AI

Stock Market

Stocks

Investing

Investment

AXT Inc ($AXTI) — substrate maker, geopolitical pivot point, and optionality engine

Executive summary

AXT is a compact but strategically placed supplier of compound semiconductor substrates (InP, GaAs, Ge and sapphire/ceramic products) used in lasers, photonic integrated circuits (PICs), RF front-ends, LIDAR, sensing and other high-frequency / high-performance applications. Near-term performance is volatile — Q2 2025 revenue contracted and margins have been pressured — largely because of China export-permit timing and cyclical demand in China, but the company sits on three structural, under-appreciated levers:

1. End-market growth: the photonics, mmWave/5G–6G and GaN/GaAs-enabled RF markets are entering multi-year expansion driven by AI/datacenter optics, connectivity, LiDAR, and power electronics. These end markets materially expand the addressable market for AXT’s wafers and substrates.
2. Operational optionality: AXT’s China subsidiary, Tongmei, and its planned STAR-market IPO are a latent value engine (capacity, local market share, potential capital injection) — but also a regulatory flashpoint. Management has repeatedly signaled Tongmei’s listing progress, and Tongmei’s export-permit story is central to 2025’s revenue rhythm.
3. Supply-chain and geopolitical arbitrage: western fabs and foundries are actively diversifying compound-substrate supply and importing materials that are difficult to substitute quickly. AXT’s mix (GaAs/InP/Ge and sapphire) positions it as a diversified supplier to multiple expanding photonics/RF ecosystems. That creates optionality for margin expansion once permit friction and China demand normalize.

This is not a low-volatility, cash-flow-safe name — near-term Q1/Q2 2025 showed negative gross margins and widening losses — but for investors who can stomach execution risk and geopolitical noise, AXT is a levered way to play compound-semiconductor industrialization. The biggest binary is Tongmei’s regulatory/capital outcome and the speed of export-permit processing: both determine short-term cash flow and whether the company can parlay demand tailwinds into profitable scale.

What AXT actually does

AXT designs, manufactures and sells substrate wafers — the foundational crystalline material (GaAs, InP, Ge, sapphire, ceramic) on which epitaxial layers are grown to create lasers, photonic devices, RF power transistors and other compound-semiconductor devices. These substrates are the raw inputs for MOCVD/MBE epitaxy and are essential where silicon cannot meet the optical/electrical requirements (for example, direct 1550 nm lasers, high-electron-mobility RF devices, photodiodes, etc.). AXT’s product mix is intentionally broad: InP for photonics (lasers, PICs), GaAs for RF and optoelectronics, Ge for infrared detectors and some silicon integration use-cases, and sapphire/ceramic for LED/GaN and specialized optical applications.

Two operational structures matter:

• U.S. HQ / global sales and R&D (Fremont) — sells into global OEMs and foundries.
• Beijing Tongmei Xtal Technology Co., Ltd. — a China-based manufacturing subsidiary that provides additional capacity, nearer to large Asian customers, but is subject to Chinese export regulation and a proposed STAR-market listing. Tongmei is both a growth vehicle and regulatory nexus.

AXT’s customers are device makers and foundries that then sell modules and chips into telecom, data center optics, lidar/autonomy, industrial lasers, defense and consumer RF markets.

The hard facts from 2025 that shape the near-term outlook

• Q2 2025 revenue: ~$18.0M (down from Q2 2024’s $27.9M; down sequentially from Q1 2025’s $19.4M). Gross margin was 8.0% (non-GAAP improvement vs Q1 negative gross margin). Net loss after minority interests: about $7.0M in Q2. Cash and restricted cash around $35.1M at June 30, 2025.
• Guidance & misses: The company revised its Q2 preliminary revenue guidance downward in early July 2025 — citing slower-than-expected issuance of export control permits for GaAs products and weaker demand in China as drivers. Export-permit processing (for certain finished or raw materials) introduced timing risk: shipments that would have closed in Q2 were delayed. Management framed this as administrative timing rather than permanent loss of business — but it created margin/loss volatility.
• Tongmei STAR-market process: Tongmei’s application for a STAR-market IPO was accepted and progressed through rounds of questions since 2022; the listing remains in process and subject to CSRC/SSE approval — success could unlock capital and separate valuation for Tongmei’s China operations. That process is multi-year and bureaucratic; management continues to discuss it in investor decks and filings.

