AXTI · AXT Inc

Current Thesis

AXT is the indium-phosphide (InP) substrate supplier sitting upstream of the AI optical-interconnect build the wafer material under the high-speed lasers and photodetectors that move data between accelerators. The fundamental story is real and accelerating: Q1 2026 (4/30) put InP above 50% of revenue for the first time, backlog hit a record $100M, and Q2 is guided to first profitability in years. The tape, however, just broke. After a parabola from the $64.25 April secondary to a $143.16 ATH on 5/26, the stock fell -15.99% on 6/5 to ~$89.04 on news of a >$22M CEO insider sale, slicing through the $90 base that prior analysis flagged as the line between pullback and distribution. The narrative is intact; the price structure is unwinding. This is a stand-aside until a higher-low base reforms, not a knife to catch at $89.

Bull Case

• Q1 2026 (reported 4/30): revenue $26.9M, +39% YoY / +17% QoQ; InP revenue surpassed 50% of total for the first time; non-GAAP gross margin 29.9% vs 21.5% prior quarter; EPS -$0.01 vs -$0.05 consensus (80% beat).
• Record $100M InP backlog (Q1 call, 4/30), up from ~$60M in Q4 2025 hard visibility into the AI data-center optical ramp as 800G→1.6T transceivers scale.
• Q2 2026 guide (4/30): revenue ≥$34M and first GAAP + non-GAAP profitability in years the inflection is guided, not modeled.
• Capacity doubling in 2026 and again in 2027 (4/30 call). AXT designs its own crystal-growth furnaces and controls raw materials via its Jinmei subsidiary supply-constrained InP implies pricing power.
• NVIDIA's 3/2/2026 $2B-each optical partnerships with Coherent and Lumentum ($4B total) validate optical interconnect as the AI bottleneck; AXT sells the InP substrate those lasers are grown on, one step upstream of the partners.
• Northland PT raised to $125 from $90 (Outperform, early June); estimates lifted 5/8 sell-side catching up to the narrative after the move.
• Sector cluster 6/2: Aehr (+15%), AAOI, AOSL and AXTI bid together a compound-semi/optical group move, not a lone-name pop.

Bear Case

• 6/5: -15.99% (-$16.95) to $89.04 on a violent $83.15–$104.55 range; -38% from the $143.16 ATH (5/26) in ~7 sessions. The $90 base is lost, with no confirmed higher-low underneath.
• CEO Morris Young disclosed a >$22M insider sale (early June) Insiders selling into strength is the cleanest tell that the easy money is behind the move.
• Valuation still rich: ~$5.67B market cap (6/5) on ~$27M quarterly revenue is north of 50x sales. A single InP order push-out re-rates this fast.
• US InP export permits remain pending/under active review. A chunk of backlog cannot ship to US customers without approval; non-US permits improved but US permits are the gating risk on converting that $100M to revenue.
• 52-week range $1.67–$143.16 describes a reflexive small-cap that round-trips hard the same reflexivity that drove the run cuts the other way.
• The story is 100% forward (backlog + guide); trailing revenue is still only $26.9M. Any slip on the capacity-doubling timeline breaks the setup immediately.
• China/Tongmei (Jinmei) geopolitics plus an unfixed STAR Market IPO timeline for the subsidiary remain an overhang.

Setup & Price Structure

• Archetype: Picks & Shovels by fundamentals a multi-billion-cap InP supplier with a genuine revenue inflection, no longer the thin-float micro-cap of the early squeeze. But the tape is trading with retail-squeeze reflexivity, so sizing should respect a6-style risk even though the classification is 2.
• Key levels: $89.04 (6/5 close); ATH $143.16 (5/26); the April secondary cleared at $64.25 (4/21); the $90 shelf was lost 6/5. The rising 20-EMA that buyers leaned on near ~$100–110 has been broken to the downside.
• A -16% session through a defined base, catalyzed by an insider-selling headline, reads as distribution rather than a bull-flag consolidation. Buyers have no structural reference below current price until the secondary-clearing zone far beneath.
• Re-engagement blueprint: wait for selling volume to dry up, a higher-low to print, and a reclaim that holds above $100 (ideally a weekly close back over the broken shelf). Chasing at $89 mid-waterfall is the lowest-quality way to express a thesis whose entire easy leg has already run.

Catalyst Calendar (next 30 days)

• No scheduled hard catalyst inside 6/6–7/6. The absence of a near-term driver into a broken tape removes any momentum reason to be early.
• US InP export-permit decisions pending / under active review; can land any week, no fixed date. Approval unlocks shippable backlog; denial caps it.
• Insider Form 4 flow watch for continued CEO/insider sales after the >$22M early-June disclosure; more selling extends the distribution.
• Annual meeting reconvened 2026-06-04 (passed; previously delayed for lack of quorum) housekeeping, not a price catalyst, but a sentiment marker around the 6/5 break.
• Q2 2026 earnings: ~2026-07-29 (est.) just outside the 30-day window; the next binary and the test of the guided first-profit quarter. Confirm the exact date via IR before any pre-print positioning; treat the 3-trading-day pre-print window as binary risk.
• Tongmei/Jinmei STAR Market IPO timeline unfixed; a structural wildcard, not a dated event.

What Would Change Our Mind

• Bullish re-engage: a reclaim and hold above $100 on normalizing volume with a clean higher-low = structure repaired and the optical narrative re-asserting; that is the level to act on, not $89.
• Thesis confirmation: Q2 (~7/29) printing ≥$34M revenue with the guided first profit and backlog still ≥$100M = the inflection is real and the theme re-accelerates for the name.
• Overhang removal: granted US InP export permits convert the can't-ship backlog into revenue.
• Thesis degradation toward DEAD: InP backlog shrinking QoQ from the $100M record, permits denied, further large insider selling, or a daily structure that fails to reclaim $100 and instead grinds toward the $64.25 secondary zone.

Correlation Notes

• Direct read-throughs: COHR and LITE (NVIDIA's optical partners), plus AAOI, AEHR, AOSL the group moved together 6/2. Watch whether peers hold while AXTI breaks: divergence points to a single-name dilution/insider problem rather than a sector roll; peers breaking too would signal the optical theme itself is cooling.
• IQE.L (GaAs/compound-semi) remains the cleanest sector read-through for substrate ASP repricing vs single-name overhang.
• Photonics ETF (EUV) basket 15 of ~40 holdings up triple-digits YTD as of 5/13; if the basket rolls over, the theme is saturating and AXTI's bounce odds fall.
• Broad AI-semi infrastructure tape hyperscaler capex +67% YoY in 2026 (Stifel, May) underpins structural InP demand, but the group's risk appetite drives the reflexive small-caps first on the way down.

The underlying optical-interconnect theme remains ACCELERATING sector-wide, but for this single name the status has flipped from ACCELERATING to a SATURATED/distribution phase after the insider-selling break. Re-classification back to a buyable accelerating setup requires the structure to repair above $100.