The fresh news: next-gen may be further off
The most concrete GPU story of the last two weeks is a discouraging one for anyone waiting to upgrade. On 21 September 2026, tbreak reported that the well-known leaker Kopite7Kimi indicated Nvidia's next consumer generation, the RTX 60 series, could slip to 2028, later than the 2027 window earlier leaks had suggested. tbreak tied the reported delay to Nvidia's focus on AI accelerators, noting that data-centre products command much higher prices than gaming graphics cards, so supply and engineering capacity could be prioritised elsewhere.
The caveat is large, and tbreak made it plainly. The outlet stressed the claim is unconfirmed and that Nvidia has not announced a consumer launch date, so 2028 is a leak rather than a release plan. We treat it as a rumor, not a roadmap, and note that leaker timelines have shifted before.
Why today's cards are not getting cheaper
The backdrop to all of this is memory. In reporting earlier in 2026, both Vice (3 January 2026) and Astute Group (2 February 2026) described memory costs becoming the dominant driver of GPU pricing, as AI data-centre demand soaked up supply of the graphics memory that cards depend on. Astute cited an industry estimate that VRAM now accounts for over 80% of the bill of materials on some high-end cards.
An industry source cited by TrendForce put it bluntly, in a line carried by Astute Group: rising memory prices are no longer a marginal cost issue, they are becoming the dominant factor in GPU pricing. When the memory is the expensive part, a new architecture on its own cannot bring prices down.
“Rising memory prices are no longer a marginal cost issue, they are becoming the dominant factor in GPU pricing.”
The numbers being floated
Those reports pointed to phased increases rather than a single jump, with AMD adjustments beginning in January 2026 and Nvidia following in February, per Vice, which traced the original report to South Korean outlet Newsis. The same coverage forecast the flagship RTX 5090 approaching $5,000 by late 2026, with custom versions already priced above $3,000 (Astute Group, Vice).
Both outlets also reported Nvidia was cutting output of the RTX 5070 and RTX 5060 Ti by 30 to 40% to steer scarce memory toward higher-margin AI chips. These were forecasts and unofficial reports at the time, and Vice noted that neither Nvidia nor AMD had confirmed the increases. We flag them as such.
What to do about it
Put the pieces together and the signals argue against waiting. A next-generation GeForce launch is, per the latest leak, potentially years out, and the memory crunch that lifted prices through 2026 shows no quick resolution. There is no obvious catalyst on the near horizon for either cheaper current cards or a new generation.
So if you genuinely need a card and can find one at a fair price, the case for buying now is stronger than the case for holding out for a newer or cheaper part that the reporting does not promise is coming soon.
Our take
We are labelling the RTX 60 timing as an unconfirmed leak and the pricing figures as 2026 reports and forecasts, not settled fact. What is not in doubt is the direction. Through 2026, GPU pricing has been shaped more by AI-driven memory demand than by any new gaming architecture, and until that supply picture eases, that is the story that matters for buyers.
Frequently asked
When will the Nvidia RTX 60 series launch?
Nvidia has not announced a date. A 21 September 2026 report relayed a leaker's claim of a possible 2028 window, later than earlier 2027 rumors, but tbreak stressed this is unconfirmed.
Why are GPU prices rising in 2026?
Reporting from Vice and Astute Group points to AI-driven demand for memory, with VRAM cited as over 80% of some high-end cards' build cost. Both noted the increases were unofficial reports at the time.
Should I wait for cheaper GPUs?
The signals do not favour waiting: next-gen is reportedly years out and the 2026 memory crunch has pushed current prices up, per these reports. Buying at a fair price now is defensible.