Can Older NVIDIA GPUs Like The A100 Earn A 143% Rental Price Premium?
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Ornn Data figures reported by a16z show rental prices for several NVIDIA GPUs increased from their January 1, 2026 levels, led by a 143% rise for the H200. The A100 rose 9%, while the data does not identify the causes of demand or account for supply, data-center capacity, and power constraints.

Rental prices for several NVIDIA GPUs rose during 2026, with the H200 recording a reported 143% increase from its January 1 level, according to Ornn Data figures reported by a16z. The A100’s increase was much smaller, at 9%, underscoring that the reported price gains were not shared evenly across older and newer accelerators.

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

The figures put the H200 at the top of the listed increases. The B200 rose 111%, the RTX PRO 6000 rose 73%, and the H100 rose 54%; the A100 recorded the smallest change at 9%. These percentages compare rental prices with their January 1, 2026 levels, rather than measuring prices across the entire GPU rental market.

The timeline was uneven. Most of the GPUs reached lower points in the first quarter, according to the reported chart. The H100 and B200 fell to roughly 23% below their January levels before prices rebounded in April and May. The H100 remained broadly flat until early May, then rose over several weeks. The H200 spiked late in May, gave back about a quarter of that rise within weeks, and then resumed climbing into October.

The report connects the general upward movement with a sharp rise in AI token consumption in spring, but the available figures do not establish that token use alone caused rental prices to rise. They also do not explain why one model’s price changed more than another’s.

At a glance
reportWhen: Data covers changes from January 1 thro…
The developmentGPU rental-price data tracked by Ornn Data and reported by a16z shows sharply different increases across NVIDIA models, with the H200 up 143% and the A100 up 9% from January 1, 2026.

Price Gains Vary by GPU Model

For customers renting compute, the figures suggest that model choice can make a substantial difference to rental costs. The reported H200 increase is far higher than the A100’s, while the H100 and B200 show their own distinct rises and dips. The percentages offer a snapshot of changing prices, not a guarantee of what a customer will pay: contract terms, location, availability, and rental duration are not detailed.

The data also adds evidence that demand for AI computing is affecting the market for more than the latest hardware. The A100’s comparatively modest reported increase does not mean it is no longer useful. Older accelerators may still appeal to users whose workloads fit their capabilities, including running medium-sized language models. The figures, however, do not provide model-by-model utilization or show how much rental activity involved older GPUs.

Price movements can matter to both buyers and renters weighing whether to secure capacity now or wait. But a price increase alone does not establish that a GPU is a better investment, that demand will persist, or that owners can resell hardware at a profit. Those conclusions would require additional information about costs, utilization, supply, and future pricing.

Products Worth Considering

Bose QuietComfort Wireless Noise-Cancelling Headphones, Sandstone

Bose QuietComfort Wireless Noise-Cancelling Headphones, Sandstone

Anker SOLIX C1000 Gen 2 Portable Power Station with 200W Solar Panel

Anker SOLIX C1000 Gen 2 Portable Power Station with 200W Solar Panel

As an affiliate, we earn on qualifying purchases.

Older GPUs Remain in the Mix

The reported figures cover the A100, H100, H200, B200, and RTX PRO 6000. They show different movements over the year rather than one uniform rise: some models fell below their January reference levels during the first quarter and later recovered, while the H200’s reported climb included a sharp late-May spike and subsequent pullback.

The article also points to the Tesla V100 as an example of an older GPU finding continued use during the AI boom. It describes the card as capable of supporting medium-sized language-model workloads and says higher-VRAM versions are being sold at a $200 premium. That is a separate observation from the rental-price percentages: no comparable V100 rental-price series or underlying sales data is provided.

While the V100 example suggests that age alone does not rule out use in AI workloads, the material does not establish how common these deployments are or whether they are comparable to enterprise-scale installations. It presents the card as an option for some consumers, not as a direct substitute for every newer accelerator.

Demand and Supply Drivers Remain Unknown

The figures leave the cause of the price changes largely unexplained. They do not identify whether the spring increase in AI token consumption came from agent-based workloads, enterprise adoption, other uses, or a mix of sources. The reported association between higher token use and rising rental prices is not, by itself, proof of causation.

The data also does not provide a breakdown of GPU supply, data-center capacity, or power constraints, any of which could affect rental availability and prices. The report does not specify its coverage, the number of listings or transactions measured, regional differences, or how the rental-price figures were calculated. Without those details, it is difficult to judge how broadly the changes apply.

It is also unclear whether the reported 143% H200 increase reflects typical realized rental rates, asking prices, or another pricing measure. The percentages should be read as the changes shown in the reported dataset, not as a universal rate for all providers or customers.

More Data Needed to Explain Pricing

The next useful development would be a fuller breakdown from Ornn Data or a16z explaining the dataset, price methodology, geographic coverage, and changes in available GPU capacity. A time series that separates listings from completed rentals and shows supply alongside demand would help clarify whether the swings reflect sustained demand, short-term availability, or both.

For now, the reported figures establish that rental prices moved differently by GPU model between January and October 2026, with the H200 showing the largest stated increase and the A100 the smallest among those listed. They do not show what prices will do next. Any continuing upward movement, or a broader increase in the value of older hardware, remains uncertain without updated pricing and utilization data.

Source: Wccftech

Key Questions

Which NVIDIA GPU had the largest reported rental-price increase?

The H200 had the largest reported increase, at 143% compared with its January 1, 2026 rental-price level.

How much did the A100 rental price rise?

The A100 rose 9% from its January 1, 2026 level, the smallest increase among the GPUs listed.

Did every GPU’s rental price rise steadily?

No. The report says most models reached lower points in the first quarter. The H100 and B200 were roughly 23% below their January levels before rebounding in April and May, while the H200’s late-May spike was followed by a pullback.

Does the data explain why rental prices increased?

No. It links the general rise with higher AI token consumption in spring, but does not identify the specific sources of demand or quantify the effects of GPU supply, data-center capacity, or power constraints.

Does the report show that older NVIDIA GPUs are profitable investments?

No. The figures describe reported rental-price changes and mention continued interest in older GPUs such as the Tesla V100. They do not provide enough information about hardware costs, utilization, resale prices, or future demand to establish profitability.

Source: Wccftech

You May Also Like

GigaToken: ~1000X Faster Language Model Tokenization

GigaToken introduces a new tokenization method that is approximately 1000 times faster, potentially transforming AI language processing efficiency.

AI Develops Cure for Common Cold Overnight – Pharmaceutical Companies Shocked

In a shocking turn of events, AI has discovered a cure for the common cold, leaving the pharmaceutical industry questioning the future of drug development.

Bitcoin Battles Unfold Live: Visualize the Crypto War

AIThis post was created with the assistance of artificial intelligence (AI).Experience the…

The Google I/O 2026 Preview: What May 19-20 Will Reveal About Google’s Agentic Bet

Google’s I/O 2026 event on May 19-20 is set to reveal major updates on its agentic AI platform, including Gemini 4.0 and multi-agent protocols, shaping the future of AI deployment.