Nebius, a prominent player in the AI infrastructure sector, is experiencing a significant and unexpected rise in AI computing prices, according to a senior executive. Marc Boroditsky, Nebius’ chief revenue officer, stated that these price increases are outpacing the company’s projections, painting a positive picture for the company even as the broader “neocloud” industry faces intense investor examination.

The business models of companies dedicated to building AI infrastructure are currently under a microscope. Concerns have been amplified by the financial challenges faced by major competitors like CoreWeave, which has seen its borrowing costs escalate. These developments, coupled with questions surrounding the substantial debt used to finance data center expansions, have cast a shadow over other neocloud firms, including Iren and Nebius, leading to notable stock price declines in recent weeks.

Boroditsky, however, differentiated Nebius from companies like CoreWeave, emphasizing that his company is not burdened by comparable levels of debt. He further expressed confidence in Nebius’ future capacity, suggesting that if its planned 2027 computing capacity were available today, it would be fully subscribed. He anticipates a similar demand dynamic for 2028, indicating a robust long-term outlook for AI computing power as businesses increasingly seek access to essential computational resources.

Nebius recently conducted an auction for computing capacity, which Boroditsky noted as evidence of continued price appreciation. He described an “upward mobility” in the price per megawatt, a critical metric for neocloud companies that reflects the revenue generated from leasing computing power. This trend is supported by other industry leaders, such as Chase Lochmiller, CEO of Crusoe, who reported rising prices for various Nvidia GPUs, including older models, attributing the demand to the enhanced capabilities of AI models and the value they create.

Key Takeaways

• Nebius reports AI computing prices are rising faster than expected, defying investor concerns.
• The company highlights its lower debt levels compared to competitors like CoreWeave.
• Strong future demand is anticipated, with current and planned capacity expected to be fully utilized.

Editor’s Analysis & Impact

The AI infrastructure market is experiencing a dual narrative: intense investor scrutiny over debt-laden business models and robust demand driving up computing prices. Nebius’ optimistic outlook, supported by rising price-per-megawatt figures and a less leveraged balance sheet, suggests a potential bifurcation in the neocloud sector. Companies that can manage capital efficiently and secure high-demand capacity may thrive, while those heavily reliant on debt could face significant headwinds. The continued strength in GPU pricing, even for older models, indicates a persistent gap between supply and demand, which could benefit well-positioned players in the medium term.

Frequently Asked Questions

Q: What is a 'neocloud' company?
A: A 'neocloud' company refers to a new generation of cloud service providers that specialize in offering high-performance computing infrastructure, particularly for artificial intelligence workloads. These companies often focus on acquiring and deploying large quantities of GPUs (Graphics Processing Units) and building specialized data centers to rent out computing capacity.

Q: Why are AI computing prices rising?
A: AI computing prices are rising primarily due to an imbalance between soaring demand and limited supply. The rapid advancement of AI models requires immense computational power, driving businesses to secure access to GPUs and data center capacity. Simultaneously, the production and deployment of these resources face supply chain constraints and significant capital investment, leading to increased rental costs.

Q: What is 'price per megawatt' in this context?
A: In the context of AI infrastructure and neocloud companies, 'price per megawatt' is a key metric used to measure the revenue generated from leasing computing capacity. It indicates how much a company can charge customers for a specific amount of power and, by extension, the computing resources it supports, often reflecting the overall market demand and pricing power.