These are the immediate anchors: revenue volatility driven by China permit timeliness, a pending (but uncertain) capital event (Tongmei IPO), and an end-market mix that looks attractive on multi-year forecasts but is lumpy quarter-to-quarter.

Why AXT matters to the AI / photonics / GaN upcycle — three deep connections

Many investors think substrates are commoditized, low-margin — which is sometimes true — but that view misses three structural shifts that make substrate suppliers strategic bottlenecks:

1) Photonic integration and InP demand

AI-scale data centers are rapidly expanding optical I/O needs — hyperscalers and networking vendors are exploring PICs and InP-based solutions for high-density optics and laser integration. Indium phosphide is the material of choice for many photonic active components (lasers, modulators, detectors) at telecom wavelengths; as PIC adoption grows, InP substrate demand scales directly. Research forecasts show PIC/InP markets expanding at double-digit CAGR through the decade. AXT supplies InP wafers that feed that ecosystem; a meaningful adoption wave materially raises wafer demand and pricing power.

2) GaAs and RF/mmWave tailwinds (5G → 6G → LIDAR)

GaAs is central to high-frequency RF front-end modules (pHEMTs, MMICs) used in 5G base stations, satellite comms and mmWave links, and some lidar / sensing technologies. The roll-out of higher-frequency spectrum (mmWave, E-band) and the densification of wireless infrastructure increase component counts and complexity — which lifts GaAs substrate demand. GaAs wafer markets are forecast to grow strongly — that top-level growth flows back to substrate suppliers.

3) GaN and sapphire adjacency

GaN power and RF devices are often grown on sapphire (for LEDs) or SiC/single-crystal GaN for power devices — AXT’s sapphire/ceramic product lines and capacity could be re-purposed or cross-sold to GaN device makers (LEDs, power, RF). As GaN power devices scale for EV chargers and data-center power conversion, the substrate ecosystem broadens; AXT’s diversified product set gives optionality to capture this adjacency.

Put differently: AXT is not a single-product vendor — it is a substrate portfolio company sitting at the cross-roads of three secular technology ramps (photonic PICs/InP, mmWave/GaAs RF, and GaN/sapphire power & LEDs). Each ramp amplifies substrate demand at scale, offering AXT the potential to move from lumpy low-margin quarters to sustained revenue expansion if supply and regulatory constraints are managed.

The geopolitics & export-permit story

Two linked realities create a distinct AXT risk/option:

1. China manufacturing footprint: AXT operates Beijing Tongmei, which means a chunk of production (and therefore deliveries to Asian customers) flows through China. That proximity gives cost advantage and proximity to large Asian customers but exposes that revenue to PRC export control regimes.
2. Export controls and permit timing: Since 2023–2025, China tightened export controls on some critical materials (e.g., GaAs earlier, and InP in February 2025 per management commentary). AXT must apply for export permits for some product shipments; processing delays can push orders across quarters. Management told investors that Q2 2025 shipments were delayed due to permit timelines — generating a revenue shortfall versus guidance. The market trades this as execution risk, but the company distinguishes administrative timing from secular loss of demand.

This creates a paradox: AXT’s China presence is strategic for cost and market access, but it also makes AXT sensitive to policy windows. Those windows are unpredictable and can create step-changes in reported revenue — and in investor sentiment. The escape valve is timing and diversification: either Tongmei’s STAR IPO unlocks onshore capital and regulatory goodwill, or AXT increases non-China production or inventory buffers to smooth shipments.

