[REVISION HISTORY]

AI demand drives surge in memory and semiconductor prices

Updated 25 times since CLSTR started tracking revisions of this situation.

What changed

2026-09-30 08:58 UTC → 2026-10-03 01:45 UTC · added ~~removed~~

The global shortage of DRAM and NAND memory components is ~~intensifying, driven by~~ intensifying as production capacity ~~being~~ is diverted to AI servers. This supply crunch has led Samsung to reportedly raise DRAM and NAND prices by 30% to 40%. Valve Corporation has ~~commented on~~ noted the worsening conditions, stating, ‘to be honest, there is no light at the end of the tunnel, in fact, things are getting even worse.’ As ~~the~~ AI ~~boom continues, TrendForce projects that~~ infrastructure expands, High Bandwidth Memory (HBM) is consuming a larger share of DRAM wafer capacity. Samsung projects that HBM will account for nearly 30% of global DRAM wafer capacity by 2027, up from approximately 20% currently. This shift is expected to create a ‘crowding out effect,’ potentially limiting supply and increasing costs for traditional memory used in PCs and smartphones. Samsung has proposed pricing for its 2027 HBM4 generation at a mid-$4 per gigabit, more than triple the current HBM3E price of roughly $1.50 per gigabit. TrendForce projects that HBM blended average selling prices will see a massive surge of 121% in 2027. To manage rising system costs, GPU and ASIC manufacturers may shift toward 8-Hi HBM configurations rather than 12-Hi, though 8-Hi products are expected to be 10-20% more expensive per Gb in 2027 due to base die costs being spread over fewer DRAM dies. Raw material costs are also rising, with 300mm silicon wafer prices projected to increase by ~~10% to over 40% in~~ 2027. By the fourth quarter of 2026, general DRAM contract prices are projected to increase by 10% to 15%, while NAND Flash contract prices are expected to rise by 15% to 20%. Enterprise SSDs are seeing the most significant price acceleration, ~~driven~~ with bit demand expected to grow by over 80% annually in 2026 as cloud ~~service~~ providers ~~expanding~~ expand AI inference and Agentic AI ~~capabilities, which require increased real-time data retrieval.~~ capabilities. Despite these ~~price~~ surges, ~~some~~ investor caution has emerged. Valuations for major manufacturers ~~like Micron and SK Hynix~~ have ~~seen reductions in~~ declined; for instance, Micron’s trading ~~multiples,~~ multiple has dropped from twelve to seven times expected earnings, suggesting skepticism regarding the long-term sustainability of the current AI-driven market cycle.

Versions

Only revisions since CLSTR began indexing content versions appear here. Select a version to see what changed compared to the one before it.