DRAM prices have surged 600% over the past year. DDR5 modules that cost $149 last spring now sell for $239. A PlayStation 6 that was supposed to launch in 2027 is now delayed to 2028 or 2029. Your next laptop will cost 15-20% more, if you can get it at all.
The cause: AI data centers are consuming 70% of global memory chip production, and the world’s three major manufacturers have made a calculated decision that your iPhone is less profitable than Nvidia’s next GPU.
The Technical Problem
High-Bandwidth Memory (HBM) - the specialized memory stacked onto AI accelerators - consumes roughly three times the wafer capacity of DDR5 per gigabyte. Every silicon wafer pushed into HBM production is a wafer denied to your smartphone, laptop, or gaming console.
This isn’t a temporary allocation shift. Samsung, SK Hynix, and Micron control virtually all DRAM production, and they’re permanently reorganizing their factories around AI customers.
According to Fortune, HBM demand is projected to increase 70% year-over-year in 2026, and will consume 23% of total DRAM wafer output - up from 19% last year. That doesn’t sound catastrophic until you realize HBM’s 3:1 wafer consumption ratio means it’s actually displacing the equivalent of far more conventional memory.
SK Hynix reported its HBM, DRAM, and NAND capacity is “essentially sold out” through 2026.
Micron Abandons Consumers Entirely
In December, Micron made the industry dynamic explicit: they’re exiting the consumer memory market entirely.
The company will wind down its Crucial consumer brand by February 2026, redirecting all capacity to enterprise and AI customers. Micron CEO Sanjay Mehrotra told analysts they can currently “meet about 50% to two-thirds of demand from several key customers” - and those key customers aren’t PC builders or phone manufacturers.
The math is straightforward. Enterprise HBM contracts deliver higher average selling prices, multi-year commitments, and predictable demand. Consumer RAM is comparatively low-margin with volatile demand. Micron’s fiscal 2025 revenue hit a record $37.38 billion, with data center and AI applications accounting for 56% of total revenue - up from roughly 30% two years ago.
Samsung and SK Hynix haven’t announced similar consumer exits, but they’re following the same incentives. As Samsung President Wonjin Lee warned: “In 2026, there’s going to be issues around semiconductor supplies, and it’s going to affect everyone, not just Samsung.”
The Price Reality
The numbers are stark:
- DRAM costs surged 75% in a single month (December 2025 to January 2026)
- 600% increase year-over-year according to Counterpoint Research
- 32GB DDR5 modules: Rose from $149 to $239 (60% increase)
- Contract DDR5 pricing: Surged over 100% to $19.50 per unit versus roughly $7 earlier in 2025
- Custom PC prices: Up approximately $1,500 per unit
Gartner projects an additional 47% DRAM price increase through 2026. DDR5 64GB RDIMM modules could cost twice as much by year-end compared to early 2025.
Analyst Mark Li described memory prices as going “parabolic.”
Who’s Getting Hit
The impact extends across consumer electronics:
Smartphones: DRAM could reach 30% of a low-end phone’s bill of materials, up from 10%. Apple’s Tim Cook has acknowledged the company is “currently constrained” on memory and expects “impact to Q2 gross margin.”
Gaming consoles: Sony has pushed the PlayStation 6 from a planned 2027 launch to 2028-2029. Nintendo is considering price increases for the Switch 2.
PCs: Major OEMs including Lenovo, Dell, HP, Acer, and ASUS have warned of 15-20% price hikes. IDC forecasts the PC market could contract 4.9% to 8.9% in 2026.
Automobiles: Tesla, Qualcomm, and automotive suppliers have flagged memory constraints affecting production. Elon Musk told analysts: “We’ve got two choices: hit the chip wall or make a fab” - and noted that “currently there are no advanced memory fabs at scale in the United States.”
More than a dozen major companies - including Tesla, Apple, Cisco, Qualcomm, Dell Technologies, Lenovo, Samsung, Sony, Nintendo, Xiaomi, and Oppo - have signaled production constraints.
No Quick Fix
The shortage will persist through at least late 2027, and potentially into early 2028.
Building new semiconductor fabrication plants takes years. Samsung’s planned Pyeongtaek facility won’t reach mass production until 2028. Equipment spending hits $135 billion in 2026, but there’s insufficient cleanroom capacity to deploy it rapidly.
The fundamental issue isn’t just capacity - it’s prioritization. AI customers pay more and sign longer contracts. Until that changes, memory manufacturers have no incentive to reallocate to consumer products.
As CCS Insight analysts noted, new supply won’t “materially address constraints until well into 2027 and potentially into early 2028.”
What This Means
The AI boom isn’t abstract. It’s making your next phone more expensive, delaying gaming consoles, and forcing electronics manufacturers to compete for memory against companies spending hundreds of billions on data centers.
This is the hidden cost of the AI infrastructure buildout. Microsoft, Google, Meta, and Amazon are locking up memory capacity years in advance. The manufacturing system is reorganizing around their needs. Everyone else - including you - pays the difference.
Mid-market firms face particular pressure. As one analyst noted, they lack the procurement leverage of hyperscale cloud providers but face the same price increases.
The Bottom Line
The AI memory crisis demonstrates how quickly infrastructure bottlenecks can reshape entire markets. When three companies control an essential input and one category of customer offers dramatically higher margins, the rest of the market doesn’t get a vote.
Expect higher prices, longer waits, and compromised specs on consumer electronics through at least 2027. The AI boom has a bill, and you’re paying it whether you use AI or not.