Tongmei’s STAR-market IPO

Tongmei is both a growth vehicle and a regulatory lever:

• Upside: A successful Tongmei IPO would likely (a) unlock capital for expanded capacity in China (where most global PIC/GaAs production sits), (b) separate valuation of the China operations (lessee of AXT’s manufacturing IP and capacity), and (c) give AXT the optionality to harvest proceeds and de-lever or reinvest into U.S. capacity. The investor-deck repeatedly references Tongmei’s IPO submission and SSE approval rounds — a parallel storyline investors watch for value-unlocking.
• Downside: the listing is subject to regulatory review and geopolitical sensitivities, and the process since 2022 has been multi-year and uncertain. Furthermore, Tongmei’s listing and local ownership could centralize certain regulatory relationships but also subject the business to PRC domestic investor dynamics (PE redemptions, policy changes). AXT’s filings explicitly warn the STAR-market process is lengthy and uncertain.

Implication for us: Tongmei is a classic optionality asset — if it lists and markets well, it materially derisks capacity funding and can be a near-term catalyst; if it stalls or runs into politics, AXT remains exposed to permit timing and localized demand softness.

Operations & margin dynamics

AXT’s P&L illustrates the typical substrate-business shape: revenue is lumpy, gross margins can swing materially with mix and utilization, and fixed costs (process control, quality, R&D) are significant. Q1 2025 showed negative gross margins and a sizeable net loss; Q2 2025 improved to single-digit gross margin, but the company is still loss-making on GAAP. Management has cut OpEx sequentially, and non-GAAP gross margin improved, but the near-term story is recovery-from-permit/timing shocks, not yet structural margin expansion.

Two levers can improve margins materially:

1. Capacity utilization — wafer production has high fixed cost; higher steady utilization drives leverage. That requires smooth demand and permit-free shipping windows.
2. Product mix — selling more InP and specialty substrates (higher ASP) versus commodity sapphire can boost blended margin. As PIC adoption grows, InP ASPs should be higher than commodity sapphire volumes.

AXT’s margin risk is therefore a blend of demand timing and product mix. The company’s investor materials highlight the strategic push into AI/datacenter connectivity and industrial lasers — markets with higher ASPs — which, if penetrated, could structurally raise gross margins.

Strategic moves AXT can make

This is where creative, company-level strategy matters. If I were advising the board or sizing scenarios as an investor, I’d watch for — or pressure for — the following moves:

1) Lean vertical integration into downstream epitaxy/service

AXT could selectively move from pure substrate sales into co-packaged substrate+epitaxy supply or become a “foundry friendly” materials partner (sell epi-ready, tested substrate stacks). By bundling substrate + process recipes + QA for specific foundry partners, AXT could command premium pricing and longer-term contracts. This is not quick — it requires new capabilities — but it shifts the company from commodity supplier to trusted supply-chain partner for PIC and RF foundries. (Think of the step from raw polysilicon to ingots vs. fully characterized wafers.)

2) Expand non-China capacity selectively (supply diversification)

AXT should consider targeted capacity expansion in jurisdictions with stable export regimes (e.g. U.S., Taiwan, Southeast Asia) to reduce permit timing exposure. That’s capex-heavy — but even partial geographic diversification of finished product shipments would mute quarter-to-quarter volatility. Tongmei’s IPO could partly fund this or AXT could seek JV partnerships with foundries that pre-pay capacity.

3) Product focus: InP + higher-margin specialty wafers

AXT benefits if it captures PIC-focused InP demand (lasers, integrated photonics). That may require process upgrades, tighter spec control and longer-term supply agreements with PIC vendors. The market for InP is forecast to grow aggressively with PIC adoption — capturing even a small share of that growth changes revenue mix materially.

4) Services and testing play

Offer added services (substrate metrology, pre-epi inspection, supply-chain traceability and certification) as fee-based products. Customers paying for reduced integration time and yield ramp are willing to pay for reliable substrate quality — a high-margin adjunct to wafer sales.

5) Strategic partnerships with GaN and SiP foundries

AXT can sign binding multi-year supply contracts with GaN, PIC and SiP foundries to lock in volumes and give financing partners visibility. This reduces cyclicality and builds stickiness.

Each move trades capex/complexity for margin stability and valuation multiple expansion. The trick is sequencing: start with binding customer commitments (paid pilot volumes), then invest capacity.

Competitive landscape — who wins if substrates go hot?

