Nvidia Discontinues RTX 5090 to Prioritize RTX Pro Chip Supply
A well-regarded leaker has reported that Nvidia has ceased production of the GB202 silicon for the RTX 5090, a chip that is presently commanding close to $16,000 in Nvidia’s professional series.
The assertion was made by an account on X, identified as Jukan, on October 10, stating: “The RTX 5090 has been discontinued.
The GB202 will now be allocated solely for the RTX Pro lineup. It’s over.” Subsequently, VideoCardz swiftly disseminated the information, with numerous media outlets such as Notebookcheck, Wccftech, and Xenospectrum echoing the report throughout the day.
If accurate, this development signifies that Nvidia’s premier gaming card, previously priced at $1,999 and above, is being quietly retired in favor of deploying the same chip into cards that fetch notably higher prices.
At the heart of this transition lies the GB202 chip, the robust Blackwell architecture powering the RTX 5090, which boasts an impressive 21,760 CUDA cores and 32GB of GDDR7 memory.
It also serves as the foundation for the RTX Pro 6000 Blackwell, a workstation card designed explicitly for AI training and inference applications.
Nvidia debuted the RTX Pro 6000 at a staggering $8,565 in April 2025. By June 2026, that price surged to $13,250, ultimately peaking at $16,000 by August of the same year, as reported by Tech Insider—a remarkable increase of approximately 87% over a span of sixteen months without any alterations to the hardware.
When a single chip can command such exorbitant margins, the rationale for offering it to gamers at a fraction of the cost diminishes significantly.
The implications for gaming hardware are noteworthy: in the absence of the RTX 5090, the RTX 5080 ascends as Nvidia’s premier consumer gaming card.
Wccftech has suggested that instead of a direct successor to the 5090, gamers might encounter a new RTX 5080 24GB variant.
Nvidia, however, has countered reports of the RTX 5090’s discontinuation, asserting that production continues alongside the RTX Pro lineup.
The company has a history of downplaying supply challenges, as seen earlier this year when it withdrew RTX 5090 and RTX 5080 Founders Edition listings from its online store in various regions, later justifying this as a temporary stock issue and characterizing these models as “limited edition products” subject to variable availability.
This situation is not unfolding in a vacuum. The global supply of GDDR7 and HBM memory is predominantly controlled by Samsung and SK Hynix, both of which are reportedly sold out.
SK Hynix disclosed in February 2026 that its HBM3e production capacity is fully allocated to confirmed clients through 2027.
Concurrently, Samsung has secured a contract to provide OpenAI with up to 800 million gigabits of HBM4 memory in the latter half of 2026 for their proprietary Titan AI chip.
Each wafer that is shipped to data center clients is a wafer that is not available for GeForce cards. Nvidia has previously acknowledged a 15% to 20% decrease in RTX 50-series supply attributable to GDDR7 constraints.
This shortage elucidates the price inflation of your forthcoming GPU, laptop, or game console. Prices for graphics cards, laptops, consoles, and NAS devices are now elevated by 6% to 18% over inflation, predominantly driven by increased production costs and scarcity of the memory components.
In contrast, Samsung is experiencing a revenue boom, with its semiconductor division reporting substantial profit growth, fueled by robust demand for HBM and enterprise memory—demand that historically was shared more evenly with the consumer GPU segment.
Herein lies a pressing concern for prospective graphics card purchasers in the coming year. The RTX 5090 is not merely an isolated incident; it serves as an ominous forecast.
If Nvidia can fetch $16,000 for a GB202 chip in a professional context compared to approximately $2,000 for the same component in a consumer-grade GeForce, then it follows that every future high-end consumer GPU is subject to mathematical scrutiny similar to that which affected this model.
The RTX 6090, whenever it materializes, will likely be produced by the same manufacturer employing the same logic amid ongoing memory shortages.
The RTX 5090 did not fall victim to poor design or a faltering market; rather, it was outbid by its more lucrative counterpart.
Source link: Startupfortune.com.
