Nebius has announced a comprehensive price increase across its GPU cloud services, averaging roughly 20%, effective October 1. The move, coinciding with NVIDIA's stronger-than-expected quarterly results, sent the company's shares up nearly 7% in after-hours trading and reinforced the prevailing market view that AI compute supply-demand dynamics remain exceptionally tight.

This latest hike, covering H100, H200, B200, and B300 chip models, marks the second round of increases from Nebius since May. Investors interpreted the development as a clear signal that AI compute demand continues to overheat while supply stays constrained. The sentiment lifted Coreweave shares nearly 4% in after-hours trading, with the optical communications sector also gaining ground — AAOI and Credo Technology rose close to 3%, while Marvell Technology and Astera Labs advanced over 2%, and Coherent climbed roughly 2%.

The pricing news, arriving alongside NVIDIA's robust guidance, further solidified confidence in the longevity of the AI data center buildout cycle, sparking a broad recovery in sentiment across cloud and compute infrastructure stocks. Meanwhile, data center operators are seeing their bargaining power increase, with contract terms beginning to tilt in favor of the supply side — a subtle but significant shift in the industry's dynamics.

Rate Hike Details: Significant Increases Across Multiple GPU Models

Pricing screenshots shared by users on the X platform reveal notable adjustments in this round. The H100 price per GPU hour rises from $3.85 to $4.50, an increase of roughly 16.9%; the H200 jumps from $4.50 to $5.40, up 20%; the B200 climbs from $7.15 to $8.50, an 18.9% gain; and the B300 increases from $7.85 to $9.50, representing a 21% rise.

This marks Nebius's second price adjustment within a few months. In May, the company raised prices by an average of 29% for on-demand capacity and 51% for interruptible capacity. For the B300 specifically, prices have climbed roughly 56% cumulatively from around $6.10 per hour prior to May.

In a May email to customers, Nebius stated that the adjustment "reflects the continued strong demand for high-end GPU compute, and even with updated pricing, Nebius continues to offer some of the most competitive GPU infrastructure prices on the market." As of this writing, Nebius has not issued a public response to the latest round of reported price changes.

Supply-Demand Logic: Demand Visibility Exceeds 24 Months, Capacity Remains Critically Tight

Market analysts point to deeper supply-demand dynamics behind this price increase. According to analysis from X platform user @MelvinInvests, Nebius had previously attempted manual pricing adjustments, but management indicated that demand was so strong that price changes still failed to balance supply and demand. The company subsequently ran an auction test for scarce Blackwell compute, with customers actually paying 15% to 20% above the previous maximum pricing to secure access.

The analyst also noted that Nebius management revealed some customers have already reserved compute capacity for the first and second quarters of 2028, with certain orders involving tens of thousands of GPUs. The company's current demand visibility now exceeds 24 months, a notable extension from the previous 18 months.

Significantly, this round of increases covers both older Hopper-architecture GPUs and the newer Blackwell series, indicating that demand is not confined to a single chip generation but is broadly distributed across multiple types of AI compute.

"Nebius is simultaneously benefiting from two forces — expanding compute capacity on one hand, and charging more for existing capacity on the other," @MelvinInvests wrote. This suggests that with utilization remaining at high levels, revenue generated per available GPU hour will rise substantially, helping to improve profit margins and accelerate infrastructure investment payback.

Fundamental Backing: Major Contracts in Hand, Expansion Accelerates

Nebius's recent fundamentals also support the market's positive sentiment. Since securing a major contract with Microsoft in September of last year, the company has accelerated its expansion, acquiring Staryps, Eigen AI, and Tavily. In March, Nebius announced a $27 billion contract with Meta Platforms and received a $2 billion strategic investment from NVIDIA.

NVIDIA's quarterly results on Wednesday further confirmed the sustained strength of upstream demand, reinforcing the market logic that big tech's AI data center buildout boom will continue to drive compute supply chain demand.

X platform user Jonah Lupton commented that Nebius's average price increase of roughly 20% indicates significant bottlenecks in AI ecosystem compute supply. Based on this, he estimates that Nebius's current short-term customer contract pricing may already exceed $60 million per megawatt, with future contract pricing for Vera Rubin architecture products potentially surpassing $80 million per megawatt.

Nebius shares have gained approximately 150% year-to-date, making it one of the standout performers among AI-related stocks. Stocktwits data shows that NBIS-related discussions have surged over 120% in the past 30 days, with follower count up 7.3%. As of Thursday morning, retail sentiment remains "bullish," with message volume at "high" levels.

Negotiating Power Shift: Contract Terms for Data Center Operators Begin to Reverse

The supply-demand imbalance reflected in Nebius's price increases is reshaping contract negotiation dynamics across the data center industry.

Previously, large cloud providers held a dominant position in negotiations, often imposing extremely stringent requirements: server racks were expected to operate at nearly 100% uptime, and data center temperature and humidity standards were set at very strict levels. One data center executive revealed having seen contract terms where a single rack going down due to power failure, overheating, or switch malfunction would allow the cloud provider to cancel six months of rent. Accumulated SLA breaches could even result in the cloud provider terminating the lease entirely.

The executive noted that negotiating SLAs essentially involves a trade-off between "optimal pricing" and "contract durability" — the stricter the terms, the higher the price, but also the greater the risk. "Securing SLAs with lighter penalties is worth it even if the price is somewhat lower," he said.

However, as operator bargaining power rises, these extreme terms are gradually softening. Payment terms are also showing signs of a shift. A credit executive cited a case where a customer leasing only a small portion of a large data center facility was contractually obligated, upon payment default, to cover the full rent for the entire facility over a specified period. When questioned about this requirement, the data center owner was blunt himself: "He said, 'Look, we know this is absurd... but we can still make it happen.'"

With cloud providers like Microsoft eager to get NVIDIA server racks operational as quickly as possible, operators like CoreWeave are gaining leverage. The supply side of compute is increasingly asserting its upper hand in contract negotiations.