Nebius Accelerates AI Infrastructure Build-Out: $25 Billion Capex, 4 Gigawatt Target, and a $50 Billion Order Book

Nebius secures $44B in contracts from Meta and Microsoft, plans 4+ GW capacity, reports 700% revenue surge, and sees analyst price targets soar above $280.

Nebius is no longer just a fast-growing cloud provider — it is transforming into an industrial-scale infrastructure developer. The Amsterdam-based company is committing up to $25 billion in capital expenditure this year alone, backed by a swelling order backlog that now approaches $50 billion. At the same time, it has lifted its contracted capacity goal to more than 4 gigawatts by the end of 2026, up from an earlier target of 3 gigawatts and already exceeding 3.5 gigawatts across multiple sites.

That spending is underpinned by two blockbuster customer commitments. Meta Platforms signed a five-year contract for AI infrastructure worth up to $27 billion, while Microsoft has a separate deal running through 2031 valued at $17.4 billion. Chip giant Nvidia also injected $2 billion directly into Nebius as part of a deepening technology partnership that gives the cloud firm early access to next-generation hardware.

The financial firepower is visible in the latest quarterly numbers. Nebius reported revenue of $399 million for the first quarter, a jump of nearly 700 percent year over year. Adjusted EBITDA came in at roughly $130 million, well ahead of analyst expectations and a stark reversal from the red ink of a year ago. The company is guiding for full-year revenue of $3 billion to $3.4 billion, with recurring revenue in 2026 forecast to land between $7 billion and $9 billion.

To turn those projections into reality, Nebius is racing to bring new data centers online. During a May 14 conference call, chief executive Arkady Volozh unveiled plans for a 1.2-gigawatt campus in Pennsylvania — the company’s second U.S. site of that scale after its Independence facility in Missouri. Initial capacity from the Pennsylvania campus is expected to go live by the end of 2027, starting at 250 to 350 megawatts and adding roughly 300 megawatts annually thereafter. Construction on a new "AI factory" in Missouri has already begun.

Wall Street has taken notice. Citigroup analyst Tyler Radke raised his price target on Nebius shares to $287 from $169, maintaining a buy rating and citing strong quarterly results and extreme scarcity of AI infrastructure. He noted that more than four customers are currently competing for each available GPU on Nebius’s network. Citizens and Compass Point also turned more bullish, lifting targets to $270 and $260, respectively.

Despite the optimism, the stock ended Friday at $213.85, down 3.3 percent on the session. The year-to-date advance still stands at 136 percent, and the shares have more than quintupled over the past twelve months, having hit a weekly high of $233.73. Some investors are clearly taking profits.

Insider activity has added to the mix. On May 13, chief infrastructure officer Andrey Korolenko exercised options on 500,000 shares at a strike price of $100 and then sold them in the market at an average price of $203.24, generating total proceeds of roughly $102 million. The transactions were executed under prearranged trading plans. Separately, director Elena Bunina sold 10,894 shares for about $2.25 million the same day, cutting her direct stake by roughly half. Those sales also fell under planned trading arrangements.

Nebius enters its expansion phase with a heavily fortified balance sheet. Cash and equivalents stand at $9.3 billion, including Nvidia’s $2 billion investment and more than $6 billion raised through convertible bonds. That liquidity will be crucial: building AI data centers absorbs enormous capital long before the revenue streams kick in.

For 2026, management has penciled in capital spending of $20 billion to $25 billion — a figure that underscores how aggressively the company is moving to lock in its infrastructure lead. The real test, analysts say, is no longer demand but execution: converting power commitments, GPU orders, and construction timelines into live capacity on schedule.