nebius logo

Nebius

GPU cloud provider spun out of Russia's Yandex search engine platform

Fast Facts
Datacenter Tier: UnknownFounded: 2022Founders: Arkady VolozhFunding: UnknownGpu Variety: LowHeadquarters: Amsterdam, NetherlandsMerits: Nvidia Cloud PartnerPersistent Storage: Supported

$$$$
Price

High

Availability
Available Regions

Overview

Nebius spun out of the Russian search giant Yandex to serve as a standalone AI infrastructure provider. Nebius focuses on the latest generation of GPUs like the H100 and L40s while also using the highest end interconnect infrastructure across all machines. Nebius works with a number of high profile customers including Mistral AI and Luma AI.

Highlights

Dots
Nvidia Cloud PartnerMember of Nvidia Cloud Partnership program
Dots
Regional FocusFocused on performance in a specific region
Dots
Great AvailabilityGPUs are readily available for on-demand use

Instances

Features

01.

Shadeform Integrated

Integrated with Shadeform for console and API

02.

On-Demand Clusters

Provides interconnected GPU clusters on-demand

03.

Persistent Storage

Disk storage that can be used across multiple instances

04.

NVLink

Instances can be NVLinked to enhance GPU to GPU communication

Pricing

Nebius offers primarily high end, latest generation GPUs like H100s and L40Ss and price them on the higher end of providers. One of the reasons why Nebius H100s are priced higher is because all machines are SXM5 and come with NVLink by default. Nodes can also optionally be interconnected with infiniband which further increases the hardware costs.

Availability

Nebius has historically always had availabilty for its H100 and L40S gpus.

© Shadeform, Inc