NASA’s new dark energy space telescope can also detect killer asteroids
technologyreview.com·3h ago
Data Science
Simon Willison's Blog·May 22·5 min read
The memory shortage is causing a repricing of consumer electronics
TL;DR
Memory manufacturers face a capacity constraint: only three major companies exist with fixed wafer (silicon disk) processing capacity split between DDR (desktop RAM), LPDDR (mobile RAM), and HBM (GPU memory). AI datacenter demand has pushed HBM allocation from 2% to expected 20% by 2026, with HBM consuming 3x more wafer capacity per gigabyte than alternatives.
✦ Why It Matters
Engineers designing consumer products should anticipate sustained memory cost increases and plan component sourcing strategies accordingly.
Key Takeaways
- Monitor memory pricing trends to anticipate cost changes in products.
- Evaluate alternative memory solutions to mitigate supply constraints.
- Consider targeting markets less affected by rising memory costs.
- Stay informed about AI developments impacting memory demand.
Full Summary
Memory manufacturers are facing a critical shortage due to limited wafer processing capacity, which is divided among different types of memory: DDR for desktops, LPDDR for mobile devices, and HBM for GPUs. Recently, the allocation for HBM has increased from 2% to an anticipated 20% by the end of 2026, driven by the rapid growth of AI data centers.
Notably, HBM requires over three times the wafer capacity compared to DDR or LPDDR. As a result, memory companies are prioritizing HBM production to maximize profits, leading to constrained supply for consumer RAM.
This situation is already affecting the sub-$100 smartphone market, which is crucial for regions like Africa and South Asia, potentially making affordable technology less accessible.