NVIDIA Bets $2B on Nebius AI Cloud Partnership, NBIS Stock Jumps

NVIDIA is putting $2 billion behind Nebius Group to build out what both companies call the next generation of hyperscale AI cloud infrastructure. The deal sent Nebius shares climbing nearly 2% in Wednesday trading, with NBIS touching $89.19 against a market cap now sitting at $24.26 billion.

The partnership targets a specific buildout: more than 5 gigawatts of NVIDIA systems deployed by the end of 2030. For context, that’s roughly the power consumption of a mid-sized city dedicated entirely to AI compute.

Nebius, which spun out from Yandex’s international operations and now trades on Nasdaq from its Amsterdam headquarters, already runs NVIDIA hardware in its existing AI factories. The new capital will accelerate that infrastructure across multiple product generations.

What’s Actually Being Built

The collaboration covers four key areas: AI factory design and operational support, an inference and agentic AI technology stack, deployment of upcoming NVIDIA hardware including the Rubin platform and Vera CPUs, and fleet management optimization using NVIDIA’s BlueField storage systems.

That last piece matters for enterprise customers. Managing thousands of GPUs at scale remains one of the messier problems in AI infrastructure, and BlueField integration suggests Nebius is positioning for the operational complexity that comes with gigawatt-scale deployments.

The Circular Financing Question

Some market observers have raised eyebrows at the deal structure. NVIDIA, which derives massive revenue from selling chips to cloud providers, is now directly investing in a cloud provider that will buy more chips. The arrangement has drawn scrutiny over what some are calling “circular financing” dynamics.

NVIDIA’s own stock traded about 5.2% below its session peak on Wednesday, with the chip giant carrying a $4.58 trillion market cap. Whether the Nebius investment represents strategic positioning or creative demand generation likely depends on who you ask.

The Agentic AI Angle

Both companies are framing the partnership around “agentic AI” – autonomous systems that can execute complex tasks without constant human oversight. Jensen Huang’s team has been pushing this narrative hard, arguing that agentic workloads will drive compute demand well beyond current training-focused infrastructure.

If that thesis plays out, cloud providers purpose-built for inference and agent deployment could capture meaningful market share from hyperscalers still optimized for training workloads. Nebius appears to be betting its entire roadmap on that transition.

The 2030 timeline gives both companies runway to execute, though AI infrastructure timelines have a habit of compressing. First hardware deployments from the expanded partnership should offer early signals on whether the 5-gigawatt target is realistic or aspirational.

Image source: Shutterstock