September 17, 2026, (Inside AI) — Nebius will raise pay-as-you-go prices for leasing selected Nvidia chips starting October 1, the company said on Thursday. The move marks its second price increase in three months.
The Amsterdam-based cloud provider did not disclose the size of the increase or specify which Nvidia models are affected. The change applies only to pay-as-you-go customers, not to those on long-term contracts.
The decision reflects a broader surge in demand for computing capacity to train and run artificial intelligence models. Rival CoreWeave said earlier the same day that it is signing new contracts at higher prices, signaling that the supply-demand imbalance in AI infrastructure is pushing costs upward across the industry.
Nebius, which trades on the Nasdaq under the ticker NBIS, has positioned itself as a specialist in high-performance cloud computing for AI workloads. The company operates data centers equipped with Nvidia's latest GPUs, including the H100 and H200 series. These chips are essential for training large language models and running inference at scale.
Read: Nvidia Faces Growth Test as Rubin Debut Meets AI Financing Scrutiny
The price hike comes as AI companies race to secure enough compute to power increasingly complex models. Training a single frontier model can require thousands of GPUs running for weeks. That demand has strained global supply chains and driven up costs for cloud providers, who pass those costs to customers.
Nebius is not alone in raising prices. CoreWeave, another major AI cloud provider, confirmed on Thursday that it is signing contracts at higher rates. The two companies compete directly for customers who need raw GPU capacity without the broader services offered by hyperscalers like Amazon Web Services, Microsoft Azure, and Google Cloud.
"Demand for computing capacity to train and run AI models is surging," a Nebius spokesperson said, confirming the price adjustment. The spokesperson declined to provide further details on the magnitude of the increase.
The timing of the hike is notable. It follows a similar increase by Nebius in July, when the company raised prices for certain GPU instances by an undisclosed amount. Two increases in three months suggest that the company is struggling to keep pace with the cost of acquiring and operating Nvidia hardware.
Nvidia's GPUs remain the industry standard for AI training. The company's data center revenue has soared over the past two years, but supply has struggled to meet demand. That scarcity gives cloud providers like Nebius and CoreWeave pricing power, at least in the short term.
Analysts say the price increases could accelerate a shift toward more efficient models and alternative hardware. Startups and research labs with limited budgets may delay projects or seek out cheaper options, including older GPUs or chips from AMD and Intel.
Read: Meta Touts Cost-Saving Benefits of In-House AI Chips
"If compute costs keep rising, we'll see more pressure to optimize," said Sarah Chen, a senior analyst at TechInsights. "Companies will look at quantization, pruning, and other techniques to reduce their GPU hours."
The price hike also raises questions about the sustainability of the AI cloud business. Nebius and CoreWeave have both invested heavily in data centers and GPU inventory. If demand cools or supply catches up, they could be left with excess capacity and falling prices.
For now, the market favors sellers. Nebius reported strong revenue growth in its most recent quarter, driven by AI customers. The company has expanded its data center footprint in Europe and the United States.
Customers affected by the October 1 increase will need to decide whether to absorb the higher costs, lock in longer-term contracts, or move workloads elsewhere. For many, the choice is not simple. Switching cloud providers can require significant engineering effort and downtime.
Nebius said it will notify customers directly about the new pricing. The company did not say whether further increases are planned.
