There’s a recurring headline every launch cycle: prices are finally coming down. Then you go to buy a mid-range card and it costs what a whole decent PC used to. This isn’t bad luck or a temporary shortage anymore — a few things have structurally changed.
The mid-range moved up a tier in name only
What used to be called an “xx60” card — the one most people actually buy — has crept up in price while staying roughly the same slice of the stack. The naming makes it feel like you’re getting more; the price-to-performance curve says otherwise. Compare cost per frame at your resolution across the last three generations and the improvement is real but small, and mostly delivered by upscaling rather than raw silicon.
AI is buying the good wafers
The same fabs and the same advanced process nodes that make gaming GPUs also make data-center accelerators, and those sell for far more per wafer. When a manufacturer has to choose what to allocate capacity to, consumer graphics is not the priority. That pressure isn’t going away on any timeline that helps your next build.
“MSRP” is a launch-day myth
A handful of reference cards sell at the announced price for about a week. After that the models you can actually buy — better coolers, factory overclocks, the ones in stock — sit $80–150 above it. Reviews benchmark the MSRP card; you buy the $150-more one. Budget for the second number.
What to actually do about it
- Buy one tier down from where you think you need to be, and turn on upscaling. The quality gap is now small; the price gap isn’t.
- Buy last generation. The moment a new series lands, the previous one is the value play for a few months.
- Skip a generation. If your current card still hits your monitor’s refresh rate at settings you’re happy with, the upgrade math almost never works. Wait.
None of this is a reason to feel ripped off buying a GPU — it’s a reason to buy a smaller one than the marketing wants you to.