Nvidia has reportedly stopped producing the GeForce RTX 5090, redirecting the GB202 silicon behind its flagship gaming card to data center and professional boards — a decision that would push the consumer top end into an artificially scarce market.
According to Tom's Hardware, the notice covered every RTX 5090 model, including the “D” variants built for China, and pointed to an RTX 5080 with 24GB of memory as the new gaming flagship. The RTX 5090 launched in January 2025 at $1,999; since then it has traded well above list price, in some markets past €5,000.
The logic is straightforward arithmetic. A GB202 die sold as a data center or workstation board earns far more than one shipped in a consumer card, and AI demand has not slackened. With memory prices also climbing sharply, the bill of materials for a large consumer flagship looks worse every quarter.
The report is not yet confirmed by Nvidia, and supply signals in this segment have been wrong before. But the direction of travel is hard to miss: the top of the consumer GPU stack is increasingly a by-product of the AI market rather than a priority of it. If the RTX 5090 does end production, buyers should expect the halo card to become more expensive and harder to find — with the mainstream tier absorbing the consequences.
