– GPU

SKS: GPU MIG partitioning support

Tue, 21 Jul 2026 00:00:00 +0000

Exoscale SKS now supports GPU MIG partitioning for MIG-capable GPU nodepools.

With Multi-Instance GPU (MIG), customers can split a supported physical GPU into multiple isolated GPU instances, each with dedicated compute and memory resources. This makes it easier to run several smaller GPU workloads on the same SKS node instead of assigning a full GPU to each workload.

Availability and setup

MIG partitioning is available for Exoscale A30, RTX Pro 6000 and B300 GPUs.
Standard SKS nodepools can be configured with MIG profiles through the Exoscale Portal, CLI, or Terraform provider.
Karpenter-managed GPU nodepools can use MIG through the ExoscaleNodeClass configuration.

Why use MIG?

Run multiple smaller GPU workloads on one physical GPU.
Improve GPU utilization and reduce underused capacity.
Keep workloads isolated at the GPU partition level.
Better accommodate AI inference, development and testing, and smaller batch workloads.

Note: Existing nodes must be cycled before a changed MIG profile takes effect. Billing is based on the selected GPU node instance type, not on individual MIG partitions.

For setup instructions, see:

Configure GPU MIG partitioning on SKS

GPU: A40 (GPU3) price reduction - 30% off from June 2026

Tue, 26 May 2026 13:30:00 +0000

All A40 (GPU3) instances in Frankfurt will be 30% cheaper starting June 1st, 2026. Existing instances continue to run unchanged and automatically benefit from the new pricing.

Size	GPUs	Current EUR/CHF/USD/hr	New EUR/CHF/USD/hr
Small	1	1.4933	1.0453
Medium	2	2.9867	2.0907
Large	4	5.9733	4.1813
Huge	8	11.9467	8.3627

The A40 has 48 GB of GDDR6 memory with Ampere RT Cores, 336 Tensor Cores, and 10572 CUDA Cores into a single card, making it a capable choice across a wide range of GPU workloads: AI model fine-tuning, batch inference, large-context LLM serving, 3D rendering, visual effects, AR/VR simulations, and scientific computing that needs GPU memory beyond the 24 GB range.

It also runs as the GPU backend for Dedicated Inference, which lets you deploy any Hugging Face model as a production-ready, OpenAI-compatible API endpoint on fully isolated European infrastructure. With the A40 you can serve models in the 13B - 34B parameter range comfortably, with no shared resources and no data leaving your environment.

At the new price point, A40 is a strong option whether you are running GPU instances directly or building on top of Dedicated Inference.

Learn more about A40 (GPU3) instances

SKS: New GPU node template - NVIDIA RTX Pro 6000 support

Tue, 10 Feb 2026 15:12:32 +0000

A new GPU node template for SKS is now available, supporting NVIDIA RTX Pro 6000 and Kubernetes versions 1.33, 1.34, and 1.35, enabling customers to run AI and inference workloads on managed Kubernetes.

The template uses the open-source NVIDIA driver, which is aligned with current NVIDIA support and is better suited for the long-term operation of GPU-enabled Kubernetes clusters.

Existing GPU node pools are not updated automatically and must be recreated to use the new template.

Note: Due to limitations of the open-source NVIDIA driver, GPU2 (Tesla V100) is no longer supported with this template.

For more details and prerequisites, check the documentation

GPU: NVIDIA RTX Pro 6000 now available + A40 (GPU3) price reduction

Tue, 13 Jan 2026 11:33:06 +0000

We’re expanding our GPU portfolio with the launch of NVIDIA RTX Pro 6000 instances and reducing A40 (GPU3) prices by 30% to give you more choice and better value.

What’s new?

NVIDIA RTX Pro 6000 GPUs

RTX Pro 6000 instances are now available in Zagreb, Frankfurt, and Zurich. They are designed for demanding AI, inference, and high-end graphics workloads that require more memory and compute. Secure your access now during the initial availability phase before capacity runs out.

A40 (GPU3) price reduction

The price of A40 (GPU3) will be reduced by 30%, effective February 1, 2026. Existing instances continue to run unchanged and automatically benefit from the new pricing.

