GPU Servers
Solve complex problems and run your compute-intensive AI, computation and graphics workloads on GPU-backed virtual servers.
Accelerate innovation and solve complex problems
Build and manage workloads
Create your GPU-backed virtual server in the portal. The Windows Server template already has the GPU driver and licence, so no extra installation is needed.
Optimise and scale resources
Choose a full card or a 1/2, 1/4, 1/8 or 1/16 slice to suit your workload; small workloads run on small profiles.
Integrate with storage
Keep training data and model files in S3-compatible Object Storage and reach them from your GPU servers.
Secure your data
Your training data, models and inference results stay in data centres in Türkiye.
Example scenarios
A team running its AI model in Türkiye
A compute profile with 24 GB of GPU memory is chosen for inference, and model files are read from an S3-compatible bucket in Object Storage. Training data and results stay in data centers in Türkiye.
A start-up that begins on a small slice of a card
The first version runs on a one-eighth slice of a card, and a server with a larger profile is opened as the work grows. Servers without a GPU keep running beside it in the same pool.
An engineering office working in CAD
Designers connect to GPU-backed desktops through Virtual Desktop. Graphics profiles start at a 3 GB slice and support more than one virtual display.
Power your most complex workloads
AI infrastructure
Use compute profiles for AI inference and GPU computation workloads. Compute profiles come with 6, 12, 24 and 48 GB of GPU memory.
Visualisation
Run graphics-intensive applications and visualise complex data. Graphics profiles for CAD and 3D design start from a 3 GB slice and support multiple virtual displays.
Graphics desktops
Give design and engineering teams GPU-backed desktops with Virtual Desktop.
Flexible GPU assignment
Servers with and without a GPU run together in the same organisation pool; a GPU is added to an existing server later through a policy.
Driver and licence management
The driver in the template is matched to the platform's GPU software version and renewed with every major upgrade. Servers pick up their licence automatically from the licence server run for you.
Embedded security
Runs inside your organisation network
GPU servers are created on the same organisation network and behind the same firewall as your other servers.
Controlled change path
GPUs are added and changed only through the portal API, with no manual changes to the infrastructure.
Role-based access
Your team gets the right to create and manage GPU servers through portal roles.
GPU Servers pricing
Each GPU profile is a fixed item, with the server's vCPU, memory and disk items added on top.
Learn more to get started
VM and container platform
Learn how the platform that runs GPU servers is built.
Solutions for software companies
See the recommended set-up for teams running AI and software products in the cloud.
Infrastructure
See the data centre and network infrastructure where GPU servers run.