Nebius sees profits leap, hit by data centre delays

European AI infrastructure company Nebius today reported a surge in quarterly revenues to $981m but said some of its co-location data centre sites powering AI had been hit by delays.

The so-called neocloud also said it had seen a rise in adoption of open-weight models on its platform. Nebius reported a 571 per cent leap in year on year Q3 revenues to $981m, fuelled by demand for its AI infrastructure offering.

Nebius, which is sometimes referred to as a neocloud, builds and operates data centres, packing them with GPUs, then offers access to these data centres to AI and enterprise companies needing compute power, as well as offering them specialised software to run AI applications.

The Amsterdam-listed company reported a net loss of $37.5m in the quarter, a reduction compared to a loss of $119.6m the previous year. Nebius had lost more than a third of its value last month as investors questioned whether the AI infrastructure boom was becoming too expensive to sustain.

The company says it expects to break even on a pre-tax basis by the end of 2027.

Nebius today said it had been hit by delays to its co-location data centre sites, including its 300 MW facility in Vineland, New Jersey.

Nebius CEO Arkady Volozh said: “Building at this scale is hard, and delays are part of our industry right now. We are not immune. For example, some of our co-location sites, including New Jersey, are behind schedule. Customers have been kept informed and none of that capacity is cancelled.”

The delay at its New Jersey data centre campus, in which Microsoft is a major customer, is due to power delays, Nebius said.

The delay means that some of its revenue will shift from 2026 to 2027, Nebius said.

Commenting on the split between open weight and closed models on its platform, Roman Chernin, chief business officer, Nebius said: “We have absolutely seen the adoption of open models accelerate even in the last few months. 

“The bigger the diversification in the market is better for us. This diversification comes from open models or other ways to build specialised models for the use cases. Because it extends the universe of customers and partners we can work with.”