You now have more flexibility when choosing GPUs on Exoscale - lower prices for proven A40 workloads, and a new high-performance option with RTX Pro 6000 for more demanding use cases.

Learn more about all available GPU options on Exoscale

GPU1 Instances Retrofit: Upgrading from NVIDIA P100 to A30 in Geneva

Mon, 07 Jul 2025 09:00:00 +0000

We're excited to announce an upgrade to our GPU offering: instances equipped with NVIDIA Tesla P100 GPUs in our Geneva zone are retrofitted with more powerful NVIDIA A30 GPUs.

The NVIDIA A30 GPU represents a significant leap from the P100, featuring a newer Ampere architecture with improved tensor cores and 24 GB of HBM2 memory, along with 933 GB/s of memory bandwidth, enabling faster mixed-precision AI workloads and improved energy efficiency.

This upgrade allows our customers to take advantage of the newer architecture and better throughput without changing anything in their deployment.

💶 Check the GPU A30 pricing.

This change ensures our infrastructure keeps pace with the evolving demands of compute-intensive workloads while maintaining Exoscale's commitment to performance, transparency, and data sovereignty.

More details about our NVIDIA A30 offering

General Availability – Liquid-Cooled A5000 & 3080 Ti GPU Instances

Wed, 28 May 2025 14:30:00 +0000

We are excited to announce the general availability of our liquid-cooled NVIDIA RTX A5000 and 3080 Ti GPU instances. Following a successful public preview phase, these powerful GPU instances are fully integrated into our platform, featuring SLA-backed availability to ensure optimal performance and reliability.

Please note that as of June 1, 2025, these instances will transition from complimentary to billed usage, following our standard pricing guidelines. These GPUs are specifically designed to support compute-intensive workloads, including AI and ML training, rendering, and scientific simulations.

For detailed pricing information, please visit our pricing page: https://exoscale.com/pricing/#gpua5000-instances

Discover the benefits of our advanced liquid-cooling technology

Preview Availability: Liquid-Cooled A5000 & 3080 Ti GPUs

Mon, 10 Mar 2025 08:18:01 +0000

We’re excited to announce the preview availability of our two new liquid-cooled GPU instances, now available in our Vienna 2 data center:

🚀 NVIDIA RTX A5000

🚀 NVIDIA GeForce RTX 3080 Ti

These high-performance GPUs are designed for AI training, rendering, and high-performance computing—now enhanced with innovative liquid cooling solution, which can reduce energy consumption by up to 50% and utilize waste heat with up to 98% efficiency. (Learn more)

🔹 Early Access: During this phase, GPU instances are free of charge, but SLAs will only apply after the preview phase.

📩 Register Your Interest: Please reach out to us to reserve one. You will be contacted on a First Come, First Served basis to align on your needs and the timing of your preview phase.

💰 Pricing: Once the preview ends, standard pricing will apply. Check out the details here: GPUs Pricing.

Get in touch and start testing today!

Check our GPU Servers offering

GPU3 Instances released

Wed, 23 Feb 2022 07:40:00 +0000

GPU3 is now available on the Exoscale platform. With the release of the new GPU, we further broaden our product portfolio. Customers can now choose between 3 different types of GPUs, to best meet the specific needs and requirements.

GPU3, based on NVIDIA A40, offers powerful visual computing capabilities. With a new generation of RT Cores, used by the A40, workloads like rendering, VDI, CAD, AR or VR experience a massive performance increase. The GPU3 offering is a good all-rounder sitting in the middle of the A100 to A2 Ampere generation of cards from NVIDIA and can tackle AI, Machine Learning or Video applications as well.

All our GPU offerings now also support the NVIDIA GPU Cloud (NGC). Reduce time to solution by combining the Exoscale Cloud with the power of NVIDIA GPUs. NGC's rich container registry works with Compute and SKS alike.

Discover GPU3

New GPU2 instance types based on Tesla V100

Wed, 22 May 2019 03:00:00 +0000

Starting today a new GPU2 class of instances based on the Tesla V100 graphic cards is publicly available. In the meantinme, the existing P100 based instances get a 15% price drop.