Skip to content
orbyd

Dossier · AXTI · Watchlist

AXTI · AXT Inc · Stock research

LOW Cyclical recovery Catalyst · semi-foundry-equipment networking-optical ai-datacenter-infrastructure

Last analysed ·

Current thesis

InP substrate supplier one layer upstream of AI optical interconnect fundamentals still accelerating on the 7/2 Coherent master supply deal ($22.29M prepay) and $100M backlog, but the tape has round-tripped >50% off the $143 ATH toward the $64.25 April secondary price. Broken structure, improving story; the ~7/31 Q2 print (first profit in years) is the binary.

Invalidation trigger

A weekly close below $64 (the April secondary at $64.25 failing as support) retraces the entire post-raise advance and opens the low-$50s; the story also breaks if the ~2026-07-31 Q2 print lands under the $34M revenue guide or the record $100M InP backlog declines QoQ.

Thesis status

Open commitment catalyst in 12dscored if the trigger above fires How this is scored →

Latest analysis and events for AXTI —

As of 2026-06-05, orbyd's latest analysis for AXT Inc (AXTI): Structure broke closed ~$89.04 (-15.99%, range $83.15-$104.55), losing the $90 base that was the prior abort line. Driver: CEO Morris Young disclosed >$22M insider sale + valuation/AI-semi-group selloff. -38% off the $143.16 ATH (5/26) in ~7 sessions. Second management top-tick after the $64.25 secondary (8.56M sh, 4/21).

Invalidation trigger: A weekly close below $64 (the April secondary at $64.25 failing as support) retraces the entire post-raise advance and opens the low-$50s; the story also breaks if the ~2026-07-31 Q2 print lands under the $34M revenue guide or the record $100M InP backlog declines QoQ.

Next dated event on file: — catalyst in 12d.

Current Thesis

AXT sells the indium-phosphide (InP) substrate that the lasers and photodetectors inside 800G/1.6T optical transceivers are grown on one layer upstream of the AI interconnect build. The fundamental story has kept accelerating through July: the 2026-07-02 AXT-Tongmei master development and supply agreement with Coherent (effective 2026-06-25) carries a $22.29M customer prepayment and funds Beijing capacity expansion through 2028 at agreed pricing, and the 2026-06-17 Casela pact adds $25.4M of contracted 2027 InP deliveries on top of the record $100M backlog disclosed on the 2026-04-30 Q1 call. The tape has gone the other way. Benzinga's market-cap stamps mark $3.89B on 2026-07-02 against $2.97B on 2026-07-17 roughly -24% in eleven sessions on an unchanged share count, implying a high-$60s/low-$70s handle versus the $143.16 ATH set 2026-05-26. That is a >50% round-trip from the high and a decisive break of the $90 shelf and the $83 June distribution low the structure was resting on. Two multi-year customer contracts and a halving stock in the same month is not a contradiction to resolve by buying weakness; it is a supply/demand problem in the shares April secondary paper, lockup expiry, and insider monetization that the fundamentals have not yet absorbed. Stand aside until a higher-low forms and holds; the ~2026-07-31 Q2 print, guided to the first profit in years, is the event that either restarts the re-rating or confirms the distribution.

Bullish and bearish views on AXT Inc

The model's bull view on AXT Inc (AXTI), in brief: Coherent master supply agreement (announced 2026-07-02, effective 2026-06-25): $22.29M prepayment, Beijing capacity expansion 2026–2028, agreed specific pricing. The bear view: The implied drawdown is the whole story. Both cases follow in full.

