Nebius Says AI Computing Prices Are Climbing as Neocloud Investors Grow Wary

Nebius is charging more than expected for AI computing power, a senior executive told CNBC on Wednesday, pushing back against investor doubts about the business model shared by a new group of cloud providers.

Those providers, often called "neoclouds," spend billions buying Nvidia chips, building data centers and renting out that capacity. Investors have grown cautious after CoreWeave, one of the largest, saw its borrowing costs jump. Questions about how much debt is funding the data center buildout have weighed on peers including Iren and Nebius, whose shares have fallen sharply over the past month.

Marc Boroditsky, Nebius's chief revenue officer, said his company is not in CoreWeave's position because "we're not carrying the kind of debt that they are." He said that if Nebius brought its 2027 capacity online, it would sell out, and projected a similar dynamic in 2028 - a forward-looking view, he said, that supports longer-term demand as companies race to secure computing power. Asked whether price per megawatt, a key industry measure of what a company can charge for a given amount of capacity, would keep rising ahead of expectations, Boroditsky said: "Yeah."

Nebius ran an auction for computing capacity on Tuesday that showed continued price increases, he said. "We're seeing continued upward mobility in terms of the kind of price per megawatt," he told CNBC.

Chase Lochmiller, CEO of Nvidia-backed data center builder Crusoe, was also bullish, saying prices are rising for older Nvidia chips too. "We're seeing higher pricing today for A100s than we saw in 2020 or 2021," he said, adding that H100 chips also command more than when first deployed in 2023. He credited tremendous demand, driven by increasingly capable models that let people create more value from computing infrastructure.

The comments land amid an investor debate over how quickly Nvidia's chips lose economic value as more powerful ones arrive. The ability to keep renting older chips at attractive prices matters to AI infrastructure firms that must recover billions in GPU spending.

Source: CNBC