Nvidia Reportedly Halts RTX 5090 Production
By @sharedot · · 7 pages
- Tech
- Nvidia
- Gpu
- Gaming Hardware
Nvidia has reportedly stopped making RTX 5090 chips to prioritize AI cards, leaving the RTX 5080 24GB as its top gaming GPU.
The flagship GeForce card is reportedly done
Nvidia has reportedly ceased selling its GB202 graphics processors for consumer GeForce RTX 5090 boards, including the China-market 'D' variants, according to a leaked factory notice cited by leaker @hongxing2020 and confirmed by @Zed__Wang, who says GB202 silicon will now be used exclusively on data center and professional cards.
The notice states that going forward the highest-end GeForce card will be the RTX 5080 24GB, while the RTX 5080 16GB version will be discontinued. Tbreak separately reports that leaker MEGAsizeGPU posted on October 10 that 'all future GB202 supplies are RTX Pro exclusive.'
This is an unverified rumor, and Nvidia has not commented.
Why pulling the flagship makes economic sense
The alleged decision would make considerable economic sense for Nvidia, as the company can sell GB202 processors in data center and professional cards costing far more than the GeForce RTX 5090.
Tom's Hardware notes the RTX Pro 5500 features the same 21,760 CUDA cores as the RTX 5090 but with 84GB of GDDR7 instead of 32GB, while the fully-fledged RTX 6000 Pro with 24,064 CUDA cores and 96GB of memory can sell for around $15,000.
By discontinuing GB202-based GeForce products, Nvidia could keep its inexpensive consumer cards from competing with its own professional accelerators in the lucrative AI sector, redirecting limited manufacturing allocation toward higher-margin products.
The evidence is leaking, not announcing
The core claim rests on a leaked notice from an unknown factory cited by @hongxing2020, corroborated by leakers @Zed__Wang and MEGAsizeGPU, though Tbreak tags the story as unconfirmed and notes Nvidia has given no end-of-production date.
Tom's Hardware asked Nvidia for comment and has yet to receive a response, noting the company typically does not comment on rumors. Tbreak also reports that BenchLife says board partners have already stopped receiving RTX 5090 chips, another unconfirmed claim.
Notebookcheck points out that ending new GB202 allocations is not the same as officially discontinuing the card, since existing boards built from inventory can still be sold.
Supply drought and sky-high prices ahead
RTX 5090 boards are unlikely to disappear from the market overnight, as add-in-board partners, distributors, and retailers likely hold inventories that could sustain availability for some time.
However, if Nvidia no longer supplies GB202 processors for GeForce products, availability will inevitably decline and prices are expected to rise.
Tbreak reports the RTX 5090 has recently sold for $7,000 to $9,600 according to VideoCardz, though Notebookcheck puts it nearer $5,000 — market observations rather than Nvidia pricing.
Tbreak also raises the question of how warranty claims would be handled on a card that is no longer being built, referencing its earlier report on RTX 5090 connector melts.
What gamers get instead, and when
If the reports are accurate, the rumored RTX 5080 24GB becomes Nvidia's top gaming card.
Tbreak notes the configuration was first covered by VideoCardz in September 2024 and tied to a rumored RTX 5080 SUPER, whose launch was reportedly put on hold in July 2026 because 3GB GDDR7 modules were expensive — but with no more RTX 5090 32GB consuming memory supply, the new product would have better availability.
The rumored RTX 50 SUPER lineup is not tipped to use GB202, per Notebookcheck. Meanwhile, Rubin-based GeForce RTX 60-series GPUs are rumored to arrive only in late 2027 or even 2028, potentially leaving enthusiasts without a GB202-class flagship for years.
Sources
- tomshardware.com › Nvidia reportedly halts GeForce RTX 5090 production in favor of AI data center and professional GPUs
- tbreak.com › NVIDIA RTX 5090 reportedly ending, GB202 to RTX PRO
- tech.yahoo.com › Nvidia reportedly halts GeForce RTX 5090 production in favor of AI data center and professional GPUs