Bull Case

  • Coherent master supply agreement (announced 2026-07-02, effective 2026-06-25): $22.29M prepayment, Beijing capacity expansion 2026–2028, agreed specific pricing. Coherent is one of two optical vendors NVIDIA committed $2B each to on 2026-03-02 the end customer is funding AXT's capex directly.
  • Casela agreement (2026-06-17): $25.4M long-term InP wafer supply for 2027 deliveries. Two multi-year commitments inside four weeks argues the backlog is contracted, not a single-quarter order spike.
  • Record $100M InP backlog (Q1 call, 2026-04-30) versus roughly $60M in Q4 2025 visibility into the 800G→1.6T transceiver ramp rather than a modeled forecast.
  • Q1 2026 (2026-04-30): revenue $26.9M, +39% YoY and +17% QoQ; InP crossed 50% of total revenue for the first time; non-GAAP gross margin 29.9% versus 21.5% prior quarter; EPS -$0.01 against -$0.05 consensus.
  • Q2 guide (2026-04-30): revenue ≥$34M with first GAAP and non-GAAP profitability in years a guided inflection with a hard date attached.
  • Northland lifted its target to $125 from $90 (Outperform, early June) with estimates raised, and the current implied high-$60s handle sits far below that.
  • Group still bid: AXTI appeared among the day's IT gainers on 2026-07-06 alongside CRDO, ALAB, IMOS and CEVA, and again on 2026-07-14 with TSEM and SK Hynix. Compound-semi and optical names are still being repriced as a cluster.

Bear Case

  • The implied drawdown is the whole story. $3.89B market cap on 2026-07-02 to $2.97B on 2026-07-17 is about -24% in eleven sessions, on top of the -15.99% single-day break on 2026-06-05 to $89.04. Roughly half the value created off the April secondary has been given back with no confirmed base.
  • Two separate management monetizations bracket the high.
  • The China value-unlock catalyst slipped: Tongmei withdrew its Shanghai STAR Market IPO application on 2026-07-08 to pursue a Hong Kong listing. A long-promoted re-rating event is now delayed with an uncertain venue and no filed timeline.
  • Valuation still needs the ramp: roughly $3B of market value against ~$27M of quarterly revenue and a first-profitable-quarter guide. The multiple prices execution that has not printed.
  • Prepayments cut both ways. A $22.29M customer prepayment funding Beijing capacity is a commitment, but it also signals AXT could not self-fund the expansion after raising $550M three months earlier.
  • News flow has thinned to filler. Two of the last three AXTI headlines (2026-07-02, 2026-07-17) are Benzinga's automated "$1,000 invested N years ago" backtests. Narrative velocity from actual company news has slowed since the Coherent print.

Setup & Price Structure

Structure is broken, not basing. The sequence runs $64.25 secondary (2026-04-21) → $143.16 ATH (2026-05-26) → -15.99% gap-down to $89.04 (2026-06-05) → lost the $90 shelf → lost the $83 June low → an implied high-$60s/low-$70s handle by 2026-07-17. That last level puts price within a few points of the April secondary at $64.25, which is the line the entire post-raise advance was built on and the price at which underwritten paper is held. The 2026-07-14 appearance on the gainers list is a bounce inside a downtrend, not a reclaim. The name currently sits in the worst quadrant of the trap matrix: a former parabolic leader, more than 50% off its high, with an improving fundamental narrative precisely the profile that invites averaging into a falling structure. The signal to wait for is mechanical: a higher low above the prior swing, volume normalization, and a weekly close back above $90 to reclaim the broken shelf. Below $64.25 there is no visible reference until the pre-squeeze low-$50s. Any position sizing here belongs in probe territory given the realized volatility (a -16% day and a -24% eleven-session slide inside seven weeks).

Catalyst Calendar (next 30 days)

  • ~2026-07-31 (est., confirm via IR): Q2 2026 results. The binary guide was ≥$34M revenue with first GAAP and non-GAAP profitability in years. Watch the InP backlog line versus the $100M Q1 figure and any commentary on Coherent capacity timing.
  • ~2026-07-31: Q3 guidance on the same call. With Coherent prepayment capacity phasing across 2026–2028, the shape of the H2 ramp matters more than the Q2 number.
  • Ongoing through August: Form 4 activity. Further insider sales after the June disclosure would confirm the distribution read; an insider buy near the secondary price would be the first genuine contrarian datapoint.
  • Read-throughs: Coherent and Lumentum quarterly reports in the August window set the InP demand tone; IQE.L remains the cleanest real-time proxy for sector-wide compound-semi repricing versus AXTI-specific dilution overhang.

