Nvidia Corp.'s pricing power is on full display as the rental cost for its indispensable H100 AI chips has surged 20% since the start of the year, a direct reflection of insatiable demand from the world's largest technology companies.
Back
Nvidia H100 Rental Prices Jump 20% as AI Demand Outstrips Supply
The rental price for Nvidia Corp.’s essential H100 graphics processing unit has climbed 20% year-to-date, a stark indicator of the supply-demand imbalance at the heart of the artificial intelligence boom. The disclosure from Nvidia's Chief Financial Officer Colette Kress confirms that accelerating demand from model providers like OpenAI and Anthropic is granting the chipmaker significant pricing power, even as it ramps up production.
"Demand for Nvidia’s products has gone parabolic," CEO Jensen Huang said on the company's recent earnings call. Huang noted that the buildout of so-called AI factories is "accelerating at extraordinary speed," reinforcing the CFO's comments that AI is now a necessity for boosting productivity across industries.
The price pressure on rentals is underpinned by staggering growth in Nvidia's core business. The company's data center division, its main growth engine, saw revenue nearly double year-over-year to a record $75.2 billion in its fiscal first quarter, crushing analyst estimates of $73.2 billion. Nvidia projected that total revenue for the current quarter would rocket to $91 billion, far exceeding Wall Street’s forecast of approximately $86 billion.
This surge in demand creates a dual reality for the technology sector: while Nvidia solidifies its market dominance and financial might, companies reliant on renting its GPUs for AI model training face mounting margin pressure. Despite the blowout quarter, Nvidia's stock dipped more than 1.5% in late trading after the report, a sign of the sky-high expectations baked into the world's most valuable publicly traded company.
The Blackwell and Rubin Effect
Nvidia is not standing still. The company's next-generation "Blackwell" architecture is already being "adopted and deployed by every major hyperscaler, every cloud provider, and every major model maker," according to CFO Kress. Furthermore, the subsequent "Vera Rubin" line of chips remains on track for a launch in the second half of this year, with Huang telling analysts that "every single frontier model company will jump on Vera Rubin from the get-go." This aggressive product roadmap is designed to counter rising competition from rivals like Advanced Micro Devices Inc. and in-house chip development at major customers.
A $75 Billion Data Center Juggernaut
The scale of Nvidia's data center business is immense, accounting for over 92% of its total $81.6 billion in first-quarter revenue. Half of those data center sales came from its largest hyperscaler customers, including Amazon, Microsoft Corp., and Google parent Alphabet Inc. This highlights a critical dependency, as these cloud giants continue to spend billions on Nvidia hardware to power their AI services, even while investing in their own custom silicon as a long-term alternative.
The report solidifies Nvidia's role as the primary arms dealer in the AI revolution. Wedbush tech analyst Dan Ives estimates that for every $1 spent on Nvidia hardware, there is an $8 to $10 multiplier effect across the rest of the technology sector. For investors, the challenge remains assessing valuation. Susquehanna analyst Christopher Rolland, who rates the quarter an "A-minus," suggests that despite its massive market cap, Nvidia may be the cheapest AI stock available, trading at a more reasonable multiple than many peers when factoring in its growth. The company reinforced its commitment to shareholder returns by adding $80 billion to its buyback authorization and increasing its quarterly dividend to 25 cents a share.
This article is for informational purposes only and does not constitute investment advice.
[1] Nvidia earnings takeaways: Data center revenue nearly doubles, report is strong but stock slides[2] Nvidia Earnings Live: AI Chip Giant Beats Street Expectations on Profits, Sales[3] Forget the Memory Chip Arms Race. The Real Trade Is the 3 Equipment Companies Every Chipmaker Must Pay First[4] Dow futures rise 75 points: 5 things to know before market opens
