Nvidia has denied widespread hardware rumours claiming that production of its flagship GeForce RTX 5090 graphics card has been halted, with senior vice president Jeff Fisher confirming the high-end processor will continue shipping amid severe supply chain constraints.
Jeff Fisher Confirms RTX 5090 Production Continues Despite Supply Constraints
Reports surfaced across Taiwan and hardware enthusiast communities claiming that Nvidia had stopped manufacturing its largest Blackwell consumer die, the GB202 silicon, to redirect all available capacity toward more lucrative enterprise and professional AI-focused RTX Pro accelerators. Leaked notices shared by hardware sources suggested that the consumer flagship would be phased out entirely.
Nvidia Senior Jeff Fisher addressed the speculation directly by telling regional media outlets that the flagship processor remains a core product in the company’s lineup, according to coverage from the Taiwanese Central News Agency and 4Gamers.
Jeff Fisher stated that supply chain and capacity are indeed somewhat constrained, but Nvidia is going all out to get products to market. Although the RTX 5090 has other variants, it has certainly not been discontinued, and Nvidia will continue shipping it, which is very important.
Fisher acknowledged that availability remains tight, a reality reflected by retail shelves where the hardware has become increasingly difficult to secure. Major hardware retailers like Micro Center have removed the GPU from their inventory listings, while third-party marketplace pricing has climbed to as high as $9,600 in select regional listings due to acute scarcity.
Market Pressures and Professional Chip Allocation Fuel Discontinuation Rumors
The rumours gaining traction over the weekend did not emerge from a vacuum. Hardware analysts note that Nvidia maintains an eight-year pattern of building a single flagship silicon die—such as the GB202—and binning it to serve both consumer gaming and professional workstation markets, letting whichever sector generates higher revenue pull supply when capacity tightens, as detailed by Tech Insider.
A single wafer of GB202 yields higher profit margins when deployed into professional workstations like the RTX PRO 6000 Blackwell or the RTX PRO 5500 rather than consumer-grade GeForce RTX 5090 cards. Financial filings indicate that Nvidia’s data center revenue has grown to four to five times its consumer hardware revenue, shifting the company’s primary operational focus toward enterprise infrastructure.
Beyond silicon allocation, an industry-wide GDDR7 memory crunch has severely impacted manufacturing costs. Leaks regarding a potential 24GB GeForce RTX 5080 intended to replace the existing 16GB model have repeatedly slipped toward early 2027 because high-capacity 3GB memory modules remain expensive and scarce, according to reporting from Club386.
Rising Retail Costs and What PC Builders Face Next
With consumer hardware availability strained and component costs climbing, system builders are feeling the pinch across the board. A US-based custom PC manufacturer warned that retail GPU pricing is expected to climb significantly and that elevated price tags will likely persist for an extended period.
While standard consumer cards like the RTX 5080, RTX 5070 Ti, and RTX 5070 remain accessible for high-resolution gaming, enthusiast hardware remains heavily inflated.
Frequently Asked Questions About RTX 50-Series Availability
Is Nvidia officially discontinuing the GeForce RTX 5090?
No. Nvidia Senior Jeff Fisher formally denied discontinuation rumours, confirming that the flagship GPU is not end-of-life and will continue to ship to global markets.
Why are RTX 5090 cards so difficult to find at major retailers?
Nvidia has acknowledged severe supply chain constraints affecting the hardware. High demand from enterprise buyers alongside rising component and memory manufacturing costs have severely restricted available retail inventory.
What is the current status of the rumored 24GB RTX 5080?
Industry leaks indicate that a 24GB GDDR7 version of the RTX 5080 has been pushed toward the first quarter of 2027. The delay is driven primarily by high module costs and limited availability of high-capacity memory supplies.