Meet your AI needs with confidence
Cloud GPUs: Find the One Made for You
- Affordable
- Secure
- Scalable
Your GPU Match, Ready for Every AI Journey

Your challenges

Keep Your Data Secure and Under Control
Maintaining control of your data is essential when adopting AI solutions. Models can risk exposing sensitive information, so choosing a cloud provider with robust security and transparency is critical. At OVHcloud, we ensure your data is never used to train third-party models, with complete clarity in processing. For the highest level of control and isolation, our Bare Metal servers offer dedicated physical resources tailored to your specific needs.

Plan Your AI Budget with Confidence
Take control of your budget with OVHcloud’s transparent, predictable pricing. Our pay-as-you-go model and flexible rates allow you to optimise costs while meeting your AI performance demands. Plan your projects confidently with no hidden fees, and scale your GPU usage seamlessly as workloads evolve. Plus, enjoy free Managed Kubernetes Service with GPU instances, offering even greater value, efficiency, and simplicity for your AI workloads.

Find the Right GPU for Seamless Growth
Unlock the Power of AI and Visual Computing with High-Performance GPUs

Performance Meets Value: OVHcloud GPUs Lead the Way
Did you know? OVHcloud’s GPUs deliver unmatched performance at a fraction of the cost of other providers.
For example:
- Save up to 64% on Llama 3.1 inference costs with OVHcloud GPUs, while maintaining top-tier performance
When it comes to balancing price and power, OVHcloud stands out as the clear choice.

A Trusted & Secure Cloud

Cloud GPUs, Built to Perform
Unlock flexibility with our versatile range of Cloud GPUs, optimized for AI, machine learning, and visual computing. Designed to seamlessly fit any cloud environment—whether you need instances, Managed Kubernetes, or Bare Metal servers.

Sustainability and efficiency
Achieve your environmental goals with our sustainable development policies. Our Cloud GPU instances are built with energy-efficient technologies that limit your carbon footprint.

Experience the best price for performance
Take control of your budget with clear, predictable pricing. Our competitive rates make it easy to manage costs while achieving optimal performance for your AI projects. Plus, benefit from free Managed Kubernetes Service with GPU instances.
Ready to get started?
Get A$ 300 in free credit to launch your first Public Cloud project
Testimonial
YouTube conditions the playback of its videos on the deposit of tracers in order to offer you targeted advertising based on your browsing.
In order to watch the video, you need to accept the Sharing cookies on third-party platforms privacy category in our Privacy Center. You have the option of withdrawing your consent at any time.
For more information,visit the YouTube cookies policy and the OVHcloud cookies policy .

