Surging AI memory demand extends chip supply crunch

Next-generation artificial intelligence processors are demanding unprecedented amounts of high bandwidth memory, creating severe supply constraints across the semiconductor industry. To process massive volumes of data efficiently, AI chips require that data to be stored nearby on specialized HBM chips. Nvidia’s upcoming Rubin GPU is designed to utilize up to 288 gigabytes of HBM4 memory, nearly triple the capacity of its Blackwell predecessor. Rival Advanced Micro Devices is pushing this boundary even further, with its MI400 series accelerators engineered to use up to 432 gigabytes of HBM4.

This exponential jump in data requirements is actively tightening the broader memory market. Because manufacturing HBM consumes significantly more wafer supply compared to conventional memory, ramping up production directly reduces the manufacturing capacity available for standard DRAM and NAND chips. Micron Technology management anticipates that these broader memory shortages will persist well beyond 2027.

Resolving this structural bottleneck requires major memory producers to build large, highly complex new chip foundries. However, this expansion is proceeding slowly due to long construction timelines, acute shortages of skilled labor, restrictive permitting requirements, and the need for additional energy infrastructure. To navigate these physical constraints, memory producers are instead forced to invest heavily in advanced manufacturing and packaging equipment to maximize output from existing facilities.

This sustained capital expenditure cycle presents a direct financial opportunity for semiconductor equipment suppliers. The iShares Semiconductor ETF provides investors with a vehicle to capture this supply-side expansion. The fund allocates nearly 21.4% of its assets to leading equipment makers, specifically Lam Research, Applied Materials, and KLA.

Each of these companies supplies the specialized machinery required to scale advanced memory production. Lam Research provides the etching and deposition tools essential for chip fabrication. Applied Materials manufactures equipment tailored for DRAM manufacturing and HBM packaging. Meanwhile, KLA supplies the inspection tools that chipmakers rely on to detect defects and improve their overall production yields.

In addition to this heavy equipment weighting, the fund provides exposure to the companies producing memory and driving its demand, including Micron. The iShares Semiconductor ETF holds a total of 30 stocks and charges a 0.34% expense ratio.