Alex Duvall

Alex Duval from Goldman Sachs asked about Nebius Group's Q4 revenue being light versus consensus while ARR exceeded expectations, inquiring which metric is more meaningful and about timing considerations. He also sought clarification on the difference between the $7-$9 billion ARR guidance and the $3-$3.4 billion revenue guide for 2026.

Answer

Dado Alonso, CFO, stated that ARR of $1.2 billion exceeded guidance, while revenue was in the middle of guidance, as anticipated. He emphasized ARR as the more meaningful metric during hyper-growth. For 2026, he explained the revenue guidance of $3-$3.4 billion reflects a prudent approach, with the difference from ARR due to capacity deployment schedules, primarily in the second half, and the ramping of large enterprise partnerships.

Alex Duval from Goldman Sachs asked about the confidence behind Nebius Group's 40% adjusted EBITDA margin target for 2026.

Answer

Dado Alonso, CFO, explained that as Nebius Group scales, margins will increase. He noted that the core AI cloud business has a significantly higher adjusted EBITDA margin (24% in Q4) than the group (7% in Q4), and as the AI business scales, it will drive most revenue and margin. He added that other businesses are expected to operate at an EBITDA loss in 2026 but with a smaller contribution to the overall loss.

Alex Duvall from Goldman Sachs asked if it was fair to assume that 2.5 gigawatts of contracted power would equate to over $20 billion in revenue, and by when Nebius envisioned achieving this.

Answer

Andrey Korolenko, Head of Infrastructure, acknowledged that it's fair to assume the revenue potential, but clarified that CapEx for building out and deploying GPUs would be based on capital constraints and future demand. He emphasized the importance of being able to accelerate deployment when needed to avoid capacity constraints.

Alex Duvall from Goldman Sachs inquired if it's reasonable to assume that achieving 2.5 gigawatts of contracted power would translate to over $20 billion in revenue, and asked about the timeline and method for reaching this.

Answer

Andrey Korolenko, Head of Infrastructure, confirmed the revenue potential assumption is fair but clarified that actual CapEx for building out and deploying GPUs would be managed based on capital constraints and future demand. The strategy focuses on having the ability to accelerate deployment when needed and avoiding capacity bottlenecks.