Nebius reported fourth-quarter 2024 revenue of $37.9 million, up 466% on a year earlier

Written by Claude Sonnet 5.5, an AI model, from public records. Each figure in it was checked automatically against the source's text or our database before it was published.

Nebius Group N.V., the Amsterdam-based AI infrastructure company that trades on Nasdaq as NBIS, announced on Feb 20, 2025 that its revenue for the fourth quarter of 2024 was $37.9 million, up 466% from a year earlier. The release said the growth was driven primarily by the core AI infrastructure business, which grew 602% year over year and contributed over half of group revenue in the quarter. It furnished the announcement to the US Securities and Exchange Commission the same day in a Form 6-K, the report a foreign company files to pass on documents it has published.1^,2^,3^,4^,5^,6^,7

For the full year 2024, revenue was $117.5 million, up 462%. The net loss from continuing operations was $136.6 million in the quarter and $396.9 million in the year. Counting the divested businesses, which the release reports as discontinued operations, the net loss attributable to Nebius Group for 2024 was $644.8 million. Adjusted EBITDA, a measure outside US accounting rules (non-GAAP) that the company defines itself, was a loss of $75.5 million in the quarter and $266.4 million in the year. Capital expenditures were $417.6 million in the quarter and $808.1 million in 2024. The release says earlier periods were restated after the company sold its Russia-based businesses and changed its reporting currency from the Russian ruble to the US dollar, as if from Jan 1, 2022.8^,9^,10^,11^,12^,13^,14^,15^,16

In December 2024 the company announced a $700 million private placement, issuing 33,333,334 Class A shares at $21.00 a share. Founder and chief executive Arkady Volozh named Nvidia, Accel and Orbis among the partners in the raise. On products, the company said it launched a new cloud platform for AI workloads, the Nebius AI Cloud, and expanded its AI Studio offering with vision model support and new large language models, including DeepSeek R1 in January 2025. It also released Tracto.ai, a serverless platform for compute-intensive workloads. On capacity, it put a GPU cluster with NVIDIA H200 chips into operation in Paris in November 2024, and in December 2024 said it intended to deploy over 22,000 NVIDIA Blackwell GPUs in the U.S. and Finland in 2025.17^,18^,19^,20^,21^,22^,23^,24^,25

The release also reported annualized run-rate revenue (ARR), which the company defines as the last month's revenue multiplied by twelve. Core-business ARR was $90 million in December 2024, below earlier guidance; the company put that down mainly to longer lead times to win customers while it built out sales and marketing and moved customers to the new platform. Based on contracts in place, the company said the core business's March ARR would be at least $220 million, and Volozh said a projected December 2025 ARR of $750 million to $1 billion was within reach.26^,27^,28^,29^,19

The company described the results as unaudited, and its adjusted measures are its own rather than US GAAP figures.1^,14