 GB) |
| 8xH100 node | $23.92/h |
| Spot / interruptible | No |
| Cheapest dedicated server | None |
| Billing | Per hour |
| Egress | Included |
| Storage (per GB/mo) | $0.110 |
| Self-serve signup | Yes |
| GPU quota | Quota ticket |
| Hardware** | Own hardware, colocated |
| UK location | Yes |
| H100 in the UK | Yes |
| EU locations | 1 |
| Locations | 5 |
| Owned in | UK-owned |
| Owner | Independent |
| Cluster networking | No |
| Kubernetes | Yes |
| Slurm | No |
| Bare metal | No |
| API | Yes |
| Terraform | Yes |
| MCP server | No |
| Uptime SLA (%) | 99.95 |
| Trustpilot (reviews) | 4.0 (3) |
| Committed pricing | Published |
| Committed term (months) | 6 |
| Stopped instance billed | Yes |
| Free credit | $250 when a valid card is added within 21 days of signup, valid for one month from the date it is applied |
| UK cities | London |
| HQ | UK |
| Publicly listed | No |
| Sovereign offering | Dedicated sovereign cloud |
| ISO 27001 | Yes |
| SOC 2 | Yes |
| NVIDIA GPUs | Yes |
| AMD GPUs | No |
| Multi-GPU nodes | Yes |
| Dedicated servers | No |
| Object storage | Yes |
| Block storage | Yes |
| Private networking | Yes |
| CLI | Yes |
| 2FA | No |
| Phone support | No |
| Support channels | Ticket, Email, Slack community |

* One NVIDIA H100 80 GB per GPU-hour, on demand, in the currency the vendor shows a UK visitor; where the vendor sells no H100 the nearest GPU is named. Committed prices are the cheapest published rate and its minimum term.

** Who owns and runs the GPUs. "Third-party hosts" means the vendor is a marketplace for other people's machines and the SLA is theirs, not the vendor's.

## Our review

## Who owns it

Civo Limited is a UK company registered in England and Wales as number 09568551, incorporated on 30 April 2015 and named Absolute Devops Ltd until 9 April 2019. Its registered office is First Floor, 32-37 Cowper Street, London EC2A 4AW, and its Companies House filings are current: accounts to 30 April 2025 with the next due on 31 January 2027, and a confirmation statement filed on 2 June 2026. It is private and founder-controlled. The only person with significant control is Mark Alan Boost, the chief executive and a director since incorporation, holding more than 50% and less than 75% of the shares and voting rights, notified in April 2016. Co-founder Barry Fenning resigned as a director in June 2021 and Andy Jeffries in July 2022. The board has since added Simon Hansford in September 2024, Kelsey Hightower in February 2025 and Benjamin Warman in April 2026. No funding total or investor list could be confirmed from a primary source, so none is given here. Civo runs five public regions and does not say who owns the buildings behind them, but it is now building its own: on 22 September 2026 it announced it had broken ground on a liquid-cooled edge data centre in Hertfordshire, opening with 8 MW and expandable to 38 MW, as the first of a planned 40 sites totalling 1 GW, with five further sites securing 150 MW across Lancashire, Greater Manchester, London and Oxfordshire over three years. It does not resell another cloud's capacity and runs no marketplace of third-party hosts. A UK or EU buyer contracts with Civo Limited, a UK company, with no foreign intermediary.

## What it costs

Prices are in US dollars and exclusive of VAT, which Civo states plainly in its billing documentation, so a UK visitor is quoted dollars rather than sterling. Read on 22 September 2026, one NVIDIA H100 SXM 80 GB costs $2.99 an hour, or $2,182.70 a month at 730 hours, and falls to $2.49 an hour on a 36-month commitment, $1,817.70 a month, a discount of 17% against the hourly rate. The H100 PCIe variant is $2.49 an hour on demand. Commitment terms of 6, 12, 24 and 36 months are offered, but only the 36-month rates are published on the product pages. GPUs are sold in sizes of one, two, four and eight, so an eight-GPU H100 configuration works out at $23.92 an hour. H200 SXM is $3.49 an hour and B200 is $3.79. The cheapest GPU is the A100 40 GB at $1.09 an hour, $795.70 a month, dropping to $0.69 an hour on a 36-month term, which is the source of Civo's headline rate. There is no spot or preemptible option. CPU-only compute starts at $0.007440 an hour, $5.43 a month. Civo does not publish the vCPU, RAM or local disk that comes with each GPU size, and it does not mention InfiniBand or any cluster fabric anywhere.

