The dark satanic rumour mill has manufactured a hell-on-earth yarn claiming that Nvidia has reduced the supply of RTX 50-series GPUs to its board partners, squeezing availability as prices have already been creeping up for weeks.
On 16 January 2026, Nvidia issued a statement on the rumours, leaning on the familiar line that demand is high and memory is the bottleneck.
The spokesNvidia said: “Demand for GeForce RTX GPUs is strong, and memory supply is constrained. We continue to ship all GeForce SKUs and are working closely with our suppliers to maximise memory availability.”
Even with that denial-by-implication, hardware leaker @Zed__Wang claims Nvidia has cut shipments to partners by around 15 to 20 per cent, which is not the sort of trim that stays invisible at retail.
The same leak says GPUs will still be bundled with VRAM chips, which at least suggests Nvidia has not decided to make board partners play memory scavenger hunt on top of everything else.
Some chatter ties the cuts to DRAM shortages and earlier reports that certain GeForce models were discontinued, which would conveniently reduce the number of cards competing for the same pool of memory.
The knock-on rumour is that RTX 50 “Super” refreshes could be delayed or binned for the year because those parts were expected to use 3 GB GDDR7 modules, which would be even harder to source in volume.
If that lands, buyers are stuck fighting for what exists now, with no mid-cycle lifeboat to reset prices or improve configs.
The pain is not limited to Nvidia, with the Radeon RX 9070 XT surging in price and the RDNA 4 line already looking a bit sparse for a fresh launch.
As for RDNA 5 or UDNA, the word is you won’t see that before 2027, which leaves the whole market looking like it is being rationed by whoever can outbid everyone else for memory.
