NVIDIA Confirms the RTX 5090 Is Not Discontinued as GB202 Allocations Stay Active

NVIDIA has issued a public statement on the future of the GeForce RTX 5090, and the message is direct: the flagship graphics card has not been discontinued. The company says allocations of the GB202 graphics processor remain active for its add-in board partners, along with the 32GB of GDDR7 memory that ships with the card.
That statement matters because reports from distribution channels in Europe and Asia had suggested NVIDIA stopped taking new orders for GB202 silicon and matching memory kits. Those reports fed speculation that desktop flagship production was being wound down in favor of enterprise AI hardware. NVIDIA now says that is not the case, and buyers should treat the discontinuation talk as rumor rather than fact.
What NVIDIA Actually Said
NVIDIA confirmed that the RTX 5090 has not been discontinued and that GB202 allocations remain active for board partners. Jeff Fisher, Senior Vice President of Personal Computing at NVIDIA, put it plainly in the statement reported by Guru3D:
“The supply chain (capacity) is indeed somewhat constrained, but NVIDIA is fully committed to bringing products to market. Although there are other variants of the RTX 5090 chip, it has not been discontinued, and NVIDIA will continue shipping, which is very important.”
NVIDIA also explained that purchase order schedules change routinely. Foundry wafer allocation, assembly lead times and global logistics cycles all shift how orders are taken in, so temporary gaps in order intake can appear as production moves between manufacturing batches. According to the report, the TSMC fabrication lines dedicated to the GB202 processor remain operational.
The company added that desktop graphics cards based on the Ada Lovelace and Blackwell architectures will keep active production runs alongside its enterprise accelerator platforms. In other words, consumer cards and data center hardware are running in parallel, not one replacing the other.
Where the Discontinuation Rumor Came From
The speculation did not come from NVIDIA. It grew out of reports from regional distribution channels, where custom graphics card makers were said to have hit fulfillment holds on bulk wafer shipments. With board partners reportedly unable to place new GPU orders, some observers concluded that desktop flagship production was being phased out.
That conclusion was always an inference, not a confirmed fact. No NVIDIA announcement had said the card was ending, and the new statement directly contradicts the rumor. The gap between what distributors reported and what NVIDIA now confirms shows why supply chain chatter should be treated carefully until the manufacturer speaks.
How the rumor spread so fast
A discontinuation rumor does not need an official source to travel. It needs a plausible detail and a worried audience. The detail here was real enough: distribution channels in Europe and Asia reported that custom card makers had hit fulfillment holds on bulk wafer shipments, and board partners could not place new GPU orders for a time.
From there, the jump to “discontinued” was short. If partners cannot order new chips, the thinking went, production must be ending. The problem is that a paused order window and a cancelled product are two different things. NVIDIA’s own explanation covers the gap: purchase order schedules change routinely as foundry wafer allocation, assembly lead times, and global logistics shift between manufacturing batches.
Social media did the rest. A supply chain report becomes a headline, the headline becomes a certainty, and by the time the manufacturer responds, the rumor has already set prices and buying decisions. NVIDIA’s statement is the correction, but corrections travel slower than the rumor that made them necessary.
Why Retail Stock Still Feels Tight
A statement that production continues does not mean shelves will fill up tomorrow. Industry analysis in the report points to a second source of pressure: enterprise demand. Retail GeForce RTX 5090 cards are reportedly being acquired in volume by third-party integrators and reconfigured into dual-slot blower assemblies for AI compute servers. That secondary demand pulls consumer retail stock away from normal PC hardware channels, which makes the shortage feel worse than the production numbers alone would suggest.
Add-in board partners including ASUS, MSI, Gigabyte and Palit report that existing inventory allocations will continue to ship to retail distributors to fill backordered channel agreements. Even so, custom card availability is expected to stay tight over the coming quarter until silicon distribution rates stabilize. Guru3D reports that position as the current channel outlook, not as a promise of a specific restock date.
That shortage lines up with the price surge we covered earlier: RTX 5090 cards have climbed far above MSRP, with some models reportedly selling at several times their launch price. A card that is still in production can still be scarce and expensive when demand from outside the gaming market keeps absorbing supply.
The AI demand problem in plain terms
The second pressure on RTX 5090 stock has nothing to do with gamers. Third-party integrators are reportedly buying retail GeForce RTX 5090 cards in volume and reconfiguring them into dual-slot blower assemblies for AI compute servers. A card built for a gaming PC becomes a cheap way to get a lot of compute into a server rack.
That pulls stock out of the normal retail channel. The card still exists, the factory still ships it, but a share of every batch never reaches a store shelf. It goes straight into a server instead. Gamers see empty listings and high prices, and assume production stopped, when the real story is that someone else bought the cards first.
NVIDIA cannot simply make more to fix this. The GB202 processor runs on TSMC’s custom 4N process, and that capacity is shared with the company’s enterprise accelerator platforms. Consumer cards and data center hardware run in parallel, as NVIDIA itself confirmed, which means gaming supply competes for the same factory lines. Until that balance shifts, retail stock will stay tight even with production running.
RTX 5090 Supply Status at a Glance
| Supply Status Parameter | Detail / Status |
|---|---|
| GPU | NVIDIA GB202 |
| VRAM configuration | 32GB GDDR7 |
| Official status | Active production, not discontinued |
| Foundry partner | TSMC custom 4N process |
| Allocation channels | Consumer board partners and enterprise server conversions |
| Retail availability | Constrained, periodic batch shipments |
What This Means If You Are Trying to Buy One
- Do not treat the RTX 5090 as end of life. NVIDIA says shipping continues, so a listing that calls the card discontinued is working from the rumor, not from the manufacturer.
- Expect tight stock and high prices for now. Production staying active is good news for the long term, but it does not fix this quarter. Watch official retail partners for restocks as new shipments arrive, rather than paying scalper prices on impulse.
- The same supply pressure is visible elsewhere in the lineup. The RTX 5080 24GB launch is reportedly in doubt as board partners wait on design kits, which points to allocation limits that go beyond a single card.
- Rival cards are picking up the slack in the meantime. AMD outsold NVIDIA on Amazon US in October, with the RX 9070 XT topping the charts, so buyers who cannot wait have credible alternatives at lower price points.
What to Watch Next
The statement settles the discontinuation question, but three things remain open. First, whether board partner order windows reopen on a regular cadence or stay stop-start. Second, whether retail prices ease as new batch shipments reach shelves. Third, whether enterprise demand keeps absorbing consumer stock at the same rate.
Until those questions get answers from NVIDIA or a board partner, treat any new claim about RTX 5090 supply as a report, not a fact. What is confirmed today is narrow and specific: the card is not discontinued, GB202 allocations are active, and NVIDIA says it will keep shipping. For anyone saving up for a flagship upgrade, that is a better starting point than the rumor mill offered yesterday.
Source: Guru3D









