I remember when a $999 GPU felt like the “expensive but justifiable” tier, something buyers saved months for before building their dream gaming PC. Nvidia’s about to make that price feel affordable.
Back in January, reports surfaced that AMD and Nvidia had warned board partners about incoming price increases.
So how big are these price hikes actually?
A week later, the claims were followed by a report of a roughly 15-20% hike across Nvidia’s entire RTX 50 lineup. Nearly six months later, actual numbers are surfacing, and if you ask me, they’re quite rough.
Based on current Chinese market rates, which may shift slightly once US pricing lands, the RTX 5060 Ti might cost around $783, up from the $429 launch price. The RTX 5070 is heading toward $973, and the RTX 5070 Ti toward $1,408, nearly double their price at launch: $549 and $749, respectively.
The top of the stack takes the biggest hit, though, primarily because their original price was already too high. The RTX 5080, originally $999 at launch, is reportedly climbing to around $1,845, while the flagship RTX 5090 is jumping from $1,999 to a wild $3,843.
Why is this happening?
Even more problematic is the fact that these are Nvidia’s baseline prices. Board partner cards with custom cooling or factory overclocks will likely cost even more. And the driving force behind all this isn’t really gaming demand at all; it’s AI.
Nvidia’s data center business, one that brings in high-revenue orders, has made GPU manufacturing capacity a scarce, more profitable resource. Consumer-grade gaming cards are increasingly treated as the lower priority.
If you’re holding onto a current GPU, this might be the year to make it last. Between console price hikes and now this, PC gaming is edging toward becoming a genuinely expensive hobby. At this rate, building a gaming rig could soon require enthusiast-level budgets by default.
If this pattern holds, expect AMD to follow with comparable increases rather than undercut Nvidia meaningfully, since both companies ultimately compete for the same constrained manufacturing capacity at TSMC.