AXT sits between a handful of specialized suppliers and larger materials conglomerates. Competitors include IQE, Soitec/Soitec’s engineered substrates, II-VI/Coherent (now part of Coherent? consolidated landscape), Sumitomo and some regional suppliers. The substrate market has high barriers (crystal growth expertise, polishing, yield discipline) so share can be sticky — but scale matters. AXT’s advantage is its product breadth plus China manufacturing footprint via Tongmei, which can be a cost advantage if permit timing is normalized.

The threat is consolidation: larger vertically integrated players (with deep pockets) can buy scale quickly and compress margins. That said, customers often prefer multiple qualified suppliers for risk management, and qualifying a new substrate vendor is multi-quarter process — a moat for incumbents who keep quality and cadence.

Scenario analysis — three plausible futures

I’ll present three focused scenarios (Bear / Base / Bull), tie them to likely timeframes, and explain what investors should watch to move between scenarios.

Bear (40% probability) — regulatory and demand headwinds persist

• What happens: China export controls remain slow; Tongmei IPO stalls; China demand softens; AXT cannot redeploy inventory fast enough; margins stay depressed and cash draws down.
• Financial outcome: Continued quarterly losses; potential need to raise capital; multiple compression to low single-digit EV/revenue.
• Key signals: Recurrent permit delays, Tongmei listing stalled further, consecutive quarters of negative gross margin.

Base (45% probability) — normalization & selective recovery

• What happens: Permits normalize, Tongmei listing moves forward or is resolved favorably, China demand recovers modestly, AXT captures rising InP orders from PIC/AI optics customers. The company modestly improves utilization and reduces fixed-cost leakage.
• Financial outcome: Revenue recovery to mid- to high-$20M quarterly run-rate over 12–24 months; gross margins climb toward the low-to-mid teens; losses narrow and the company reaches break-even or modest profitability on a non-GAAP basis. Multiple re-rating as growth becomes visible.

Bull (15% probability) — structural step-change and re-rate

• What happens: Photonics and mmWave ramps accelerate; AXT secures multi-year contracts with major PIC and RF foundries; Tongmei IPO provides CAPEX to expand capacity quickly; AXT executes product-mix shift toward InP and specialty substrates.
• Financial outcome: Multi-year revenue CAGR in the 20–40% range; gross margins rise to 20–30% as mix shifts to higher ASP products and utilization improves; company trades at a sector-growth multiple (EV/Revenue multiple re-rating). Tongmei IPO proceeds allow rapid capacity expansion, further compounding growth.

Implication for us: the path from Base → Bull is primarily about customer contracts and capacity funding (i.e., binding orders + Tongmei listing or equivalent financing). The path from Base → Bear is permit/regulatory disruption or secular demand softness.

Catalysts & specific near-term KPIs

Concrete, observable events will move the needle on this thesis. Watch these in each quarterly cycle:

1. Export-permit cadence & management commentary — any statement that permit timelines shorten materially (or conversely lengthen) is high-signal. AXT explicitly said Q2 shortfall driven by permit timing.
2. Tongmei STAR-market updates and IPO progress — any positive CSRC/SSE step, or concrete timeline for float, is a value catalyst; conversely, new regulatory friction signals continued uncertainty.
3. Order book / backlog disclosures — management notes on multi-year supply contracts with PIC or RF foundries (binding MOUs with volume commitments) are a de-risking event.
4. Product-mix shift evidence — rising percent of InP and specialty-substrate revenue vs commodity sapphire/GaAs low-margin sales will show margin expansion in future quarters.
5. Capacity expansion announcements / JV funding — Tongmei IPO proceeds, partner-capitalized plants, or other evidence of capex funding.
6. Customer wins in hyperscale photonics or tier-1 RF — public wins by major PIC or RF-device OEMs that name AXT as qualified supplier would be transformational (these are often disclosed as supply agreements).
7. Quarterly cash runway and working-capital movement — looks for cash trends (AXT had ~$35.1M cash+restricted at June 30). Sustained cash burn without capital events is a risk.

