The Nvidia GeForce RTX 5090 is becoming increasingly difficult to find at retailers, and the reason may go far beyond normal demand from PC gamers.
A new report claims that AI companies are buying high-end graphics cards in bulk, sometimes purchasing entire pallets of GPUs at once. This puts regular consumers, PC builders, and gaming enthusiasts in direct competition with businesses that have significantly larger budgets.
For anyone hoping to upgrade to Nvidia’s flagship Blackwell graphics card, the situation could make an already expensive GPU even harder to obtain.
RTX 5090 Stock Is Becoming Increasingly Scarce
The GeForce RTX 5090 launched in early 2025 as the flagship gaming graphics card in Nvidia’s Blackwell generation.
More than a year later, availability remains an issue in several markets. Retail stock can disappear quickly, while some stores struggle to maintain consistent supplies.
The latest reports suggest that AI demand is contributing heavily to that problem.
Rather than buying one or two graphics cards, some AI-focused companies reportedly purchase GPUs in bulk. Entire pallets can be acquired for workstation systems, local AI infrastructure, model development, and other computational workloads.
That creates an unusual situation where someone building a gaming PC is effectively competing against businesses purchasing dozens or potentially hundreds of graphics cards.
Why AI Companies Want the RTX 5090
The RTX 5090 might be marketed primarily as a gaming graphics card, but its hardware makes it useful for considerably more than gaming.
The card includes 32GB of GDDR7 video memory, alongside the enormous amount of processing power provided by Nvidia’s Blackwell architecture.
That makes the RTX 5090 attractive for workloads such as AI development, rendering, content creation, machine learning, and other GPU-heavy applications.
Dedicated enterprise AI hardware remains the preferred solution for major data centers, but high-end consumer GPUs can provide an alternative for smaller companies, developers, researchers, and businesses building local AI systems.
When enough of those customers start buying cards simultaneously, gamers are left fighting over whatever inventory remains.
Gamers Cannot Compete With Bulk Buyers
This is where the situation becomes particularly frustrating for PC gamers.
An individual customer might spend thousands on a single RTX 5090. A business purchasing GPUs for AI workloads can potentially order an entire shipment without worrying nearly as much about the retail price of each individual card.
Retailers naturally cannot reserve every shipment exclusively for gamers.
As a result, stock that would normally slowly make its way into gaming PCs can instead disappear through bulk orders before ordinary consumers ever get a realistic opportunity to purchase it.
It also gives scalpers another opportunity. When legitimate retail availability becomes scarce while demand remains high, cards appearing on secondary marketplaces can command substantially higher prices.
The DRAM Shortage Is Making Things Even Worse
AI demand is not the only problem affecting graphics card prices.
The ongoing DRAM shortage has already pushed memory prices higher throughout 2026. RAM kits have become noticeably more expensive, while products relying heavily on high-performance memory are also feeling the pressure.
Graphics cards are particularly vulnerable.
The RTX 5090 carries 32GB of GDDR7 memory, meaning rising memory costs can have a significant impact on manufacturing costs and ultimately retail pricing.
Combine that with limited inventory and enormous demand from both consumers and businesses, and there is little pressure for RTX 5090 prices to move downward.
Instead, the opposite can happen.
The RTX 5090 Was Already an Expensive GPU
None of this would be quite as painful if the RTX 5090 were an affordable graphics card to begin with.
It isn’t.
As Nvidia’s flagship gaming GPU, the RTX 5090 sits firmly at the extreme end of the PC hardware market. Its performance is equally extreme, with enough power to handle demanding games at 4K while also providing considerable workstation capabilities.
For enthusiasts who simply want the fastest gaming hardware available, however, its usefulness outside gaming has become something of a disadvantage.
Gamers are no longer the only major audience interested in Nvidia’s most powerful consumer GPUs.
AI Demand Is Changing the GPU Market
The current situation feels familiar to anyone who remembers the cryptocurrency mining boom.
During previous GPU shortages, miners purchased enormous quantities of graphics cards because their computational performance could generate revenue. Gaming cards disappeared from shelves, second-hand prices exploded, and building a high-end gaming PC became significantly more expensive.
AI has created a different workload, but some of the consequences for consumers look remarkably similar.
High-end GPUs are valuable computational resources.
That means graphics cards designed and marketed toward gamers can quickly become attractive to industries willing to pay significantly more for the same hardware.
The difference this time is that AI demand could prove considerably more persistent.
Cryptocurrency mining profitability could collapse rapidly depending on market conditions. AI infrastructure, meanwhile, has become a major investment area across the technology industry.
Could Nvidia Restrict RTX 5090 Bulk Purchases?
Retailers could theoretically introduce purchase limits, although such restrictions would only partially address the problem.
Companies can purchase hardware through multiple retailers, distributors, system integrators, and business channels. Restricting individual retail customers to one RTX 5090 therefore does little if large quantities are being sold elsewhere.
Nvidia also has little financial incentive to prevent legitimate customers from purchasing its products.
From Nvidia’s perspective, an RTX 5090 sold to a gamer and an RTX 5090 sold to an AI company are both sold graphics cards.
For PC gamers, however, the distinction matters considerably.
Waiting Could Be the Better Option
Anyone currently planning an RTX 5090 build should be extremely careful about paying massively inflated prices simply because inventory is limited.
The RTX 5090 remains an incredibly powerful graphics card, but paying significantly above its intended retail pricing makes its already questionable price-to-performance proposition even harder to justify.
Unless the absolute fastest GPU is required, waiting for availability to improve or considering lower-tier hardware could make considerably more financial sense.
There is also no guarantee that today’s shortage will disappear quickly.
If AI companies continue purchasing consumer graphics cards in bulk while memory prices remain elevated, high-end GPU availability could remain under pressure well into the foreseeable future.
PC Gamers Are Once Again Fighting for GPUs
For years, PC gamers complained about cryptocurrency miners buying graphics cards faster than retailers could stock them.
Now history appears to be repeating itself with a different customer.
The RTX 5090 is not disappearing because everyone suddenly decided they needed the world’s fastest gaming PC. Its enormous computational performance and 32GB of GDDR7 memory also make it valuable hardware for AI workloads.
That leaves ordinary enthusiasts competing against companies capable of buying graphics cards by the pallet.
Combined with the ongoing memory shortage, high prices, and already limited RTX 5090 availability, upgrading to Nvidia’s flagship GPU could become increasingly difficult.
For PC gamers who thought the worst days of GPU shortages were finally behind them, 2026 may have other plans.