They work with us
Onboarding Tutorials

Collaboration at its best
OVHcloud and NVIDIA jointly deliver tailored services for deep learning, inference and high-performance computing, with the best available GPUs. AI models are becoming more complex due to the rise of conversational AI. Training and inference now require massive computing power and scalability.
AI Unlock the Power of Your GPUs
Your GPU choice is just the start. OVHcloud offers solutions like Bare Metal servers, Kubernetes, and Public Cloud to enhance performance, scalability, and control—helping you get the most out of your workloads.
Case Studies
Frequently Asked Questions
Which hypervisor is used for instance virtualisation?
GPU instances, like other OVHcloud instances, are virtualised using the KVM hypervisor, a component integrated into the Linux kernel.
What is PCI Passthrough?
PCI Passthrough is a feature that enables GPUs to be accessed directly via the physical server’s PCI bus. This allows hardware resources to be fully dedicated to a virtual machine, bypassing standard virtualisation layers for optimal performance.
Can I resize a Cloud GPU instance?
Yes, Cloud GPU instances can be upgraded to a higher model after rebooting. However, downgrading to a lower model is not supported, so plan your resources accordingly.
Do GPU instances include anti-DDoS protection?
Yes, all GPU instances come with OVHcloud’s anti-DDoS protection as standard, included at no additional cost.
Can I switch to hourly billing if I have monthly billing set up?
No, once an instance is set to monthly billing, it cannot be switched to hourly billing. Please ensure you choose the billing method that best fits your project before launching your instance.
What is a Cloud GPU?
A Cloud GPU is a scalable cloud service that provides access to powerful graphic processing units (GPUs) for high-computing workloads. These workloads can include graphic rendering, AI training, machine learning, data analysis, and scientific simulations. Cloud GPUs eliminate the need for significant hardware investments, offering flexibility, scalability, and pay-as-you-go pricing.
What are H100 and A100 servers?
NVIDIA H100 and A100 servers are high-performance computing systems designed to deliver exceptional capabilities in AI, data analytics, and high-performance computing (HPC) applications. These GPUs are purpose-built for demanding workloads, ensuring optimal results.
What is NGC (NVIDIA GPU Cloud)?
NVIDIA GPU Cloud (NGC) is a platform developed by NVIDIA to simplify GPU-accelerated computing. It provides a range of tools, including containers, pre-trained models, SDKs, and other resources, optimised for AI, machine learning, and HPC workloads. NGC accelerates deployment and enhances performance by leveraging NVIDIA GPUs.
Why use a Cloud GPU?
Cloud GPUs offer several advantages, particularly for developers, researchers, and organisations working on resource-intensive tasks like artificial intelligence (AI), machine learning (ML), high-performance computing (HPC), and graphics rendering. They provide flexible, scalable access to high-performance GPUs without the need for costly on-premises infrastructure.
What are Large Language Models (LLMs), and how do GPUs support them?
Large Language Models (LLMs), like GPT or Llama, are advanced AI models trained on vast amounts of data to perform tasks such as text generation, translation, and summarisation. GPUs play a critical role in training and deploying LLMs by providing the computational power required to handle the immense complexity and scale of these models. OVHcloud’s GPUs, including the NVIDIA H100, are optimised to support LLM training and inference efficiently, enabling faster results and cost-effective scaling.
1
Limited ‘Public Cloud Free Trial’ special offer applicable to all first-time setup and use of a Public Cloud project. Both new and existing customers can request to use this special offer, provided that they have not already created a Public Cloud project in the past, whether it is still active or not. Special offer available from 30 June 2022, 23:00 BST. The voucher must be activated when the holder creates their first Public Cloud project. The voucher is only valid for the purchase of Public Cloud services provided by OVHcloud, purchased directly from OVHcloud via the website, in all available Public Cloud regions, excluding free offers and free beta testing services. This voucher cannot be used in conjunction with any other current special offers applicable to the services concerned, including the ‘Public Cloud Free Trial’ special offer. The voucher applies to standard, non-discounted public rates (as published on the OVHcloud website). The voucher’s value is listed in the publicly displayed currency for the market/country to which the Public Cloud account with the voucher is attached, excluding taxes, and can only be used to order services in the same currency. The voucher is valid for the purchase of services usually available in the market to which the NIC handle used is linked. The voucher is assigned to a specific natural or legal person who already has an OVHcloud customer account, and is linked to their OVHcloud NIC handle (unique ID). If the natural or legal person has several NICs, the voucher shall be attached to a single NIC. The voucher cannot then be linked to a different NIC handle, and only one voucher may be used. A single natural or legal person can only use one voucher, even if that person has several different NIC handles. In the event that the voucher is used by a NIC handle other than the NIC handle to which the voucher is attached, OVHcloud reserves the right to automatically cancel or terminate the services obtained as a result, without reissuing the voucher and without legal formality or compensation. The use of the voucher is subject to the full acceptance of these conditions, as well as the General and Special Terms and Conditions applicable to the services obtained. These terms and conditions can be viewed at: https://www.ovhcloud.com/en-gb/terms-and-conditions/contracts/. By way of exception, the amounts issued from this voucher do not form part of the base monthly fee for services used in terms of service levels (SLAs) and associated extra charges — i.e. if the voucher holder uses an amount less than or equal to the voucher for a given month, no penalty will be paid for a breach of an SLA in the course of the service, while if the voucher holder uses a higher amount, only the amounts paid above the voucher value will be taken into account in calculating extra charges. The voucher is non-exchangeable, non-refundable, and non-redeemable, even partially. It may not be given away, sold or transferred in any capacity whatsoever to a third party, either free of charge or for a fee (it may not be transferred between separate legal entities of the same group of companies). The voucher can only be used for one Public Cloud project, but can be used in one or more instalments on the project, and can be used in conjunction with another payment method if required, until the balance is used up or the voucher’s validity period ends. To use this voucher, the holder must have a valid payment method saved in their OVHcloud customer account. Once the voucher has been activated in the holder’s customer account, it is valid for a period of one (1) month, and is automatically deducted from the bills linked to the Public Cloud project for which the voucher has been activated. The voucher is not replaceable in the event of loss, theft, destruction, end of validity or fraudulent use. Furthermore, in the event of non-compliance with these conditions or of fraudulent use (including one person using multiple vouchers), OVHcloud reserves the right to terminate the subscribed services using the vouchers, and/or cancel the vouchers, and the holder will in all cases be liable to pay the full amounts to OVHcloud for the services used, including the voucher amount, without prejudice to any compensation that OVHcloud may claim. Within the limits of what is authorised by the laws in force, OVHcloud’s responsibility for this special offer is limited to the voucher amount. The vouchers will be created and delivered by the OVH Groupe SA Affiliates companies (537 407 926 RCS Lille Métropole, 2 rue Kellermann, 59100 Roubaix, France), depending on which OVHcloud Groupe company the holder orders the Public Cloud service from using the voucher (this entity is named here as “OVHcloud”). For the purposes of these conditions, “Affiliates” are any company controlled by OVH Groupe SA, that controls OVH Groupe SA, or with which OVH Groupe SA is jointly controlled by a third party. The concept of control is understood in Article L233-3 of the French Commercial Code, control may be direct or indirect. Furthermore, the “OVHcloud Groupe” is defined as OVH Groupe SA and all its Affiliates. French law applies to these conditions, and any dispute that is not resolved through an amicable solution will be subject to the jurisdiction of the French courts.