Those seven items are the scoreboard (at least for me). Positive movement on multiple items accelerates the Base → Bull path; negative movement increases the chance of Bear outcomes.

Balance-sheet and capital questions — runway & funding vectors

AXT had roughly $35M in cash+restricted cash at June 30, 2025, and net losses in the recent quarters. That cash cushion is non-trivial but limited if demand recovery stalls. The company has several funding levers:

• Tongmei IPO proceeds — the biggest, albeit least certain, lever; proceeds could be used to expand capacity, pay down short-term liabilities, and smooth cash cycles.
• Customer pre-pay / long-term purchase commitments — securing downstream commitments can support asset finance or bank financing for capacity.
• Equity / convertible raises — likely if needed, but would dilute equity and compress upside.
• Strategic partnership / JV investment — partnering with foundries or materials conglomerates for financed capacity build-out reduces dilution while locking volumes.

Our preference: favor binding offtake or capital via Tongmei listing or strategic JV rather than repeated pure-equity raises, because the latter reduce option value.

Risks

I’ll list the most likely failure modes in descending order of likelihood/impact:

1. Regulatory timing risk (export permits) — manifests as repeated quarterly misses and revenue lumpiness. That is already material in 2025.
2. Tongmei listing stall or adverse conditions — if the IPO fails or is delayed indefinitely, expected capacity funding and derisking vanish.
3. Demand softness in China / Asia — because a high proportion of customers and manufacturing are in Asia, a localized slowdown hurts AXT harder than a diversified competitor.
4. Competitive pricing pressure — large integrated suppliers expand capacity and compress prices; customers consolidate supplier lists.
5. Operational yield problems at higher volumes — moving from pilot quantities to high-volume, high-quality substrates requires strict process control; yield surprises can erode margins.
6. Capital scarcity — if AXT must dilute equity to fund capacity repeatedly, shareholder returns compress.
7. Macroeconomic shocks to telecom / capital spending — the substrate market is downstream of capex cycles in telecom and datacenter operators.

Each risk has mitigants, but we must expect volatility and watch the KPI scoreboard.

Valuation framing

AXT is best thought of as a cyclical growth hardware supplier with embedded optionality. Traditional DCFs struggle with the permit and IPO binary, so treat valuation in three parts:

1. Base substrate business value: model near-term revenue (Q3–Q4 2025) assuming permit normalization, apply modest gross margins (10–15% for mixed product), and use a sector-appropriate multiple for a small industrial (EV/Revenue 0.5–1.0x depending on margin profile). If revenue re-accelerates into multi-year growth and margins expand, the multiple should expand.
2. Tongmei optionality: treat Tongmei as a call option — assign value if the IPO proceeds or if Tongmei’s listing clears regulatory hurdles. Market commentary treats Tongmei as a latent asset; successful IPO would likely be a multi-100% upside catalyst, because it could represent a large portion of AXT’s manufacturing footprint and differentiated market coverage.
3. Strategic/latent upside: quantify potential upside from product-mix shift to InP and PIC supply (high ASP) and capture of new-market growth in GaN and photonics — these are leveraged to top-line and multiple expansion if realized.

We should stress-test models across permit scenarios, and price in a binary: either permit & IPO risk resolves (higher valuation) or it remains a governance/regulatory drag (lower valuation).

Final synthesis

AXT is small today but strategically placed. Substrates are a low-glamour corner of the semiconductor supply chain until the market needs them badly — and the confluence of PIC adoption for AI/data-center optics, mmWave expansion, GaN scaling for power, and lidar/autonomy create such a moment. AXT’s product breadth (InP, GaAs, Ge, sapphire) provides a unique optionality: if it can stabilize its China export pipeline (or fund alternative capacity) and shift mix toward higher-ASP InP/PIC sales, the company could transition from a lumpy industrial to a higher-growth, higher-margin supplier.

The single largest practical lever is Tongmei and export control resolution: that determines whether AXT is a timing story for quarters or a structural growth company. Let’s watch permit cadence, Tongmei IPO progress, and customer contract wins closely. If you like asymmetric, industrial-tech bets that combine geopolitics with semiconductor supply cycles, AXT is one to follow .