Home / Technology / AI Demand Could Extend Memory Shortage for a Decade

AI Demand Could Extend Memory Shortage for a Decade

Summary

  • Local AI use risks pushing memory supply gap for up to ten years.
  • Consumer demand for AI on PCs may outpace new memory production.
  • Memory shortage may force PCs to 256GB storage, smartphones to 64GB.
AI Demand Could Extend Memory Shortage for a Decade

The increasing interest in running Artificial Intelligence (AI) programs on local hardware is poised to prolong the global memory chip shortage for several years, potentially up to a decade, predicts Khein Seng Pua, CEO of Phison. This trend adds significant pressure to an already constrained market, which is also grappling with demand from AI data centers and a scarcity of foundational wafers, as noted by SK Hynix.

The popularity of open-source AI agents like OpenClaw, designed to run on personal computers, is a key factor. Pua anticipates a future where users increasingly opt for on-premise AI solutions, driving demand for higher memory configurations on consumer devices. This surge in consumer demand for AI inference on PCs is expected to occur within the next two years.

In the interim, the memory shortage is compelling manufacturers to adopt lower storage capacities. Reports suggest that by the third quarter of this year, many PCs may ship with only 256GB of storage, a sharp decline from current 1TB configurations. This situation is further exacerbated by the anticipated release of new data center chips that will consume more SSDs.

Phison's CEO estimates that even with new manufacturing facilities coming online around 2028, the projected 40% increase in production will be insufficient to meet the burgeoning demand, which is expected to be twice the new capacity. He believes the industry will require at least three cycles of investment to balance supply and demand, a process that could span up to ten years.

Phison is actively developing memory systems capable of running AI on lower specifications and educating users about on-premise AI. However, current optimal configurations of 16GB RAM with 1TB SSD are considered merely 'good-to-use' rather than providing the best performance. The company aims to see these specifications increase to 64GB RAM and 2TB or 4TB SSDs as DRAM output improves in the coming years, presenting a significant revenue opportunity for memory suppliers if consumer devices adopt higher memory standards.

Disclaimer: This story has been auto-aggregated and auto-summarised by a computer program. This story has not been edited or created by the Feedzop team.

Read more news on

Property Code: 5571