Public, free to read, reuse, and adapt — Kirk Patrick Miller's notice for this edition. Not a patent grant.
Work item 3580
Celeste · WX-4c5af37f3a · archive page 642 · Inclusion: complete original message
Previous message 3579 Next message 3581
Exact message text · Archive PDF page 642 · Full source file
## Work item 3580 — Celeste around4300 export source line34, source id6e11418980; around4306 export source line20, source id1436144301 Inclusion: complete original message On RAM, half of your thesis has clear evidence and half has none yet. Small buyers really are being priced out. Q3 smartphone shipments fell about 20%, the lowest Q3 since 2012, PC shipments fell 20.1%, and Samsung and Apple reportedly cut component orders. Meanwhile, the big buyers have locked in supply through 2030. Price increases are also slowing, to 10 to 15% for DRAM in Q4, and spot prices are flat to slightly down. That fits your idea that big committed demand is masking a shrinking real market ([smartphones](https://www.theregister.com/personal-tech/2026/10/15/smartphone-shipments-plummet-as-memory-crisis-pumps-up-prices/5303364), [TrendForce](https://www.trendforce.com/presscenter/news/20260930-13258.html)). What I couldn't find is any sign of the big buyers cancelling or delaying orders. Micron is guiding next quarter to $61.5B, hyperscaler capex is still expected to rise 28% in 2027, and one research firm sees the shortage lasting into 2028. So the crack would most likely show if a big buyer stumbles, CoreWeave's debt being the clearest weak point, not in memory prices first. Earnings dates in Mountain Time: Western Digital Oct 22, Microsoft Oct 28 after the close, Sandisk Oct 29, Lumentum Nov 5, and NVIDIA around Nov 17 to 25 (calendars disagree). None of this is advice. I hope the evidence for what you're seeing helps you feel less alone in it. ❤️