The more I think about the AI supercycle, the less I think the investment case belongs exclusively to hyperscalers. That shift in perspective is why I now see Nebius Group (NBIS +4.53%) in a different light. The company is building an AI cloud platform that goes beyond model training, which is why it could play an even bigger role in the next phase of the AI supercycle.
The AI boom is a race to secure capacity
Nebius says that demand for AI compute remains ahead of available supply. In fact, it signed four AI cloud contracts worth more than $1 billion each during the second quarter, while contract values with new customers increased more than ninefold.
Image source: Getty Images.
That shows how hyperscalers are already spending billions of dollars to lease capacity on third-party AI infrastructure, creating more opportunities for Nebius to provide capacity and meet that surging demand. It also means that Nebius doesn't need to compete with hyperscalers. Instead, it can help them expand.
One interesting example is Meta Platforms. Nebius announced a five-year agreement under which Meta will lease $12 billion in dedicated AI capacity beginning in 2027, with the potential for an additional $15 billion in capacity over the period. That puts the total potential contract value at around $27 billion.
Meanwhile, chipmaker Nvidia is already validating Nebius' business. It announced a $2 billion investment in Nebius alongside a partnership covering AI factory design, infrastructure, and agentic AI. Through the partnership, Nebius could deploy more than 5 gigawatts of Nvidia systems by the end of 2030.
But the opportunity is now expanding beyond AI training.
The next wave could be inference, agents, and enterprise
While the first phase of the AI boom was defined by training AI models and developing consumer applications, the industry is now moving toward inference, AI agents, and enterprise. Nebius is positioning itself for that transition.
NASDAQ: NBIS
Key Data Points
Its partnership with Nvidia includes work on inference and agentic AI, as well as the development of infrastructure for robotics and physical AI, positioning the company to meet the future of compute demand instead of just relying on one established type of AI workload.
On top of that, the company's push toward enterprise use cases is gaining traction. Palantir named Nebius its preferred sovereign AI infrastructure partner. The companies plan to integrate Nebius into Palantir's enterprise environment, giving customers access to Nebius infrastructure while keeping their compute, data, and models within Palantir.
To help meet the growing demand for those use cases, Nebius is aggressively building out cloud capacity.
Nebius is rapidly expanding its infrastructure
In May, Nebius broke ground on an AI factory in Independence, Missouri. It describes the project as its first gigawatt-scale digital infrastructure project in the U.S.
Nebius is also expanding in the U.K., where it committed around 1.7 billion pounds to build capacity across four sites. The planned deployments are based on Nvidia infrastructure and are expected to reach 65 megawatts when fully ramped up.
What's more interesting is that the company is finding ways to scale without owning everything. It announced a new infrastructure model under which its partners finance and own their data centers, while Nebius supplies its systems, software, and services stack and then brings that capacity to market.
The new approach could be a faster way to expand its AI cloud footprint without needing to finance every piece of infrastructure itself. More importantly, it gives Nebius an edge in the race to secure more AI capacity.
With all of that positioning underway, the real question for investors is what it means for the stock.
Is Nebius stock a buy right now?
As the AI megatrend shifts toward inference, agentic AI, and enterprise applications, the demand for AI infrastructure will only continue to accelerate. Nebius is positioning itself right in the middle of that expansion, signing major customers and inking strategic partnerships, and expanding platforms, all of which make it a much more exciting way to invest in the AI supercycle.
