Nvidia's RTX 50 graphics cards have gotten sharply more expensive in South Korea, with prices climbing as much as 30% in the latest round of increases, according to Tom's Hardware.
The report describes the hikes as hitting the entire RTX 50 lineup, not just one or two models. Premium cards are absorbing the worst of it: Tom's Hardware says the flagship RTX 5090 has pushed past $5,100 — well above what a halo consumer graphics card was expected to cost.
Two cost pressures are named as the drivers. The first is wafer price increases from TSMC, the Taiwanese contract manufacturer that actually fabricates Nvidia's chips. When TSMC raises what it charges per silicon wafer, every chip cut from that wafer costs more before it ever reaches a card. The second is memory: Tom's Hardware points to GDDR7 modules running roughly $20 apiece. A high-end card carries a lot of them, so a per-module price at that level adds up quickly across a board's memory configuration.
Tom's Hardware frames the development as "a troubling sign" — language that suggests the concern is less about South Korea specifically and more about what it signals for pricing elsewhere. The underlying costs, wafer contracts and memory modules, are set globally rather than by any single market. The available reporting covers South Korean pricing, and does not establish that comparable increases have landed in other regions.
Why it matters: graphics cards are the single most expensive component in most gaming and AI workstation builds, and when the squeeze comes from foundry and memory costs rather than retail markup or scalping, there's no easy fix — the increases tend to stick and to spread.