Elapsed catalysts

  • Rolling, no fixed date: Hong Kong listing filing for Tongmei following the 2026-07-08 STAR Market withdrawal. A filed application with a timeline would restore a catalyst that just went dark. _(passed 11d ago)_

What Would Change Our Mind

  • A weekly close back above $90 on expanding volume, reclaiming the shelf lost on 2026-06-05, would flip the read from stand-aside to a break-and-retest entry.
  • Q2 revenue above the $34M guide with the InP backlog printing above $100M and explicit Coherent-driven H2 capacity commentary would confirm the fundamental leg is intact and the selling was purely supply-driven.
  • A Tongmei Hong Kong listing filing with a dated timeline restores the value-unlock catalyst removed on 2026-07-08.
  • On the other side: a weekly close below $64.25 means the April secondary price failed as support and the entire post-raise advance is retraced, which would put the low-$50s in play.
  • Q2 revenue under the $34M guide, or a QoQ decline in the $100M InP backlog, breaks the accelerating-demand story regardless of what the contracts say.

Correlation Notes

  • Direct customer/peer set: COHR (contracted counterparty, 2026-07-02), LITE, AAOI, CRDO, ALAB. AXTI traded with this group on 2026-07-06 and 2026-07-14; divergence from it would isolate the move as single-name dilution rather than sector demand.
  • Upstream read-through: IQE.L (GaAs/InP epitaxy). If IQE is flat while AXTI slides, the weakness is AXT's share supply, not compound-semi ASPs.
  • Theme anchor: NVDA optical-interconnect capex and the $2B commitments to Coherent and its peer (2026-03-02) are the demand source; hyperscaler capex guides are the upstream driver of the InP backlog.
  • China exposure: Tongmei's Beijing capacity and the shifted listing venue make AXTI sensitive to US-China semiconductor export policy headlines in a way pure-US optical names are not.
  • Float mechanics: post-secondary share count plus the 2026-06-21 lockup expiry means AXTI trades with more supply elasticity than its optical peers it will underperform the cluster on down days and can overshoot on squeezes.

