People think they missed the memory trade.
I don’t think they have even seen the middle of it.
Right before the AI boom hit $MU, $SNDK
, and $SK Hynix all cut production. The memory cycle had been brutal. Prices were collapsing. The rational move was to slow output and wait for demand to recover.
Then ChatGPT happened. Then agentic AI happened. Then every hyperscaler on earth started building data centers at a scale nobody had ever planned for.
Memory demand exploded overnight and the companies that make it had just deliberately slowed their production lines.
You cannot turn a semiconductor fab back on like a light switch. These are multi-year buildouts. The fabs being built today do not reach full production until 2028 to 2030.
That is the structural story. A demand curve that compounds every year with AI and a supply response that is years behind it.
Now look at the valuations.
$MU trades at roughly 10x forward earnings despite 196% revenue growth and 75% gross margins.
$SNDK is still cheap relative to its locked in revenue commitments.
$SK Hynix is printing $26 billion in operating profit per quarter on an HBM monopoly and still trades at a discount to US peers.
The forward multiples on these companies do not reflect a shortage that runs through 2029.
$DRAM is the cleanest way to own all of it. One ticker. Every single company within wins from the same structural imbalance.
The shortage created the trade. The valuations keep it going.
@PaulKeitzke Semiconductors are undeniably critical to global growth. Your price targets are quite optimistic, but the secular demand story (AI + electrification + computing) remains very strong. Long-term constructive
10 stocks. 1 decade. Millionaire potential.
2036 target:
$RGTI — $20 Must buy — Full-stack quantum computing leader, enterprise cloud deployments + gov contracts accelerating
$QBTS — $23 Must buy — Quantum annealing systems already in production, real-world optimization deals with AI & logistics
$FN — $670 Must buy — Optical assembly kingpin powering AI data center high-speed interconnects
$MOD — $254 Must buy — Liquid cooling pioneer with multi-year hyperscaler order books exploding through 2028+
$NVT — $162 Must buy — Critical power management & cooling infrastructure for next-gen data centers
$CAMT — $158 Must buy — Semiconductor inspection equipment — every advanced AI chip packaging run needs it
$APLD — $46 Must buy — AI-native data center developer & operator with major hyperscaler hosting deals
$SMR — $11 Must buy — Small modular nuclear reactors purpose-built for always-on AI power demand
$OKLO — $64 Must buy — Advanced nuclear energy for data centers — the ultimate always-available power play
$PL — $41 Must buy — Daily Earth-observation satellite constellation feeding massive AI training datasets
AI Quantum | Satellite Data | Optical Infra | Data Center Cooling | Power Management | Semi Packaging | AI Data Centers | Nuclear Energy | Earth Observation
Which 2-3 are you loading up on for the next decade? Drop your picks below 👇
NFA