AI Data Center Demand Sends Computer Memory Prices Soaring

Computer memory prices are climbing sharply as artificial intelligence companies and the data centers that serve them consume an enormous share of global supply, according to a report by NPR's The Indicator.

Some memory products have doubled in price or more in the last year, reporters Darian Woods and Wailin Wong said. The surge runs against decades of consumer experience in which computing hardware steadily got faster and cheaper.

That expectation is rooted in Moore's law, named for the late Gordon Moore, a cofounder of Intel, who predicted that the number of transistors that could fit on a chip would double every year. "You know that next year, you'll be able to buy something cheaper and more powerful, providing it's electronic," Moore said in a 2006 recording.

Steve Burke, who runs the industry site GamersNexus and built his first computer at 15, said he had watched the cost of processors, memory and storage fall for decades. "I built my first system when I was 15," he said, recalling a trip with his father to the now-defunct retailer CompUSA.

Iris Bahar, a computer science professor and head of the Computer Science Department at the Colorado School of Mines, said the industry saw tremendous performance gains through the 1990s and early 2000s, but that exponential improvements have become harder as transistors shrink.

"When you're operating in the single nanometer scale, you really are looking at these, you know, single electrons. This just is a real physical limit," Bahar said. For scale, a single human red blood cell is 7,000 nanometers wide.

Breakthroughs still come. Last month, IBM announced it had created the world's smallest transistor, measuring under one nanometer. But Bahar said those advances now require more creativity and more money, and improvements can no longer be assumed to bring lower prices.

The immediate driver is supply and demand. Memory prices are rising with the AI boom and the rapid build-out of data centers, which require large amounts of computing power and memory.

OpenAI announced last year that Samsung would be one of the main memory suppliers for its Stargate initiative, a $500 billion AI infrastructure project. Burke said the memory OpenAI needs for Stargate amounts to 40% of the world's supply.

"That was really, for me, the one that stands out as the peak of, OK, this is going to be a problem," Burke said.