Layer 4 · AXTI — AXT Inc
Thesis
Indium-phosphide substrate supplier — the upstream wafer that every InP laser diode (every 100G+ datacom transceiver, every 1.6T module of the next cycle) is fabricated on. Q1 26 printed a +80% EPS surprise on +38.7% YoY revenue and the gross-margin inflection (29.9% vs -6.1% YoY) is the kind of operating leverage that confirms a structural tightness. But the stock has rocketed from a 52-week low of $1.38 to an ATH of $134 — a 100x move — and entering at $124 with that history is hazardous. Tier B because the bottleneck is real but the entry is not.
What it does + financial print
AXT is a Beijing-listed-and-headquartered (US ADR) compound-semiconductor wafer supplier — they grow indium-phosphide (InP), gallium-arsenide (GaAs) and germanium boules and slice them into the substrates that III-V foundries pattern. InP is the relevant AI bottleneck: it is the substrate of choice for the high-speed laser diodes that drive 800G, 1.6T and beyond datacom optics. The InP supply chain is genuinely tight — there are only three credible suppliers globally (AXT, Sumitomo Electric, JX Nippon), Chinese InP capacity has been export-controlled, and the AI datacom demand inflection has stripped the safety-stock buffer out of the chain.
Q1 26 (April 30) revenue $26.9M (+38.7% YoY), EPS -$0.01 vs -$0.05 (+80% surprise). Non-GAAP gross margin 29.9% vs -6.1% YoY — that's a 36-percentage-point swing in twelve months. Guide constructive: InP capacity doubling 2026. Wedbush raised PT to $93 from $80. Stock at $123.78 on May 16 — above the analyst PT after the May 13 push to an ATH $134.
Bull case
InP demand growth runs ahead of capacity adds through 2027 because the 1.6T and 3.2T datacom optical roadmap consumes 2-3x the InP-substrate area per port versus 400G/800G. AXT doubles InP capacity in 2026 and runs sold-out; gross margin steady-state moves to 35-40% on the AI mix; EPS turns durably positive in 2H 2026 with the $1+ annualised run-rate that anchors a $200+ price target. China export-control workarounds (Tianjin facility, foreign-government qualifications) prove to be the small risks the bulls said they were.
Gap / bear case
This is a 100x stock — $1.38 to $134 in under two years. The historical comp for what happens to a 100x compound-semi name when the cycle hesitates is something like Cree in 1999-2001 (down 80% from peak). The valuation at $124 prices a multi-year ramp at near-perfect execution. China geopolitical risk is real and material — AXT's Beijing operations sit in the export-control crosshairs every time US-China optical-component licensing is renegotiated. Liquidity is thin; the move down can be as violent as the move up.
Trigger to upgrade / downgrade
Upgrade to Tier A on a pullback to $80-90 with the bottleneck thesis intact and Q2/Q3 confirming the gross-margin inflection. Downgrade on loss of $90 close + China export-control negative headline.
The trade
- Entry zone: $80-95 on pullback (would be Bucket A territory)
- Stop: $75 close (cuts 38% from current — sized for the volatility)
- Position size: 0.5% NLV at current; 1% if it pulls into zone
- Catalyst date: Q2 print July 30 2026; InP capacity ramp disclosure
- Conviction: 6/10