Billing is hourly, rounded up to the nearest full hour, with the monthly figure acting as the cap for continuous use. The important detail is that a stopped machine still costs full price: Civo's documentation says a stopped instance or cluster "is exactly the same as an active instance", and it will not shut anything down for you unless the account is suspended. Volumes are $0.11 per GB a month, object storage $0.010860 per GB a month, and a load balancer $10.86 a month per 10,000 concurrent requests. Data transfer is free and unlimited, with no allowance and no overage rate, which removes the usual source of a surprise bill. Billing is postpaid, invoiced on the first of the month for the previous month, by card through Stripe only, with American Express, Mastercard and Visa accepted and virtual and prepaid cards refused. If an invoice goes unpaid for 8 days the account is suspended, and after two weeks the resources are deleted. New accounts get $250 of credit if a valid card is added within 21 days of signing up, and that credit lasts one month from the date it is applied. For the persona here the smallest realistic monthly bill for GPU work is one A100 40 GB left running for a month plus a 100 GB volume: $1.09 x 730 hours = $795.70, plus 100 GB x $0.11 = $11, so about $807 a month with nothing to pay for transfer.

## What you get and what to watch

Signup is self-serve but there are two steps to clear. You confirm your email, then you must verify a mobile number by SMS before you can add a payment method or launch anything, and Civo says certain countries are not supported by that verification and it cannot serve them. A valid card is required to create an instance or a cluster. Accounts are usually verified automatically, and if that fails the account stays pending until a person reviews it. The bigger unknown is quota. Civo publishes no numbers at all, saying only that all customers start on a basic quota level covering instances, Kubernetes nodes, CPUs and RAM, and that increases are requested from the dashboard with an explanation. Its own wording is that it is "generally happy to increase user quotas after a period of proving that existing account quota is being handled correctly", and that prior payment history helps, so a brand new account should not assume it can launch an eight-GPU node on day one. Quotas are account-wide rather than per region, which at least makes moving between regions simpler.

There are five regions: LON1 in London, FRA1 in Frankfurt, NYC1 in New York, PHX1 in Phoenix and MUM1 in Mumbai. GPUs are in London only, for both GPU compute and GPU Kubernetes, so the H100 sits in the UK and there is no GPU option in Frankfurt or the United States. All regions carry CPU Kubernetes, CPU compute, volumes and databases, and object storage is everywhere except Phoenix. The product is built around Kubernetes: clusters are advertised as deploying in 90 to 120 seconds, the control plane is included in the node price rather than billed separately, and there is a marketplace of one-click cluster applications. GPU images ship with NVIDIA drivers pre-installed. There is a documented public API, an official Terraform provider and an official command line tool, both actively maintained on GitHub in September 2026, plus a Go SDK and a CSI driver. There is no official MCP server: the only Civo MCP server is a third party's, unsupported by Civo. There is no Slurm support and no priced bare-metal or dedicated-server product.

Civo lists ISO 27001, SOC 2 and Cyber Essentials Plus, along with Crown Commercial Service and G-Cloud listings, though it does not publish certificate numbers, scopes or auditors. It sells a named UK Sovereign Cloud with data held inside the local jurisdiction, available in the UK and India with Indonesia announced, and it says its data centres run on 100% renewable energy at around 1.2 PUE. It does not address the US CLOUD Act anywhere, which is a noticeable absence on a page otherwise about sovereignty, given it also runs two US regions. The published SLA guarantees 99.95% uptime, with a credit of 10% of the monthly fee for uptime between 99.0% and 99.95%, 40% between 95.0% and 99.0%, and 100% below 95.0%, which is strong at the bottom end. The catch is the claim: you have 30 days and you must supply server log files showing the connectivity errors with dates and times. Support runs through tickets and dashboard messaging plus a public Slack community, and Civo does not publish response targets, hours of cover or any paid support tier. There is no phone line. On Trustpilot the profile is claimed and carries a 4.0 score from three reviews, two of them in the past year, all five star and organic rather than invited, with Trustpilot noting no recent history of asking for reviews. Three reviews is too few to read much into, and there are no one-star reviews to report. G2 could not be read. No outage with data loss, regulatory action or lawsuit was found on the public record.

*Not yet tested by Saaskly. Assessed on published pricing, vendor documentation, company filings, the vendor's own status and trust pages, and third-party reviews.*

---

Researched and drafted with AI assistance; reviewed and approved by Tim Meredith. See https://saaskly.com/ai-transparency.md and https://saaskly.com/editorial-policy.md.

Canonical HTML: https://saaskly.com/category/gpu-compute/civo · Markdown index: https://saaskly.com/llms.txt