AI memory crunch pushes MacBook Air into shortage
Apple's highest-volume laptop is now short of supply as DRAM makers divert wafers to AI data centres, with custom orders slipping into September.
The memory shortage created by AI data-centre buildouts has reached Apple's best-selling computer. Bloomberg's Mark Gurman reported on Sunday that MacBook Air supply has tightened globally, with configured orders on Apple's own store slipping into late August or September in North America, Europe and India. Earlier this year the squeeze was confined to Apple's lower-volume desktops, the Mac mini and Mac Studio; the Air is the company's highest-volume Mac and its primary education-market product.
Apple's response, according to the report, has been a mix of pricing and marketing adjustments rather than a fix. The company has raised prices on affected models, begun sourcing memory from Chinese suppliers to widen its vendor base, and delayed its back-to-school promotion — originally planned for June — while steering campaign emphasis toward base-model MacBook Pro units that are easier to keep in stock. Apple has also started attaching availability disclaimers to Air marketing materials.
The underlying cause is allocation, not capacity loss. DRAM fabricators have been shifting wafer starts toward high-bandwidth memory, the stacked DRAM that sits beside AI accelerators and commands far higher margins than the LPDDR modules that go into laptops. Consumer PC lines now effectively bid against hyperscaler orders for the same fab output. DDR5 and NAND contract prices have climbed sharply since January, and most announced greenfield memory capacity is not scheduled to produce volume before 2027.
Why it matters
This is the clearest consumer-visible evidence yet that AI infrastructure spending has begun to price other industries out of the semiconductor supply chain. Apple has among the strongest procurement positions in electronics — long-term contracts, prepayments, enormous volume — and it still cannot hold stock on its flagship laptop. That sets a floor expectation for everyone below it: PC OEMs, smartphone makers, console and automotive buyers face worse terms and longer waits. For AI economics specifically, memory is now a hard constraint on both sides of the ledger, raising the bill of materials for on-device AI hardware at the same time it inflates the cost of the accelerators the cloud runs on.