Nebius Surges on Mega AI Deals—But Capacity Crunch Threatens Its $9 Billion Cloud Ambition

Written byGavin Maguire
Tuesday, Nov 11, 2025 8:46 am ET3min read
Speaker 1
Speaker 2
AI Podcast:Your News, Now Playing
Aime RobotAime Summary

-

reported $146.1M Q3 revenue (355% YoY), but missed $155.7M estimates due to capacity constraints.

- Signed $17.4B

and $3B AI deals, with Meta capacity to be deployed in 3 months.

- Capacity limits hinder 2025 growth, despite $7-9B ARR and 2.5GW power targets by 2026.

- Raised $4.3B in financing, plans $5B 2026 CapEx for GPU/data center expansion.

- Faces margin pressures and debt risks as it scales AI infrastructure amid high demand.

Nebius (NBIS), the Amsterdam-based neocloud infrastructure company powering the AI revolution, delivered another

of growth but fell short of Wall Street’s revenue expectations as capacity constraints capped upside. The company’s Q3 results reinforced its position as one of the key infrastructure players in the global race to build compute capacity for AI workloads, Shares are trading higher by approximately 4% in pre-market trade as upbeat commentary and a new multi-billion-dollar deal with Platforms (META) drive price action.

Nebius operates what it calls a “full-stack AI cloud”—a vertically integrated platform that provides compute, storage, and data services tailored for artificial intelligence workloads. Its infrastructure supports hyperscalers, enterprise AI users, and startups alike, offering GPU capacity at scale to train and deploy large language models and other AI applications. The company’s growing role in the AI ecosystem places it in the same competitive tier as CoreWeave and IREN—often referred to collectively as “neoclouds,” or next-generation cloud providers purpose-built for the AI era.

For the

ended September 30, reported revenue of $146.1 million, a 355% increase year over year, but just shy of analyst expectations of $155.7 million. Management attributed the shortfall to having sold out of all available capacity—a consistent theme this year as demand for AI compute continues to outpace supply. Adjusted net loss widened to $100.4 million from $39.7 million a year ago, while adjusted EBITDA loss improved 89% to $5.2 million, signaling strong operating leverage as the company scales.

CEO Arkady Volozh called the quarter “another period of exceptional momentum,” noting that the company signed two landmark AI infrastructure deals in recent months—a $17.4 billion agreement with Microsoft and a new $3 billion, five-year contract with Meta. “Demand for capacity was overwhelming,” Volozh said, explaining that the Meta deal size was limited only by the company’s available power and infrastructure. He added that Nebius plans to deploy the necessary capacity to service Meta within the next three months.

Despite the near-term strain on capacity, Nebius maintained a bullish long-term outlook. Management guided for annualized run-rate revenue (ARR) of $7 billion to $9 billion by the end of 2026, roughly half of which is already under contract. This compares with a current ARR of roughly $550 million. The company also raised its contracted power guidance to more than 2.5 gigawatts (GW) by year-end 2026, up from a previous goal of 1 GW. Of that total, it expects 800 megawatts to 1 GW of “connected power” by late 2026—capacity that will be fully provisioned and ready to activate upon GPU installation.

Still, the outlook wasn’t without caution. Like CoreWeave, Nebius is grappling with capacity constraints that are slowing near-term revenue growth. The company said its 2025 revenue trajectory will be shaped by how quickly it can unlock new power and data center capacity. “The only real limitation on our revenue growth in 2025 has been the amount of capacity we’ve been able to bring online,” management noted, emphasizing that demand continues to exceed supply across all regions.

From a financial perspective, Nebius continues to scale aggressively. Total operating expenses surged 145% to $276.3 million, driven by expansion-related costs and GPU investments, though as a percentage of revenue, expenses fell sharply—evidence of improving operating leverage. Cost of revenue rose 333% to $42.9 million but declined as a percentage of sales, while product development expenses grew 43% to $44.9 million (31% of revenue, down from 98% a year ago). Sales, general, and administrative costs rose 87% to $89.5 million, reflecting higher share-based compensation and marketing outlays. Depreciation and amortization climbed 321% to $99 million, underscoring heavy GPU and data center investments.

Nebius ended the quarter with a strengthened balance sheet and expanded financing options. The company raised $4.3 billion through convertible notes and a follow-on equity offering in September to support GPU purchases, land acquisition, and power infrastructure expansion. Additionally, Nebius plans to establish an at-the-market (ATM) equity program for up to 25 million Class A shares, which will provide ongoing access to capital while maintaining sensitivity to shareholder dilution. Management said it intends to balance financing across corporate debt, asset-backed loans, and equity, leveraging customer creditworthiness to secure favorable terms.

Capital expenditures totaled $955.5 million in Q3, primarily for GPUs and data center development, marking a ramp-up from earlier quarters as Nebius accelerates its buildout across the U.S., Europe, and the Middle East. The company expects 2026 CapEx to reach approximately $5 billion, up from $2 billion previously, to support its goal of 2.5 GW in contracted capacity. Management reiterated that its financing plans, coupled with strong customer commitments, provide clear visibility into both its funding runway and capacity roadmap.

Operationally, Nebius continues to evolve beyond infrastructure toward higher-value software and enterprise services. The company recently launched Nebius AI Cloud 3.0 “Aether,” an enterprise-grade cloud platform designed to give organizations more control over AI workloads, as well as Nebius Token Factory, an inference platform that helps businesses deploy open-source models at scale. These offerings aim to deepen customer relationships and create a defensible moat in an increasingly competitive market.

Beyond AI cloud operations, Nebius’s subsidiaries also showed progress. Its TripleTen edtech unit grew revenue by 100% year-over-year, while its Avride autonomous vehicle division secured a $375 million strategic investment from Uber and plans to launch public robotaxi service in Dallas by late 2025. While these businesses remain small contributors, they underscore the company’s broader diversification and technological reach.

In sum, Nebius’s Q3 results showcased exceptional growth, robust deal flow, and rising financial discipline, but also underscored the physical limits of AI infrastructure expansion. With demand still far outstripping supply, the company’s challenge now lies in execution—ramping capacity fast enough to meet multi-billion-dollar contracts without overextending its balance sheet. The stock’s muted reaction likely reflects investor caution after a massive run in AI infrastructure names this year, as markets weigh short-term bottlenecks against the company’s clear path toward multi-gigawatt scale.

If Nebius can deliver on its ambitious targets—$7–9 billion in ARR and 2.5 GW of contracted power by 2026—it will cement its place among the foundational infrastructure players of the AI era. But with margins tight, debt rising, and capacity stretched, execution will define whether it becomes a durable leader or another overextended player in the race to power artificial intelligence.

Comments



Add a public comment...
No comments

No comments yet