The $1,999 Flagship That Now Costs $4,299: Nvidia Reportedly Halts RTX 5090 Production to Free GB202 Chips for AI Cards

Nvidia's fastest gaming card is reportedly on its way out, not because gamers stopped wanting it, but because artificial intelligence customers can pay four times as much for the same piece of silicon. According to a supply-chain report from Wccftech, corroborated by the leaker MEGAsizeGPU and picked up by TweakTown, HotHardware and HWBusters, Nvidia has stopped assigning new GB202 chips to GeForce and is winding down production of the RTX 5090, including the China-specific RTX 5090 D v2. Nvidia has issued no statement and no board partner has confirmed the change, so the reports should be treated as unverified supply-allocation leaks rather than an official end-of-life announcement.

The claim is that every GB202 die, the largest consumer Blackwell chip, will now go exclusively to the RTX PRO workstation and AI lineup: the 96GB RTX PRO 6000, the 84GB RTX PRO 5500, and the 72GB or 48GB RTX PRO 5000. The economics are not subtle. A RTX 5090 launched at $1,999 in January 2025, while the same class of silicon powers professional cards that list above $8,000. With memory prices climbing and AI demand consuming fab capacity, the report describes a factory notice stating that GeForce's new ceiling is a refreshed RTX 5080, with the 16GB model also discontinued.

That successor, the RTX 5080 24GB, is essentially the long-rumored 5080 SUPER revived under a different name. Leaked specifications point to the GB203-450 chip with 10,752 CUDA cores, 24GB of GDDR7 built on 3GB modules, a 256-bit bus with memory speed raised from 30Gbps to 32Gbps for about 1,024GB/s of bandwidth, and board power up from 360W to 415W. Retail expectations fall between $1,999 and $2,999, which would put the top gaming card at roughly the RTX 5090's original launch price.

Anyone who wanted a 5090 at retail has not been able to get one at that price for a long time. The card was already trading around $4,299 at Walmart, above 5,000 euros in European listings, and between $6,500 and $9,500 from third-party sellers before the production reports surfaced, against a 2,099-euro MSRP in Europe. Tech Insider's report notes that no leaked document quantifies the production cut, and that existing RTX 5090 stock will sell through until it is gone.

The timing matters for what comes next. The RTX 60 series has reportedly slipped to 2028, which would leave the RTX 5080 24GB sitting at the top of the GeForce stack for a long time. It also means the decision, if accurate, is a permanent re-ranking rather than a temporary supply pause: Nvidia is telling enthusiasts that the largest chip of the generation belongs to the customers paying professional prices for AI memory capacity.

The episode crystallizes how the consumer GPU market now works. Gaming cards have become the residual claimant on silicon, memory and packaging capacity that the data-center business does not absorb first. A flagship dying quietly at age 21 months, with no launch event and no press release, is what that hierarchy looks like from the inside.