Memory prices are set to surge as AI hardware demand intensifies, according to a new forecast from Aletheia Capital. The firm predicts that HBM average selling prices could double annually in 2027. Server DRAM ASPs are expected to jump 30% in Q3 2026 and another 10-15% in Q4 2026. These increases reflect memory's growing role as the most critical component in AI systems.
Memory content value in AI hardware to exceed 70%
The forecast covers high-bandwidth memory (HBM) and server DRAM, key components in AI hardware. Aletheia Capital, an investment research firm, issued the report. The projections highlight Micron as a major beneficiary. The firm expects Micron's earnings per share to jump 8.5x in calendar year 2027, followed by a further 1.8x expansion in 2028.
Memory devices' combined content value in AI hardware could exceed 70% by 2027, up from the mid-40% range in 2025. This shift makes memory a dominant cost driver in AI systems. Aletheia notes that the NVIDIA Vera CPU rack could cost up to $26 million due to high memory prices.
The pricing outlook implies significant revenue growth for memory makers. Aletheia forecasts roughly 15x cumulative EPS growth for Micron and $350-400 billion of free cash flow generation in fiscal years 2026-2028. The firm's analysis underscores memory's transition from a commodity to a strategic component in AI infrastructure.
Discussion
0 comments
Log in to join the thread with a thoughtful take, question, or correction.