HBM, DRAM, flash and hard drives — where the data an AI model trains on physically sits.
Sorted by one-month move. “Vs target” compares the current price with the mean analyst price target — a figure from analysts, not from us.
| Company | Price · todayThe latest closing price, and how far it moved from the previous close in dollars and percent. | 1 MonthPrice change since the last close on or before the same date one month ago — a calendar month, not 21 trading days. | 1 YearPrice change since the last close on or before the same date one year ago. | Vs targetHow far the price sits below the average of analysts’ published 12-month price targets. Negative means the price has already risen past that average.Analysts’ figures, not ours. Not a forecast we make or endorse. |
|---|---|---|---|---|
| SNDKSanDisk | $1791.82+$177.43 (+10.99%) | ▲ 10.2% | ▲ 1712% | ▲ 19%$1000 … $3600 |
| MUMicron | $1015.80+$38.30 (+3.92%) | ▲ 8.0% | ▲ 502% | ▲ 49%$361 … $2200 |
| NTAPNetApp | $197.79+$0.87 (+0.44%) | ▼ 3.4% | ▲ 59% | ▼ 0%$170 … $219 |
| STXSeagate | $858.79+$55.66 (+6.93%) | ▼ 5.0% | ▲ 300% | ▲ 31%$700 … $1600 |
| WDCWestern Digital | $441.36+$17.49 (+4.13%) | ▼ 11.0% | ▲ 321% | ▲ 51%$420 … $1050 |
As of close, Sep 18, 2026 · prices from Yahoo Finance
Data: daily close · Reporting, not recommendations · Not financial advice.
Memory and storage is where the data an AI model trains on physically sits. Modern accelerators pair logic with high-bandwidth memory stacked directly alongside it, because feeding the processor fast enough is as hard a problem as the computation itself.
This sector has a long history of brutal boom-and-bust cycles: capacity takes years to add, demand turns quickly, and pricing swings hard in both directions. The AI build-out has pushed it into an unusually tight stretch, but the cyclicality has not been repealed.
What moves it: memory contract pricing, capacity announcements, and qualification wins on each new accelerator generation.
The Weekly Pulse covers Memory & Storage alongside every other sector — one email, Sunday, free.
High-bandwidth memory — DRAM stacked vertically and placed next to the processor so data has a much shorter distance to travel. It is what lets an AI accelerator stay fed, and supply of it has been a recurring constraint.
Capacity is expensive and slow to build, so supply arrives in large steps while demand moves continuously. When they fall out of step, pricing overshoots in whichever direction the gap runs.
Yes, for the storage tier rather than the training tier. Training data has to live somewhere before it reaches the accelerator, and at that scale mechanical drives remain the cheapest place to keep it.