Aletheia Capital Raises Micron Target Price to $1,600
Taylor Wilson
Independent research firm Aletheia Capital raised its Micron (MU) target price from $650 to $1,600, arguing that memory's share of AI hardware value will jump from ~40% to over 70% — making storage the single most important component in an AI system.
Why more than double the target price?
Aletheia lifted Micron's target from $650 to $1,600 — a 140%+ increase, still implying roughly 47% upside from the current share price.
Micron's stock surged 10% overnight after the report dropped.
This means → the call is not a tweak; it is a wholesale re-rating of how Micron should be valued.
Why change the valuation framework entirely?
Aletheia abandoned its prior approach — pricing Micron off historical peak price-to-book (P/B) — and switched to a 10× forward P/E on FY2027 earnings.
In plain terms = memory used to be treated as a cyclical commodity — buy low, sell at the historical peak. Aletheia now argues AI-driven demand is structural, so Micron deserves a growth-tech multiple instead.
The earnings forecast is equally aggressive: FY2027 EPS up 8.5× from current levels, then another 1.8× in FY2028 — a cumulative 15× increase.
How does memory end up as 70% of AI hardware value?
Aletheia projects memory's share of total AI hardware system value will rise from the mid-40% range in 2025 to over 70% by 2027.
This means → in today's AI servers, the GPU is the biggest cost item; by 2027, memory overtakes it — whoever controls premium memory capacity controls AI hardware pricing power.
The report cites Nvidia's Vera CPU as proof: SoC-plus-memory (SoCAAM) alone accounts for over 70% of bill-of-materials cost in H2 2026, and a fully configured Vera CPU rack could carry an ASP as high as $26 million.
How much further can HBM and server DRAM prices climb?
Aletheia expects server DRAM ASPs to rise another 30% quarter-on-quarter in Q3 2026 — far above the prior consensus of 10–15% — with a further 10–15% QoQ gain in Q4.
For HBM — high-bandwidth memory, ultra-fast storage chips purpose-built for AI-scale compute — the firm forecasts ASPs doubling year-on-year in 2027.
Two pillars support the call: relentless demand from AI training and inference + high technical barriers that keep supply from expanding quickly.
Where is the biggest risk in this forecast?
Aletheia's numbers sit well above the Street consensus — validation hinges on two variables.
Variable one: can HBM and server DRAM prices actually sustain the projected climb, or will supply catch up and cap pricing at some point?
Variable two: can Micron's share of AI memory supply expand in lockstep — if prices rise but Micron loses share, the earnings forecast falls apart just the same.
In plain terms = price and share must both deliver; miss either one and the aggressive forecast takes a haircut.
Content is for reference only, not financial advice.