The RX 9060 XT followed a parallel trajectory, peaking above ¥85,000 in late January before falling to roughly ¥71,000 by late February, a 20% drop that some retailers deepened further via spring sale discounts to ¥59,800 .
The 2026 GPU market has been shaped by one dominant force: an AI-driven memory shortage that pushed GDDR6, GDDR7, and HBM costs sharply higher . Both AMD and Nvidia planned phased consumer GPU price increases starting in early 2026, and spot-market pricing for RX 9000 series cards initially jumped 10–18% in Europe and China
. Yet in Japan, those hikes unraveled within weeks.
Three factors explain the divergence:
1. Demand hit a hard price ceiling. Japanese coverage consistently reports that after the January spike, Radeon buyers simply stopped purchasing . GAZ:log data described "demand cooling as prices came down" and notebookcheck.net noted that "gamers simply stopped buying the aforementioned GPUs at such ludicrous prices"
. The demand drop was immediate enough to force retailers to act.
2. Retail inventory piled up. With buyers on strike, Japanese retailers accumulated unsold stock. The April 2026 price slide was directly attributed to inventory buildup, forcing sellers to cut prices aggressively to clear shelves . This is a classic correction pattern: speculative price hikes that outrun real demand produce inventory gluts, and retailers discount to restore volume.
3. AMD lacks Nvidia's pricing power. During the same period that RX 9070 XTs were falling toward ¥90,000 in Japan, Nvidia's RTX 5080 was still selling around $1,349 (35% above its $999 MSRP) and the RTX 5090 was reported at $3,000–$3,700 versus a $1,999 MSRP . The mid-range RTX 5070 Ti faced similar 25–50% premiums in reported retail data
. The contrast is stark: Nvidia's brand and feature set—including CUDA, DLSS, and AI/data-center mindshare—give it the ability to sustain elevated pricing that AMD's Radeon division cannot match.
The memory shortage narrative is real, and it is hitting Nvidia hard too—but in the opposite direction. GDDR7 supply was tight through early 2026, GDDR6 pricing reportedly increased, and HBM allocation for data-center GPUs remained heavily committed months in advance . Nvidia's RTX 50 cards, particularly those with 16GB or more VRAM, saw manufacturer and AIB price hikes of 10–20%
.
Unlike AMD, Nvidia's retail pricing absorbed and even exceeded these cost increases:
The memory-driven cost pressure is real and market-wide—it pushed RX 9000 spot prices up in January and has kept RTX 50 prices elevated all year. The difference is that Radeon demand collapsed when prices got too high, while GeForce demand remained sufficient to sustain large premiums.
The Japanese price data exposes a structural difference between AMD's and Nvidia's market positions:
AMD's Radeon brand is heavily value-dependent. When RX 9070 XT prices climbed toward ¥140,000 (roughly $875–$900 USD equivalent), demand evaporated . The subsequent discounting brought cards back into a range where buyers perceived value—close to or below the $599 USD MSRP in some cases
. This suggests a large portion of Radeon's addressable market is price-elastic and will defer purchases or switch to used/previous-gen options rather than pay premiums.
Nvidia's GeForce brand commands premium tolerance. Despite identical memory cost pressures and reports of supply cuts for mid-range RTX 50 cards, buyers continued paying 35–185% over MSRP . The combination of stronger gaming feature perception (DLSS, ray tracing performance), professional ecosystem lock-in (CUDA), and tighter supply from data-center competition for VRAM and silicon has created inelastic demand at elevated price points.
Inventory risk is asymmetric. Japanese retailers accumulated RX 9000 stock and cut prices; the same didn't happen to the same degree with RTX 50 cards, where pricing remained elevated into mid-2026 . This suggests AMD and its partners are more exposed to demand shocks when pricing overshoots perceived value.
The DRAM shortage is real, but it doesn't hit both brands equally. The same GDDR6/GDDR7 cost increases that squeezed AMD's margins also hit Nvidia, but Nvidia's ability to pass costs through to retail buyers—and then some—insulated it from the demand destruction AMD experienced in Japan .
For gamers frustrated by a year of punishing GPU prices, the RX 9070 XT's correction is a genuine opportunity. By May 2026, Japanese buyers could purchase the card below its initial launch MSRP, and US pricing was tracking in a similar direction, with the ASRock Challenger variant at $629 in March and the card available near $599 in select listings . European retailers also offered limited-time discounts pushing select models below euro MSRP
.
The catch is that these price levels depend on continued soft Radeon demand—and may not last if Nvidia's supply tightens further or if AMD cuts production in response to margin pressure. The memory shortage is expected to persist through 2028 according to some industry analysts, meaning the baseline cost floor for both brands remains elevated . But for buyers willing to choose Radeon over GeForce, early-to-mid 2026 offered the best value since the RX 9000 series launched.
The bottom line: AMD's RX 9070 XT didn't defy the DRAM shortage—it got punished by the pricing it enabled. When Radeon cards crept into Nvidia price territory, Japanese gamers walked away, and the market corrected with remarkable speed.