Notes

  • 2026-04-18: seed: Serenity/attention list
  • Q1 2026 earnings ~2026-05-05/08 confirm via IR before any pre-print sizing; reduce to probe 3 trading days out (binary risk rule).
  • Attention stacking REAL (5 Benzinga prints in 11 sessions) but price confirmation PENDING do not anticipate the breakout; buy the break-and-retest.
  • No live price context supplied operator MUST confirm 20-EMA relationship
  • relative volume
  • and 52-week pivot proximity before entry.
  • If IQE.L is dead while AXTI rips
  • the GaAs ASP narrative is likely false use IQE tape as real-time read-through confirmation.
  • Micro-cap <$150M with elevated short interest hard 1–2% per-name cap if setup upgrades to a6 squeeze on breakout.
  • Prior decision 2026-04-19 was avoid/MEDIUM
  • no real synthesis) current refresh is first real trader-grade read.
  • Squeeze leg complete. NO CHASE.
  • 2026-04-20: Q1 pre-announced $26-28M rev / -$0.03 to -$0.05 EPS 2026-05-06 print is de-risked AND neutralized as upside catalyst.
  • Re-entry blueprint: 10-30 trading days for underwriter distribution to clear; require higher-low + volume normalization + weekly close >$64.25 + bullish InP/use-of-proceeds language.
  • Pattern saved: 5 Benzinga prints in 11 sessions (04-06 to 04-16) successfully front-ran the squeeze on a thin-float micro-cap. Replicable signature for next analog.
  • IQE.L is the cleanest real-time read-through sector-wide GaAs repricing vs single-name dilution overhang.
  • Prior decisions 2026-04-19
  • 2026-04-21 both avoid. System never got long. Missed trades cost zero chasing post-secondary drift costs real money.
  • Theme membership flipped ACCELERATING → SATURATED in 2 weeks. Watch the next compound-semi name that prints the same Benzinga-stack signature.
  • LESSON: 3x avoid (4/19-5/15) on 'management top-ticked the squeeze' was WRONG the $64.25 secondary FUNDED the InP capacity doubling that drove $64→$143.16 ATH (5/26). Dilution is NOT a top when the raise funds an accelerating demand story. Don't reflexively fade secondaries in ACCELERATING themes.
  • Archetype corrected 6→2: not a retail squeeze, it's an InP picks-and-shovels substrate supplier with a real revenue inflection (InP >50% of rev, $100M backlog, Q2 guided to first profitability in years). Drop the 1%/name squeeze cap; volatility still warrants disciplined sizing.
  • Q2 2026 print ~2026-07-29 (next hard binary). Reduce to probe 3 trading days out per binary-risk rule. Guide: ≥$34M rev + first GAAP/non-GAAP profitability in years.
  • Cluster read-through: COHR, LITE (NVIDIA-backed InP laser makers AXT supplies, $2B each 3/2), AAOI, AEHR watch their tape as real-time confirmation. EUV photonics ETF flows = mechanical bid/pressure.
  • Valuation ~50-65x sales; thesis is 100% forward backlog/guide. A single InP order push-out can halve it. Stop = daily close <$90.
  • Re-entry quality gate: require higher-low above ~$100 + reclaim of $115-120 on >1.5x volume before upgrading MEDIUM→HIGH. We already missed the first leg don't compound it by chasing chaotic $100-116 intraday chop.
  • 20-day SMA quoted at ~$60 by data vendors is STALE/lagging after the $64→$143→$103 move; true rising 20-EMA est. ~$95-110, meaning ~$105 is sitting on support.
  • 2026-06-05: Structure broke closed ~$89.04 (-15.99%, range $83.15-$104.55), losing the $90 base that was the prior abort line. Driver: CEO Morris Young disclosed >$22M insider sale + valuation/AI-semi-group selloff. -38% off the $143.16 ATH (5/26) in ~7 sessions. Second management top-tick after the $64.25 secondary (8.56M sh, 4/21).
  • Require volume normalization + higher-low + reclaim/hold above $100 (ideally weekly close back over the broken shelf) before re-engaging. The easy leg ($64->$143) is gone; only a clean re-based setup is worth expressing.
  • Export-permit overhang is the real fundamental swing factor: US InP permits pending/under active review backlog can't fully ship to US customers without approval. Track permit headlines as a discrete catalyst with no fixed date.
  • Q2 2026 earnings ~2026-07-29 (est., outside 30d) first guided-profit test (>=$34M rev). Confirm exact date via IR; treat 3 trading days pre-print as binary risk and avoid sizing into it.
  • Archetype kept at 2 (Picks & Shovels — $5.67B-cap InP supplier with real revenue inflection), but the tape behaves with a6 reflexivity (52-wk $1.67-$143.16); size with squeeze-grade risk even at.
  • Read-through watch: COHR/LITE/AAOI/AEHR/AOSL moved with AXTI 6/2.L for substrate ASP. Peers holding while AXTI breaks = single-name insider/dilution problem; peers breaking too = optical theme cooling. Photonics ETF (EUV) basket as saturation gauge.
  • Theme status: optical-interconnect ACCELERATING sector-wide, but for THIS name flipped ACCELERATING -> SATURATED/distribution post-6/5. Reclassify to buyable only on a structure repair above $100.
  • Q2 2026 earnings ~2026-07-31 (est.) CONFIRM via IR before any pre-print sizing. This is a first-profit-in-years print, i.e. a live binary, not de-risked; shrink/avoid fresh entries 3 trading days out.
  • Coherent master development + supply agreement (announced 7/2, effective 6/25): $22.29M customer prepayment, Beijing capacity expansion 2026-2028, agreed pricing. Customer-funded capex from an NVIDIA optical partner watch for a backlog raise on the Q2 call as the confirmation.
  • Casela $25.4M long-term InP supply deal (6/17) for 2027 deliveries second multi-year commitment inside four weeks.
  • Tongmei withdrew Shanghai STAR Market IPO application (7/8), pivoting to a Hong Kong listing long-touted China value-unlock catalyst now delayed/venue-changed; monitor for an HK filing.
  • the last known supply overhang event has now passed; removes a seller but doesn't repair structure by itself.
  • No buyable setup until a reclaim + weekly-close hold above $100 with a higher-low and normalizing volume. Micro/small-cap with elevated short interest and thin float hard 1-2%/name cap if a squeeze setup re-emerges on a clean breakout.
  • IQE.L (GaAs/InP peer) and COHR/CRDO/ALAB/LITE are the read-through cluster; if IQE is dead while AXTI rips, treat it as single-name mechanics, not sector ASP repricing.
  • Market-cap stamps are the only price proxy in the feed: $3.89B (2026-07-02) → $2.97B (2026-07-17) = ~-24% on an unchanged share count, implying a high-$60s/low-$70s handle. Confirm live quote before any sizing.
  • The $83 June distribution low broke during the week of 2026-07-14; the prior structural read is stale below that.
  • $64.25 is the reference level that matters Underwritten paper is held there; it is both support and supply.
  • Q2 print ~2026-07-31 unconfirmed verify via IR. Binary risk rule applies inside 3 trading days.
  • Tongmei STAR Market IPO withdrawn 2026-07-08 for a Hong Kong listing value-unlock catalyst is now undated. Do not model it.
  • IQE.L remains the read-through for sector-wide compound-semi repricing vs AXTI-specific dilution overhang.
  • Beginner-trap flag: former parabolic leader >50% off high with improving fundamentals is the classic average-down invitation. Require a higher low + weekly close above $90 before treating the structure as repaired.
  • Lesson carried: the 2026-04/05 'management top-ticked the squeeze' read was wrong on direction the $64.25 raise funded the InP capacity that drove $64→$143. A funding event is not automatically bearish; the timing of the tape is separate from the quality of the raise.
  • Pattern saved: 5 Benzinga prints in 11 sessions (2026-04-06 to 04-16) front-ran a thin-float compound-semi squeeze. Watch for the same signature on the next analog.

