We’re starting pre-orders for NVIDIA Blackwell GPU-powered clustersWe’re starting pre-orders for NVIDIA Blackwell GPU-powered clusters

We are now accepting pre-orders for NVIDIA GB200 NVL72 and NVIDIA HGX B200 clusters to be deployed in our data centers in the United States and Finland from early 2025. Learn more and secure your cluster today.

Based on NVIDIA Blackwell, the architecture to power a new industrial revolution of generative AI, these new clusters deliver a massive leap forward over existing solutions. Over 22,000 NVIDIA Blackwell GPUs will be deployed on the Nebius AI-native cloud.

Putting ourselves on the map in the United StatesPutting ourselves on the map in the United States

Our first availability zone in the US is due to go live in Q1 2025. The Kansas City data center will become Nebius’ third AZ in addition to our own facility in Finland and the colocation in France.

Supporting our growth plans in the country, we now also have offices in the iconic Ferry Building in San Francisco and on Dallas Parkway in Texas. Nebius’ third office will open in New York in December.

Our latest AI R&D study: Leveraging training and search for better agentsOur latest AI R&D study: Leveraging training and search for better agents

Discover how Nebius is advancing AI-driven software engineering through a combination of search and learning to build better agents.

The research article written by Alexander Golubev and fellow engineers from our in-house AI R&D team explores how this approach enhances both open-weight and frontier LLMs, driving further improvements in software development automation.

Register for a webinar: Fine-tuning LLMs with MLflowRegister for a webinar: Fine-tuning LLMs with MLflow

We will gather on Dec 18 to explore best practices for experiment tracking, model management and collaboration in LLM projects with MLflow’s powerful features.

Things to consider when building GenAI appsThings to consider when building GenAI apps

• Nebius AI Studio has been benchmarked by Artificial Analysis. For each model we offer — and we have the most popular ones like Llama, Mistral, Qwen and many others — Studio’s prices are the best, according to AA.

• Follow this guide to build an AI-powered finance planner using the Meta Llama 3.1 70B model via Nebius AI Studio and a full-stack Next.js app. Or simply draw inspiration from some of the ideas.

• Staying on the subject of demo apps based on open-source models, we also detailed the process of creating a powerful code generator and reviewer. You can customize it to your exact needs and extend further.

In-house designed hardware explained: Watch our Open Compute Project Summit talkIn-house designed hardware explained: Watch our Open Compute Project Summit talk

Throwback to this fall’s OCP Global Summit, where our hardware R&D team leaders Igor Znamenskiy and Oleg Fedorov gave a talk on Nebius’ in-house server and rack design.

Our designs are based on several concepts and principles from the Open Compute Project, so this year, we came full circle by bringing our own developments to the OCP community.