$MU $SKHY
I'd rather own Micron than SK Hynix, and it's not close.
"Build memory in America, or pay a 100% tariff." — US Commerce Secretary, said at Micron's groundbreaking.
$200B in US fabs vs $3.9B of packaging that doesn't run until 2028.
Long MU. Sell SK.
$MU $SKHY
SK Hynix missed revenue by 5.6%. SKHY ADR: all-time low, below its $149 IPO price.
Everyone's selling memory on China fear. But Samsung and SK run fabs in China. Micron doesn't — it's the one lobbying to keep CXMT out.
Wrong stock to sell.
Long MU. NFA.
$MU
Kimi K3 shock: the winner is clearly MEMORY.
2.8T params = 1.4TB of weights that must sit in memory. Compute got cheap; memory didn't.
Jul 21 rebound:
🔥 MU +12% / SNDK +14% / SKHY +14%
😴 NVDA +2%
Jevons paradox again — but the rent moved from FLOPs to HBM.
🚨BREAKING: SpaceX just ordered about 1 MILLION new GB300 GPUs pushing Colossus compute towards 2GW
Foxconn named as manufacturing partner
>13,000+ racks = 936,000 GPUs
>$4M per rack = $52 billion order
>deliveries late 2026
SpaceXAI is playing to WIN.
$MU $SKHY Memory stocks trade at single-digit P/Es because the market assumes peak earnings always vanish. Micron just broke that assumption: multi-year fixed-price contracts, $18B prepaid by customers. SK Hynix still reprices every year. This isn't an earnings story
$NBIS just signed a 10-year, $2.6B / 328MW deal with Bloom Energy.
Meta's $27B + MSFT's $19.4B contracts don't justify 328MW alone. Power follows demand — not hope.
Nebius also raised FY26 capex guidance by $4.5B on the same day.
H200 hit $6.40/hr, up 29% overnight, and is now above B200
The GPU shortage is getting worse
Massive win for neoclouds that have already purchased GPUs
Source: Ornn
Not just another AI-themed stock — this is a geometric match to the hyperscaler-launch pattern.
But: $NBIS only has 1 post-inflection quarter. The real test is Q2 2026 (Aug).
Hyperscalers don't get born. They launch.
We may be watching one in real time.
Hyperscalers aren't born. They launch.
The moment of launch = their first major consensus beat. After it: consecutive beats → exponential mcap growth.
Aligned 6 hyperscalers at t=0 (their inflection). The geometry is undeniable.
And $NBIS is standing there right now. 👇