Memory prices up sixfold. Companies and consumers pay the AI bill

Digital inflation has a specific source, and it is not oil. An analysis by Corriere Comunicazioni, citing the WEF, notes that memory prices have risen roughly sixfold in the past year, according to Morgan Stanley. That shift has reset the cost calculation for hardware within just a few quarters.

Gartner's April forecast projected average DRAM price increases of 125 percent and NAND of 234 percent in 2026. Analysts see a possible turning point only in the second half of 2027. The scale is unprecedented because memory is a component in every device, from servers with accelerators to the phone in your pocket.

The impact is already visible in demand forecasts. Gartner expects PC shipments to drop 10.4 percent and smartphone shipments 8.4 percent in 2026. Institutional clients pay higher prices for the same hardware, while consumers delay purchases.

A second factor is the concentration of value in the supply chain. SIA and Deloitte report that processors, memory, networking chips, storage, and power chips account for about 95 percent of the hardware value in an AI compute rack. This explains why the entire investment is so sensitive to the prices of a few component categories.

Spending scale is striking in absolute terms too. By 2028, global data center infrastructure spending is set to exceed 4 trillion dollars, with as much as 2.8 trillion potentially going to semiconductor makers. HBM memory, essential for training and inference, is now a market worth over 30 billion dollars.

For end users, the conclusion is unpleasant but simple: until the supply chain increases output, prices will not fall. Memory prices have become a macroeconomic variable. They enter corporate inflation baskets, cloud bills, and IT budgets for the coming year.

The global semiconductor market is expected to be worth more than 1.3 trillion dollars in 2026. Forecasts were revised upward to 1.56 trillion dollars as late as June. According to SIA and Deloitte, processors, memory, networking chips, storage media, and power chips make up roughly 95 percent of the hardware value in a single AI server rack. These are the components whose prices are rising fastest, which is why their cost increases ripple through the rest of the bill.

Consequences for users appear in sales forecasts. Gartner anticipates PC shipments falling 10.4 percent and smartphones 8.4 percent in 2026. Devices are simply getting more expensive. Higher costs for conventional memory, combined with HBM chips worth over 30 billion dollars absorbed by AI accelerators, narrow the manufacturing capacity available to the consumer market.