Nebius to roll out NVIDIA Vera Rubin NVL72 in H2 2026

Nebius has announced plans to deploy NVIDIA Vera Rubin NVL72 through its AI Cloud and Nebius Token Factory. Customers will be able to access it from H2 2026. The company says that this will make it among the first NVIDIA partners to deploy the technology, and it will be available in its US and European datacentres. It is the latest buy-in to the NVIDIA AI technology from Nebius and continues its commitment to that relationship.

Arkady Volozh, founder and CEO of Nebius, said: “We are proud to be one of the first on the market to offer Vera Rubin GPUs as we fuel the next wave of AI innovation.

“By integrating Vera Rubin into Nebius AI Cloud and our inference platform Nebius Token Factory, we’re giving AI innovators and enterprises the infrastructure they need to develop agentic and reasoning AI systems faster and more efficiently.”

What is NVIDIA Vera Rubin NVL?

Announced in September 2025 and finally unveiled at CES 2026, the NVIDIA Vera Rubin NVL is a rack-scale AI supercomputer. It consists of 72 Rubin GPUs and 36 Vera CPUs. It is NVIDIA’s flagship AI solution that, on paper, outperforms its existing solutions and is designed to deal with complex AI workloads.

When NVIDIA announced it in 2025, it made some significant claims for the capabilities of this AI supercomputer. Among those claims were:

• The NVIDIA Rubin CPX GPU is purpose-built to handle million-token coding and generative video applications.
• The NVIDIA Vera Rubin NVL144 CPX platform packs 8 exaflops of AI performance and 100TB of fast memory in a single rack.
• Companies can monetise at an unprecedented scale, with $5B in token revenue for every $100M invested.
• It delivers 8 exaflops of AI compute, delivering 7.5x more AI performance than the NVIDIA GB300 NVL72.

To deliver all of that, NVIDIA had to design a new architecture. That architecture means that each rack is its own self-contained machine, while also acting as part of a larger AI factory. In a change to traditional computing, the company has tightly integrated compute, networking and infrastructure into the architecture.

NVIDIA has also designed six new chips, each with its own discrete function. The chips have more cores, threads and L2 and L3 cache than previous generations. The memory bandwidth is now 3x that of earlier chip generations, which is a key element in its performance.

The question now is how many generations of this architecture will NVIDIA deliver? How long before we see a replacement for the Vera Rubin NVL? Companies buying into the technology need to be able to monetise their investment.

What does this mean for Nebius?

That latter question will have been part of board-level discussions at Nebius. On December 17, 2025, Nebius announced that it had gone live with its first GB300 NVL72 deployment in Finland. It had also previously announced plans for that system in multiple locations. Now it is announcing the replacement for that system, which is many times more powerful.

It raises several questions. Will it be able to recover its investment in the GB300 family? Will it continue with those deployments over the next six months as it waits for its first Vera Rubin NVL? What are the data centre requirements of the Vera Rubin NVL? Will Nebius be able to reconfigure its existing datacentres or will it need to redesign for greater power and cooling needs? How many Vera Rubin NVLs is Nebius planning to buy?

Perhaps the biggest question is, will this impact customer plans and delay their spending as they wait for the next generation?

What we do know is that it will be available through the Nebius AI Cloud and Token Factory. The company says that the Vera Rubin will “serve as the computational layer in Nebius Token Factory, complementing existing NVIDIA GB200 NVL72 and NVIDIA Grace Blackwell Ultra NV72 capacity and expanding customers’ choice of platforms optimized for different AI workload profiles.”

That statement suggests that Nebius sees its investments as a tiered approach where customers can access different levels of AI technology based on need and cost. It will be interesting to see how this works out across its customers and how it impacts cash flow.

Enterprise Times: What does this mean?

This is a key announcement by Nebius that keeps it in the top tier of AI cloud providers. While everyone looks at Google, AWS and Azure as the leaders, the investments by Nebius give it a seat at that table. Importantly for Europe, it continues to deliver on its commitment to the European AI industry.

However, it will take several months for the ramifications of this announcement to play out. The costs associated with potential data centre improvements and the acquisition of the technology need to be seen. The impact on potential customer business will also be watched carefully, as will the revenue, as customers look for lower costs when using older technology.

Last year, the company signed a large deal with Microsoft and raised significant capital. This announcement should mean we will see other large deals this year. That should mean it has all the money it needs to support this technology acquisition.

The post Nebius to roll out NVIDIA Vera Rubin NVL72 in H2 2026 appeared first on Enterprise Times.