Related · shared themes

MRCY

Mercury Systems Inc

Defense-electronics turnaround re-rating inside an accelerating modernization tape: Q3 FY26 (2026-05-06) printed record bookings $348M (+73.7% YoY), 1.48 book-to-bill, record ~$1.6B backlog and +46% EBITDA, with FCF guided positive. But the stock just reversed ~16% off its $128.45 ATH; the ~2026-08-10 Q4/full-year print is the next binary.

MEDIUM

OKTA

Okta, Inc.

Identity-security re-rating entering a second, sell-side-led leg: a post-print upgrade wave (KeyBanc street-high $175 on 07-10, Scotiabank upgrade to $165 on 07-06) is still building six weeks after the 05-29 beat-and-raise, lifting the top target 17% in a month. Theme ACCELERATING; the risk is extension and hot retail into a digested 52-week-high gap.

MEDIUM

SNX

TD SYNNEX Corporation

Q2 blowout (rev $19.575B vs $16.8B est, 6/25) broke the seasonal guide-down bear case and the sell-side is chasing higher (MS $374, UBS $352). The channel re-rating is now confirmed by fundamentals, but the catalyst is spent and the next binary is ~3 months out current levels chase the post-earnings gap rather than buy the setup.

MEDIUM

XMTR

Xometry, Inc.

AI-native manufacturing marketplace re-rating on a profitability inflection Q1 marketplace revenue +40% YoY, FY guide raised, Siemens embed plus ~$50M stock buy. Wedbush 7/16 Outperform $126 reopens Street upside above spot after June's above-cluster trade; the ~2026-08-06 Q2 print is the next binary.

MEDIUM

See also · stocks to watch

optical networking stocks to watch semiconductor equipment stocks to watch