Soaring RAM prices are bad news for the entire tech industry—but Apple stands to lose the most.
Memory prices have surged 100% to 400% in just six months. A 32 GB RAM kit that cost $95 last summer now sells for as much as $400. Some U.S. retailers have even removed price tags, quoting figures only at checkout—an unusual move that underscores the volatility.
Why it matters
Samsung and SK Hynix have reportedly committed around 40% of global RAM production to OpenAI’s Stargate data center project. Together with Micron, the three companies control 93% of the market and are prioritizing server-grade memory over consumer devices.
TrendForce now expects entry-level smartphones to revert to 4 GB of RAM by 2026, while budget laptops stagnate at 8 GB. For the first time in decades, hardware specifications are moving backward.
The AI paradox
AI is driving the memory shortage—but that same shortage undermines the ability to run AI locally. Data centers absorb most available RAM to train massive models, leaving consumers unable to run those models on their own devices. The result: more AI in theory, less in practice.
Apple’s unique exposure
Apple is the most vulnerable player. While Meta, Google, and Microsoft rely heavily on cloud-based AI, Apple has spent two years betting on on-device intelligence as its key differentiator—privacy-first, fast, and independent of servers.
Apple Intelligence depends on ample local memory and compute power. iPhones have steadily increased RAM to support it, and Macs with Apple Silicon have normalized 16 GB as a baseline, ending years of 8 GB stagnation.
An impossible choice
Apple’s financial muscle and supply contracts help secure memory—but they don’t solve the core dilemma:
- Raise prices to preserve specs, risking consumer pushback.
- Cut RAM to control costs, undermining the very advantage Apple has been selling.
Neither option is attractive.
Reading between the lines
Other manufacturers can lower specs with minimal impact—Samsung can ship a mid-range phone with 6 GB of RAM and lean on Google’s cloud AI. Apple cannot. Its architecture requires powerful hardware in users’ hands, and that hardware is becoming dramatically more expensive to build.
Private Cloud Compute may ease some pressure, but it doesn’t change the core promise of local AI.
The bigger picture
Apple Intelligence may end up far more expensive than planned, not because AI software is costly, but because the raw materials to run it are now scarce.
Apple is well positioned to weather the storm, as it did during the pandemic chip crisis. Yet strategically, it also has the most to lose. The company chose a different path just as that path became prohibitively expensive.
Cloud AI scales with rentable servers.
Local AI scales only if every user owns powerful hardware—and that hardware just got a lot pricier.
Bottom line: The era of cheap memory is over. And one of its most significant consequences may be the economic viability of Apple’s boldest AI bet.
