GPU Rental Data Challenges Nvidia Depreciation Concerns as Older Chips Retain Demand

Nvidia’s Older GPUs Become a Test of AI Infrastructure Economics

A debate over the useful life of artificial intelligence hardware has become an important issue surrounding Nvidia (NVDA) stock. Investor Michael Burry argues that major buyers of Nvidia GPUs are depreciating those assets too slowly, potentially making their annual profits appear higher than they would under shorter depreciation schedules.

Nvidia CEO Jensen Huang is pointing to the rental market as a counterpoint. A100 GPUs introduced in 2020 are still being contracted for use through 2029, while rental prices for newer H100 processors increased 22% in one month. Short seller Jim Chanos has challenged Huang's argument, keeping attention on whether AI computing equipment can retain its economic value as new generations arrive.

Key Points

• Michael Burry argues that AI companies using six-year useful lives for GPUs are spreading depreciation costs over too long a period, potentially increasing reported annual profits.
• Nvidia's A100 GPUs, introduced in 2020, are being contracted for use through 2029, while H100 rental prices recently increased 22% in one month.
• The dispute centers on economic useful life: whether older GPUs can remain productive and revenue-generating even as Nvidia introduces newer architectures.

Why Does GPU Depreciation Matter for Nvidia?

The controversy starts with a basic accounting question: how long does an expensive AI processor remain economically useful?

The supplied information says Google, Microsoft and Oracle estimate useful lives of about six years for their AI chips. Under that approach, the cost of the hardware is spread across those years as depreciation expense.

Burry argues that two to three years is more appropriate because Nvidia introduces increasingly advanced architectures at a rapid pace. He estimates that cloud providers could understate depreciation costs by approximately $176 billion between 2026 and 2028 by using longer useful lives.

The distinction matters because a longer depreciation schedule reduces the amount recognized as an expense each year. That results in higher annual reported profits than would be recorded if the same hardware were depreciated more quickly.

Meta provided an example of the accounting effect. After extending the useful life of certain servers to 5.5 years, the company said the change reduced its 2025 depreciation expense by approximately $2.9 billion.

The issue therefore extends beyond Nvidia's own financial statements. The debate concerns the economics and accounting of companies purchasing large quantities of Nvidia hardware to operate AI infrastructure.

Chanos has raised a related concern around the businesses renting GPU capacity. He is short data-center and neocloud companies and has described GPU rental as a commodity business, while warning that processors could become economically obsolete within three to four years even if they remain operational.

Older Nvidia GPUs Provide a Real-World Test

Recent GPU-market data provide measurable evidence for evaluating those concerns, although they do not settle the broader accounting debate on their own.

Nvidia launched its Ampere A100 GPU in 2020. CoreWeave has since signed a contract to rent A100 processors through 2029, extending their commercial use to nine years after their introduction.

Huang highlighted that agreement in August, describing Nvidia compute as durable, fungible and highly rentable.

The H100 provides another data point. The processor launched in 2022, yet rental prices increased 22% in one month to $3.28 per hour, according to information Huang shared from financial market platform Ornn Exchange on Sept. 8.

CoreWeave has reported similar behavior within its own GPU fleet. CEO Mike Intrator said a group of H100 GPUs coming off an expired contract was immediately rebooked at 95% of its original price.

Residual values provide another measure. Silicon Data placed the resale value of six-year-old A100 processors at approximately $5,000 as of September and reported that A100 pricing stopped declining in late 2025. H100 and B200 values were also higher in 2026 as rental rates increased.

These figures support Huang's argument that older Nvidia hardware can continue generating revenue years after introduction. They also sharpen the distinction at the center of the debate: technological age does not necessarily mean that a processor has stopped having economic value.

What Does the Huang-Chanos Exchange Add to the Debate?

Huang's public response has focused heavily on evidence from the rental market rather than a detailed accounting rebuttal.

His Sept. 8 post highlighted the 22% monthly increase in H100 rental prices. Chanos responded by asking why Nvidia does not rent the processors itself or continue raising prices if the assets retain such attractive economics.

Chanos later clarified that his criticism was aimed at companies purchasing Nvidia hardware and renting computing capacity rather than Nvidia itself.

That distinction is important. Burry's argument focuses on whether large technology companies are using depreciation schedules that are too long. Chanos is questioning the economics of businesses that acquire GPUs and then sell access to their computing capacity. Huang, meanwhile, is using utilization, rental pricing and older-chip demand to argue that Nvidia hardware retains productive value.

