Supercycle Themes

8 structural investment themes across AI photonics, memory, grid, and quantum. Each represents a multi-year supply chain inflection point.

Volume Ramp

CPO Lasers

The chokepoint of next-gen AI photonics

Goldman GS projects CPO at $91B. SIVE at $1.5B is the laser chokepoint.

TAM$15B$154Bby 2030
$SIVE$AAOIMSSCorps$MRVL+3
Qualifying

Glass Core Substrates

Next-gen advanced packaging — 2027 mass production

LPKF LIDE monopoly. 80%+ of semis chose LPKF. Volume H2 2026 not in any forecasts.

TAM$2B$20Bby 2030
$LPK
Mass Production

HBM4 Memory Capex

Next memory supercycle — SK Hynix, Samsung, Micron racing

Three memory makers all increasing HBM4 capex simultaneously. Towa monopoly = unavoidable inflows.

$TOWAAuros$SNDK$EWY
Active Bottleneck

InP Substrate Supply Chain

Active chokepoint — Western photonics depends on this

AXT controls raw materials + pBN crucibles + InP substrates. Not just 'InP company.'

$AXTI$IQE
Active Bottleneck

Silicon Photonics

The substrate layer beneath all CPO architectures

SOI: 100% of next-gen AI DCs use Photonics-SOI. Monopoly substrate layer.

$SOI$SIVE
Active Bottleneck

Transformer & Grid Infrastructure

National emergency declared — 2–5 year bottleneck

Defense Production Act invoked. 2-5 year lead times. Federal funding incoming.

$HPS.A$AMSC
Mass Production

Neoclouds

Quality Matters: NBIS vs. IREN

NBIS: favorite neocloud, sum-of-parts discount. IREN: $6B ATM dilution machine.

$NBIS$RDDT$HIMS$ARM
Early R&D

Quantum & MBE

Early R&D — 3–4 years, but front-running now

Riber: MBE duopoly, MSFT quantum supplier. Apollo could buy entire float as rounding error.

$ALRIB$YSS