Via nytimes.com
Memory chip prices have quadrupled as Big Tech's spending spree collides with finite semiconductor supply, raising uncomfortable questions about what's actually driving the AI gold rush.
Memory chip prices have surged roughly fourfold, driven by data-center operators hoovering up every available unit of high-bandwidth DRAM and HBM they can find. Hyperscalers, the Microsofts and Googles of the world, have redirected production capacity toward high-margin AI hardware, starving other industries of the chips they need to build everything from cars to consumer electronics.
Micron Technology reported quarterly profits nearly 15 times higher year-over-year, a figure that reflects just how dramatically AI-driven memory demand has reshaped the semiconductor landscape. Overall semiconductor revenue is forecast to exceed $1.3 trillion in 2026. Big Tech companies are expected to pour over $650 billion into AI capital expenditures in 2026, with a significant chunk flowing directly into memory chips.
Non-AI industries have responded by ramping up lobbying efforts in Washington, pressing for relief from memory prices that are making their products more expensive to manufacture. The automotive industry finds itself competing for semiconductor allocation against companies building AI data centers with seemingly unlimited budgets.
Memory chip shortages are expected to persist through late 2027 and potentially stretch into the 2028-2030 window, a timeline that could delay AI projects themselves.
Skeptics point out that supply crunches have a way of inflating perceived demand. When companies can’t get enough chips, they over-order. When they over-order, it looks like demand is even stronger than it actually is. This cycle, sometimes called the “bullwhip effect” in supply chain economics, has burned semiconductor investors before. The last major chip shortage, during and after the pandemic, ended with an inventory glut that hammered chipmaker valuations.
Investors watching this space should pay close attention to memory chip inventory levels, hyperscaler capex guidance, and any signs that AI-related orders are being driven by fear of future shortages rather than current deployment needs.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our