Skip to content
theregister·

🧠DRAM Revenue Jumps 369% as AI Demand Drains Memory Supply

Your next laptop just got more expensive

TL;DR

DRAM revenue hit nearly $200 billion in Q3 2026, up 369% year over year, while bit shipments barely moved. Suppliers are setting records and buyers are fighting for supply. Expect pricier PCs and phones well into the decade.

DRAM revenue jumped 369 percent year over year in Q3 2026, and it's not because anyone shipped more memory. Bit shipments rose only about 4 percent. The money came from price, and the price came from AI. Buyers now compete for supply instead of haggling, and suppliers are on track to post another earnings record next quarter. If you're buying hardware or budgeting cloud spend, this hits you directly. Laptop and desktop prices rose by double-digit percentages across Europe this summer, and PC shipments fell because vendors couldn't get parts and buyers balked at the bill. Smartphone shipments dropped by at least a fifth as memory costs pushed budget models out of the market. Your fleet refresh just got harder to justify. The numbers: DRAM revenue reached nearly $200 billion, close to five times its level a year ago. Average DRAM selling prices hit $1.90 per gigabit. Samsung led with a 38 percent revenue share, while SK hynix recovered some ground with stronger sequential growth. Micron grew more slowly than rivals but its broader business held up. Watch the long-term agreements. Major new fabs won't start production until 2028, and LTAs are propping up prices, so shortages could persist until at least 2030. Short-term results will differ by supplier, but they should all trend up together.

DRAM Revenue Jumps 369% as AI Demand Drains Memory Supply — theregister

Key Points

1

Average DRAM selling prices reached $1.90 per gigabit, a level that is reshaping hardware budgets across the industry.

Keep reading — it's free

Enter your email to keep reading — plus the best of AI & tech, daily. Free, forever.

Also get
or

Already a member? Sign in

DRAMmemory pricesAI infrastructuresemiconductorshardware costs

Comments

Subscribe to join the conversation...

Be the first to comment