AI's Memory Tax

Unimicron supplies advanced IC substrates and high-density interconnect boards needed for large AI accelerator packages carrying processors and HBM.

Although focused on conventional DRAM rather than leading-edge HBM, Nanya can benefit when HBM absorbs industry capacity and tightens broader DRAM supply.

IBIDEN produces high-end package substrates used with data-center processors and accelerators, whose integration with HBM drives larger and more technically demanding substrates.

AI accelerators and HBM assemblies require extensive testing to prevent costly package-level failures.

ASE's advanced packaging and testing capabilities position it to capture greater outsourced content from increasingly complex AI accelerator and memory packages.

Resonac supplies advanced semiconductor-packaging materials, including die-bonding films and encapsulation products, that benefit from the greater packaging complexity and material intensity of HBM-based AI accelerators.

Enterprise flash demand can benefit as AI workloads increase the need for fast local and tiered storage.

Entegris provides specialty materials, filtration and contamination-control products consumed across memory fabrication, so added HBM capacity increases recurring process-material demand.

HBM qualification and a concentrated global DRAM market give Micron substantial operating leverage to higher memory prices.

AI training and inference create large volumes of data that sustain demand for high-capacity enterprise hard drives.

Transsion competes heavily in price-sensitive emerging-market smartphones, where higher DRAM and flash costs are difficult to pass through without hurting demand.

Dixon's low-margin electronics and smartphone assembly operations can face working-capital and margin pressure when memory costs rise faster than customer pricing adjusts.

As a high-volume, relatively low-margin smartphone and computing-device ODM, Huaqin is vulnerable when memory costs rise faster than customer pricing resets.

Memory inflation pressures the deliberately thin hardware margins of its smartphones and connected devices.

Wingtech’s low-margin smartphone, notebook and consumer-electronics manufacturing operations face margin pressure when DRAM and storage costs rise faster than contract pricing can be adjusted.

Longcheer's contract manufacturing of smartphones and smart devices leaves it exposed to higher DRAM and NAND bills of materials with limited pricing power.

Apple consumes large volumes of DRAM and NAND in premium devices, so a sustained memory-price upcycle raises hardware bills of materials despite its substantial purchasing power.

Compal's notebook and device manufacturing operations operate on narrow margins, making a PC-memory cost upcycle a risk when customer reimbursements lag.

Higher component costs can squeeze profitability across Acer's price-sensitive consumer PC portfolio.

Pegatron's thin-margin consumer-electronics assembly model is vulnerable to component-cost volatility and delays in passing higher memory prices to brand customers.