[Memory shortage] AMD and Nvidia GPU Price Hikes Take Shape as Memory Surge Triggers Chain Reaction

Rising memory prices accelerate cost inflation
Price pressure from components to boards, PCs, and used products
Consumers across the IT market face mounting cost burdens

As surging demand from AI data centers drives memory prices sharply higher, manufacturing costs for graphics processing units are coming under significant pressure. Major GPU makers including Nvidia and AMD have notified customers of price increases as they move to protect margins. The impact is now spreading to finished PC prices, while higher costs for new products are pushing demand toward older and used models. With memory supply instability colliding with adjusted launch schedules, pricing structures across the IT hardware market are being reshaped.

Memory accounts for more than 80% of GPU manufacturing costs

According to IT-focused outlet Tom’s Hardware on January 21, Nvidia and AMD have informed board partners of impending price increases due to memory supply shortages driven by data center demand. AMD is set to raise prices starting in January, with Nvidia following in February. The increases are expected to hit consumer GPUs first, including Nvidia’s GeForce RTX 50 series and AMD’s Radeon RX 9000 lineup. Nvidia’s flagship RTX 5090, which carried a launch price of $1,999 last year, is now expected to climb to around $5,000.

The core driver is the dramatic expansion of memory’s share in GPU cost structures. Industry estimates put memory at slightly over 80% of total GPU manufacturing costs. While chipsets, packaging, and power components once dominated cost breakdowns, large-capacity, high-speed memory has become the decisive factor. Premium graphics memory such as GDDR6 and GDDR7 faces tight supply, with prices rising rapidly and directly amplifying cost pressure on GPU makers.

Crucially, memory price increases are cumulative rather than transitory, leaving GPU manufacturers with limited room to absorb costs. Prices for DDR5 16Gb memory used in GPUs have jumped from about $5.50 in May last year to more than $20 recently—more than a threefold increase in just over six months. Counterpoint Research forecasts an additional rise of over 40% through the second quarter, warning that maintaining current launch prices would severely erode profitability as memory suppliers continue raising prices.

Cost pressure is spreading beyond consumer GPUs to AI data center and server markets. While AI GPUs are typically sold under long-term supply contracts, new agreements signed this year are likely to reflect higher memory costs. Nvidia’s H200 AI accelerator uses six units of fifth-generation high-bandwidth memory (HBM3E), with supply prices for HBM3E already reported to be up about 20%. As GPU prices rise across server and AI markets, pricing adjustments are emerging as a systemic variable reshaping cost structures across the IT industry.

On the supply side, relief appears unlikely in the near term. Japan’s Kioxia has stated that its production for the year is already fully sold out, noting that while it is expanding capacity at its Yokkaichi and Kitakami plants, the pace falls short of surging AI data center demand. Taken together, Nvidia and AMD’s notifications to board partners suggest a preemptive response to what they see as a medium- to long-term memory price upcycle rather than a temporary adjustment.

The specter of “chipflation” draws closer

Industry watchers are now tracking how price increases originating in memory and GPUs are cascading through boards and into finished PCs. As GPU price hikes broaden, PC makers face cost burdens unlike those seen in previous cycles. Lenovo, the world’s largest PC vendor, along with HP and Dell, has begun reviewing product roadmaps for upcoming AI PCs and tablets. AI PCs typically ship with at least 16GB of memory, meaning higher DRAM prices directly inflate bill-of-materials costs.

With high-performance SSDs now effectively standard, memory and storage costs per device are climbing rapidly. Yet passing those costs on to consumers is difficult. AI PCs require a degree of price competitiveness to achieve mass adoption, but soaring memory prices leave manufacturers choosing between higher launch prices and downgraded specifications—both unattractive options. TrendForce expects DRAM and NAND flash shortages to persist through late this year and possibly into the first half of next year, noting that some vendors, including Lenovo, have secured inventories more than 50% above normal levels in preparation for a prolonged squeeze.

Pressure is not confined to GPUs and memory. Hard disk drives are also seeing sharp price increases amid AI data center demand and supply constraints. According to German tech outlet ComputerBase, average prices for 12 popular HDD models have risen 46% over the past four months, with some models up as much as 60%. Dell has raised prices on business laptops by up to 30%, while Asus has also adjusted pricing. As memory demand and prices enter a supercycle, fears of broad-based “chipflation” across electronics are becoming reality.

Used and older products also feel the squeeze

A more troubling dynamic is emerging as price hikes and launch delays push up prices for older and used products. As new-device prices climb, consumers are shifting toward previous-generation and secondhand options, triggering an immediate substitution effect. As of November last year, used prices for Nvidia RTX 30 and 40 series GPUs were 18% to 27% higher than a year earlier across major online marketplaces. With new products less accessible, older models offering comparatively better price-to-performance ratios are increasingly favored.

Launch delays have reinforced this trend. In December, Chinese smartphone maker Meizu announced the cancellation of the commercial launch of its Meizu 22 Air, citing rising memory costs. The company stated that the sharp increase in memory prices had become a major obstacle to its business plan. Memory typically accounts for up to 15% of smartphone manufacturing costs, but once that share exceeds 20%, options narrow to price hikes or launch postponements. Delays divert pent-up demand toward older devices, increasing turnover in secondary markets and propping up prices.

Industry observers also expect price increases for Samsung Electronics’ Galaxy S26 series, set to be unveiled in February. Samsung has attempted to mitigate costs by using its in-house Exynos 2600 application processor, but amid the broader semiconductor price uptrend, analysts expect price hikes of roughly $54 to $61. Samsung Electronics President Roh Tae-moon acknowledged the risk at CES 2026, saying the company is concerned about rising raw material costs for key components—especially memory—and warning that the increases will affect product pricing across the industry.