Nvidia's older H100 graphics processing units (GPUs) are proving far more durable moneymakers than skeptics expected, with rental rates climbing 22% last month to $3.28 per hour. This development underscores a counterintuitive reality in the artificial intelligence boom: even as newer, faster chips flood the market, companies continue paying premium prices to rent aging hardware, keeping older systems profitable far longer than typical technology cycles would suggest.

The finding comes as Nvidia reported a blockbuster quarter with $96 billion in revenue, up 106% year-over-year, and guided next-quarter revenue to approximately $108 billion. But beneath those headline numbers lies a more nuanced story about how the AI infrastructure market actually works. Nvidia CEO Jensen Huang highlighted the staying power of the H100 chips, which debuted roughly three years ago, as evidence that demand for AI computing power continues to outpace supply across the industry.

Why Are Older Chips Still So Valuable?

The 22% monthly increase in H100 rental rates reveals something important about the current state of AI infrastructure: companies desperate for computing power are willing to pay premium prices for whatever hardware they can access, even if it's not the absolute latest generation. This dynamic has profound implications for Nvidia's business model and for the broader economics of AI deployment. When rental rates rise, it signals that demand for compute capacity is outstripping the available supply of newer systems, forcing customers to bid up prices for older equipment.

This phenomenon also challenges the narrative that newer always means better in the AI hardware space. While Nvidia's latest generation chips offer performance improvements, the H100s remain capable enough to handle most AI workloads. The rental market is essentially saying that having access to computing power today matters more than waiting for marginally faster hardware tomorrow. For companies training large language models or running inference at scale, the difference between a three-year-old chip and a brand-new one may not justify the wait or the premium price tag.

How to Understand Nvidia's Expanding AI Ecosystem Strategy

Beyond the H100 rental story, Nvidia is making strategic moves to deepen its position across the entire AI infrastructure landscape:

• Firmus Partnership: Nvidia invested in Firmus, which signed a multiyear deal to supply OpenAI from two Malaysian data centers and will deploy Nvidia Vera Rubin processors across Firmus's Asia-Pacific facilities, expanding Nvidia's reach into critical AI infrastructure projects.
• Mistral AI Funding: Nvidia funded Mistral AI to support model training and frontier AI research, expanding potential customers for Nvidia's chips and growing its AI compute ecosystem beyond OpenAI and Anthropic.
• Video Workflow Innovation: At IBC 2026, Nvidia and VAST demonstrated an AI video workflow using Nvidia Video Super Resolution and Beamr CABR on RTX PRO GPUs to upscale archival video and cut storage and delivery costs, showing how Nvidia chips enable practical applications beyond pure language model training.

These moves paint a picture of a company that is not resting on its dominance in data center GPUs. Instead, Nvidia is actively building out an ecosystem where its chips become essential infrastructure across multiple layers of the AI stack, from raw compute capacity to specialized applications like video processing.

What Do Analysts Make of Nvidia's Valuation?

Wall Street remains largely bullish on Nvidia despite ongoing skepticism from short-sellers. Analysts maintain Buy ratings on the stock, with price targets ranging from approximately $300 to $390, and a consensus target near $349.15. Citigroup recently raised its target to $315, signaling continued confidence in the company's growth trajectory. However, short-seller Jim Chanos has raised pointed questions about Nvidia's chip economics, specifically questioning why Nvidia does not rent chips directly, why it has not raised prices more aggressively, and whether cloud GPU providers are capturing most of the value in the rental market.

These questions highlight an ongoing debate about whether Nvidia's current valuation fully reflects the competitive dynamics of the AI infrastructure market. If cloud providers like Amazon Web Services, Microsoft Azure, and Google Cloud are capturing the majority of profits from renting Nvidia chips to end customers, then Nvidia's own margins may face pressure over time. The 22% increase in H100 rental rates suggests that demand remains strong, but it does not necessarily answer whether Nvidia itself is capturing enough of that value to justify its current stock price.

The H100 rental story ultimately reveals a market in transition. Nvidia has built an unassailable lead in AI chips, and older hardware continues to generate revenue as demand outpaces supply. But the company's long-term profitability will depend on whether it can maintain pricing power and ecosystem control as competition intensifies and cloud providers become more sophisticated at managing their own hardware investments.