Nvidia's $5 Trillion Valuation Faces a Reality Check as Chip Rental Prices Surge 114%

Nvidia's stock nears €180 amid AI boom, but rental prices for B200 chips skyrocket to $5/hour, with some firms spending more on compute than payroll.

The numbers coming out of Nvidia’s ecosystem are starting to look like science fiction. The chipmaker’s stock is trading near €180, just shy of its 52-week high, having climbed 25% over the past month and 88% over the past year. But beneath that rally, a more complicated picture is emerging — one where the cost of accessing Nvidia’s hardware is spiraling out of control even as the company prepares to report its next quarterly results on May 20.

A Market in Fever Pitch

According to the “Ornn Price Index,” the rental price for a single Nvidia B200 chip has surged to nearly $5 per hour — a 114% jump in just six weeks. On the secondary market, a single B200 chip now commands an estimated $500,000. The bottleneck is so severe that Google is reportedly locking up a significant chunk of available capacity, leaving other players scrambling. Analysts don’t expect any relief until the second half of 2026.

This price explosion is reshaping the economics of the entire AI industry. Nvidia Vice President Bryan Catanzaro has flagged a startling new trend: for some companies, the cost of AI compute is now exceeding their total employee payroll. Industry insiders have coined the term “Tokenmaxxing” to describe this phenomenon. Tech giants are on track to spend a staggering $740 billion on AI infrastructure this year, straining balance sheets across the sector. Uber, for instance, has already exhausted its entire 2026 AI budget.

Even the industry’s biggest names aren’t immune. OpenAI is expected to burn through $25 billion this year, and after missing its internal revenue targets for the first quarter, the ChatGPT developer ended its exclusive partnership with Microsoft Azure. It’s now shopping around for cheaper compute capacity from other cloud providers.

Nvidia’s Software Offensive

Against this backdrop of hardware scarcity, Nvidia is doubling down on software. The company recently launched Nemotron 3 Nano Omni, a new AI model that processes images, audio, and text simultaneously. Early adopters include Palantir and Foxconn. The model runs across all major Nvidia architectures and offers significantly higher capacity for complex document analysis, according to technical specifications.

The software push comes as Nvidia also moves into new verticals. French telecom operator Orange is exploring a strategic shift to use Nvidia graphics processors for its 5G infrastructure, moving away from custom-designed chips. The company is partnering with Nokia on the initiative, though Orange’s technology chief Laurent Leboucher has warned of the risk posed by heavy dependence on Nvidia’s development platform.

Supply Chain and Competition

On the manufacturing front, Nvidia’s supply chain is scrambling to keep up. Tower Semiconductor, which supplies critical components for Nvidia’s data center modules, is quintupling its production capacity to meet surging demand.

Meanwhile, the competitive landscape is heating up. AMD and Intel have jointly proposed new standards for their architectures, aiming to unify AI acceleration across the industry. Google is also exploring alternative technical paths for its infrastructure. Nvidia CEO Jensen Huang, however, has dismissed these as immediate threats, insisting the company isn’t resting on its laurels.

What to Watch on May 20

All eyes are now on Nvidia’s fiscal first-quarter results, due after the close of European trading on May 20. Management has guided for revenue of approximately $78 billion, excluding its China business. Chief Financial Officer Colette Kress will provide written commentary, followed by a Q&A session with analysts.

If Nvidia hits that $78 billion target, it would provide strong support for the company’s current valuation of over $5 trillion. But with hardware prices spiraling and customers’ budgets under unprecedented pressure, the sustainability of that valuation is becoming an increasingly urgent question. Nvidia is already developing its next-generation platform, Vera Rubin, slated for release toward the end of 2026, while the current Blackwell generation continues to ship — at increasingly eye-watering prices.