The available evidence establishes that older Nvidia processors remain commercially active. A100 GPUs are being contracted through 2029, H100 rental prices have recently increased, and expiring H100 capacity has been rebooked at close to its previous rate.

What those data do not establish by themselves is one universally appropriate accounting life for every GPU owner. Useful-life estimates and economic demand address related but distinct parts of the debate.

What It Means for Investors

The depreciation controversy matters because Nvidia increasingly presents AI compute as more than rapidly replaced technology.

The supplied information says the company has been positioning its computing systems as durable financial assets capable of retaining economic value. The continuing demand for older A100 and H100 processors provides current market data relevant to that argument.

At the same time, Burry's criticism raises a broader question about the financial results of Nvidia's customers. If GPU useful lives prove materially shorter than the schedules those companies use for accounting purposes, faster depreciation would increase annual expenses. If older processors continue generating meaningful revenue for longer periods, longer useful-life assumptions have more economic support.

The disagreement is therefore not simply about whether an older GPU still functions. It is about how long that hardware can remain economically productive as Nvidia releases newer processors.

Nvidia shares were down roughly 2.3% in the supplied market data. The depreciation controversy was also described as continuing to weigh on the stock, keeping GPU longevity and residual values relevant to the broader NVDA stock discussion.

Future rental prices, residual values and utilization of older GPU generations can provide additional evidence about how quickly AI hardware loses economic value.

Conclusion

The debate over Nvidia GPU depreciation now has a growing set of real-world data points.

Burry argues that two-to-three-year useful lives better reflect the rapid development of AI processors and that longer depreciation schedules inflate the reported profits of major GPU buyers. Chanos has separately questioned the economics of companies purchasing Nvidia processors to rent computing capacity.

Huang's counterargument is visible in the market for older hardware. A100 processors introduced in 2020 are contracted through 2029, H100 rental rates increased 22% in one month, and expired H100 capacity has been rebooked at 95% of its previous rate.

Those figures do not independently determine the appropriate accounting life of every AI processor, but they show that technological obsolescence and economic usefulness are not necessarily the same thing. How long Nvidia's older GPUs continue producing revenue remains a measurable part of the debate over AI infrastructure economics.

FAQs

What is Michael Burry's argument about Nvidia GPUs?

Michael Burry argues that major technology companies are depreciating Nvidia GPUs over useful lives that are too long. He believes two to three years is more appropriate than the roughly six-year estimates used by some companies.

Why does a GPU's useful life affect reported profits?

A longer useful life spreads the cost of a GPU across more years, reducing annual depreciation expense. Lower annual depreciation expense results in higher reported annual profits than would occur under a shorter depreciation schedule.

Are older Nvidia GPUs still being used?

Yes. Nvidia's A100 processors, introduced in 2020, are being contracted for use through 2029. The supplied information also says six-year-old A100 processors had resale values of approximately $5,000 as of September.

What happened to Nvidia H100 rental prices?

H100 rental prices increased 22% in one month to $3.28 per hour, according to data shared by Nvidia CEO Jensen Huang. CoreWeave also reported rebooking H100 capacity at 95% of its previous contract price.

Does the rental data settle the GPU depreciation debate?

No. The data show that older Nvidia GPUs continue to have commercial demand and economic value, but they do not independently establish the appropriate accounting useful life for every company or GPU deployment.

This article was created with AI assistance and reviewed by an editor. For details, please refer to our Terms of Use.

Go Beyond the Market Brief with Market Edge

Follow SharperTrades’ complete approach to trading and investing, combining active trade opportunities through Block Orders, long-term research through Stock Investor, and structured market education through the Swing Trading Masterclass. Try Market Edge for $19 your first month →

Explore Research with Stock Investor

Stock Investor is SharperTrades’ platform for long-term investing research and portfolio management. Members receive research reports, portfolio updates, conviction tracking, and in-depth analysis designed to support disciplined investment decisions.

Explore Active Trading & Income Strategies

Block Orders tracks institutional activity and highlights active trade setups and price behavior across long and short opportunities.

For options-focused traders, Essential Option Income provides a structured approach to options income strategies, while Pro Option Trader offers a broader range of options strategies and trade opportunities.

Think More Clearly with SteadyCapital

SteadyCapital is SharperTrades’ decision-support system for long-term investors, built around the SteadyCapital Method™. Review investment ideas, challenge assumptions, evaluate valuation and risk, compare companies, and think through important buy, hold, add, trim, or sell decisions before you act.

Risk Disclosure

All content is provided for educational purposes only and does not constitute investment advice. Trading involves risk, and past performance is not indicative of future results. Please review our full Risk Disclosure for additional information.