# AXTI - AXT Inc
## Date: 2026-03-24
## Score: 47/100

### Thesis (2-3 sentences)
AXT controls 60-70% of global indium phosphide (InP) substrate production -- the foundational wafer material that every AI optical transceiver laser is built on. This is the ultimate bottleneck play: without InP wafers, there are no EML lasers, no 800G/1.6T transceivers, and no AI optical buildout. However, the company is tiny ($99M TTM revenue), unprofitable, heavily dependent on its Beijing facility (geopolitical risk), and has been plagued by Chinese export permit delays.

### Key Metrics
| Metric | Value | vs Sector |
|--------|-------|-----------|
| Market Cap | ~$2-3B | Micro-cap (volatile) |
| Revenue (TTM) | $99.4M | Tiny |
| Gross Margin | 24.0% | Low |
| Operating Margin | -14.9% | Still losing money |
| Net Income (TTM) | -$11.6M | Unprofitable |
| InP Market Share | 60-70% | Dominant |
| 52-Week Range | $1.13 - $65.24 | Extreme volatility |
| Analyst Consensus | Hold/Buy (3-4) | Target $24-31 |
| Capacity Expansion | July 2026 | Critical milestone |
| Capital Raise | $100M (Dec 2025) | For InP expansion |

### Moat Analysis
**Rating: Strong but Fragile (3.5/5)**

AXT's moat is geological and process-based:
- **60-70% InP substrate market share**: Near-monopoly on the critical wafer material for all AI optical transceivers. Lumentum, Coherent, and every transceiver maker depends on AXT substrates.
- **Process know-how**: Growing InP crystals is extremely difficult, requiring years of process optimization. Few competitors can match AXT's crystal quality and yields.
- **Bottleneck leverage**: With transceiver demand exceeding supply by 2x, InP substrates are the binding constraint. AXT sits at the choke point of the entire value chain.
- **CRITICAL RISK -- Beijing manufacturing**: Primary production is in China. Export permits have already caused revenue disruptions. Any escalation of US-China tensions could permanently impair AXT's ability to deliver to Western customers.
- **Competitor threat**: Japan's Sumitomo Electric and JX Metals hold smaller shares. If demand persists, these competitors will invest to break AXT's dominance.

### Catalysts
1. **InP capacity expansion (July 2026)**: Major capacity addition funded by $100M raise. If successful, revenue could step-change.
2. **Export permit resolution**: Q4 revenue fell on permit delays. Permits awarded = immediate revenue pop.
3. **Structural demand**: Every 800G/1.6T transceiver needs InP. With 63M units expected in 2026 (2.6x YoY), substrate demand is assured.
4. **Pricing power**: Supply deficit gives AXT leverage to raise prices.

### Risks
1. **China manufacturing concentration**: Beijing facility risk is existential. A US-China export ban or sanctions escalation could cripple the company overnight.
2. **Unprofitable**: -15% operating margin. Growth isn't translating to profits yet.

### Update 2026-08-17 — InP tightness + LITE/COHR lock-ups

Bloomberg 17 Aug: $81.64, mcap $5.35bn, TTM P/S 33.5x, TTM GM 12.7%, OM −24.9%, TTM sales $88m / NI −$21m. 1M +78%, YTD +399%, 1Y +3,882%. 5 analysts, rec 4.60, BEST TP $91.60.

The P&L has started to inflect even if TTM is still a loss: 华泰 31 Jul said Q2 InP sales >$30m, blended GM 45%, non-GAAP OP $11.2m, backlog >$100m, YE26 quarterly InP capacity ~$60m → YE27 ~$130m. BEST FY27E sales $218m (+$75m in 4 weeks) / EPS $0.87; FY28E sales $459m / EPS $1.92.

LITE $87m prepaid through 2031 and COHR $22.3m prepaid are confirmed; LITE’s 11 Aug call said they “found additional substrate help from AXTI.” Semiconsight: >50% of Q2 revenue already China-domestic (no license), US license still the swing.

The 17 Aug 經濟日報 >10% Q4 hike is consistent with Citi’s 16 Jul 15% extra LITE FY27 price and with the Feb-2025 China InP export-license shock, but it is not a new AXTI filing. Fragile-choke framing is unchanged: this is still Tongmei + MOFCOM wearing an InP multiple. Repo verdict stays AVOID on quality-of-listing grounds even though the constraint is real.

3. **Extreme stock volatility**: $1.13 to $65.24 in 52 weeks. This is a speculative vehicle.
4. **$2-3B market cap on $99M revenue**: 20-30x sales while unprofitable. The AI premium is enormous.
5. **Capital raise dilution**: $100M equity raise dilutes existing shareholders.
6. **Competitor response**: Japanese suppliers investing in InP capacity. AXT's share could erode over 2-3 years.

### Valuation
At 20-30x TTM sales while unprofitable, AXT is priced purely on its strategic bottleneck position. If capacity expansion succeeds and revenue reaches $200-300M with normalized 30%+ gross margins, the company could justify a $2-3B valuation. But the China risk creates a binary outcome -- either the geopolitical situation remains manageable and AXT captures the InP supercycle, or it doesn't and the stock returns to single digits. This is a lottery ticket, not an investment.

### Verdict: AVOID
AXT owns the most strategically important position in the AI photonics supply chain -- the InP substrate bottleneck -- but the China manufacturing risk, unprofitability, extreme volatility, and inflated valuation make it uninvestable for most portfolios. The thesis is "own the bottleneck material," but the execution risk and geopolitical risk are too high at current prices. Monitor for: (1) successful capacity expansion outside China, (2) profitability inflection, (3) significant price correction. If the company establishes non-China manufacturing and reaches profitability, this could be a compelling BUY.

**Final Score: 47/100**