ticker semi-equipmentcompound-semimocvdgansic updated 2026-04-05

AIXA -- Aixtron SE

Identity

Field	Detail
Full legal name	AIXTRON SE
Ticker / Exchange	AIXA / XETRA (Frankfurt, Prime Standard)
OTC ADR	AIXNY (US)
ISIN	DE000A0WMPJ6
Indices	MDAX, TecDAX
Sector	Information Technology / Semiconductor Equipment & Materials
Industry	Deposition Equipment (MOCVD / CVD)
HQ	Dornkaulstrasse 2, 52134 Herzogenrath, Germany (Aachen region)
Founded	1983 (spin-off from RWTH Aachen University)
Employees	1,117 (end FY2025; down from 1,207 at end FY2024)
Website	aixtron.com
Legal form	Societas Europaea (SE) -- governed by AktG + EU SE Regulation
Shares outstanding	113,456,120
Market cap	~EUR 1.96B (Dec 2025 at EUR 17.31/share)
Enterprise value	~EUR 1.74B (net cash ~EUR 225M)

One-sentence description: Aixtron builds the machines that grow the exotic semiconductor materials (GaN, SiC, InP, GaAs) used in EV inverters, data center lasers, fast chargers, and AI server power supplies. If you want to make a compound semiconductor chip, you almost certainly need an Aixtron machine.

Thesis

The Setup

Aixtron is a 77% MOCVD market share monopolist currently caught in a cyclical downturn (FY2025 revenue -12%, FY2026 guided -7%). The market is pricing near-trough earnings while the structural growth drivers -- AI data center power demands, optical interconnect buildout, EV recovery, and the GaN power inflection -- remain fully intact. This is a classic "buy the dominant franchise during the downturn" setup.

Primary category: Undervalued cyclical compounder.

Single most important thing that must go right: The SiC overcapacity must absorb and order intake must inflect upward by late 2026 or early 2027, driving revenue recovery toward EUR 600M+ by FY2027-2028.

Expected holding period: 2-3 years (through one full cycle turn).

Bull Case

GaN power inflection is real and imminent. The GaN power device market reached USD 355M in 2024 and is projected to hit USD 2.9-4.4B by 2030 (42-49% CAGR). Data centers specifically show 53% CAGR. Aixtron has 70-90% market share in GaN power MOCVD. Management explicitly states AI "has the potential to become the largest single application for GaN power semiconductors."
Nvidia's 800V HVDC architecture forces the industry's hand. Rubin (H2 2026) pushes GPU power to 1,800W per chip. Rubin Ultra (2027) doubles that to 3,600W. At the rack level, power density explodes from 120kW (Blackwell NVL72) to 600kW for Rubin Ultra's Kyber rack architecture. Traditional 48V power delivery fails -- at 1MW rack power, you would need 200 kg of copper busbar per rack. The 800V HVDC transition is not optional; it is mandatory. And GaN is irreplaceable in that architecture -- 100x faster switching than silicon, enabling MHz-frequency operation at the 800V-to-48V and 48V-to-point-of-load conversion stages where power density matters most.
Aixtron is in Nvidia's supply chain. At GTC 2025, Nvidia announced partnerships with 14 power semiconductor companies for 800V architecture, including Infineon, Texas Instruments, Navitas, and EPC. Multiple confirmed paths connect Aixtron directly to Nvidia's AI infrastructure:
- Texas Instruments honored Aixtron with its 2022 Supplier Excellence Award for GaN MOCVD. TI subsequently announced 4x GaN capacity expansion and successful 300mm GaN pilots using Aixtron tools. TI is a confirmed Nvidia 800V HVDC partner.
- Infineon acquired GaN Systems for $830M (Oct 2023) and operates from European fabs where Aixtron equipment dominates. Infineon announced at GTC 2025 it is co-developing next-gen 800V power systems with Nvidia. The company is targeting 300mm GaN manufacturing with samples expected Q4 2025.
- EPC's CEO Alex Lidow stated: "Our collaboration with NVIDIA is focused on developing compact, efficient, and cost-effective board-level conversion to power future AI factories at gigawatt scale." EPC's 6kW 800V-to-12.5V converter is exactly the topology required for Rubin/Kyber architectures.
The optoelectronics supercycle is accelerating now. AI is driving explosive demand for optical transceivers (800G+). Aixtron's G10-AsP platform has become the "leading solution" per management, with repeat orders from blue-chip customers. Optoelectronics grew to 23% of systems revenue in FY2025 -- this was the one segment that accelerated during the downturn.
300mm GaN is a massive TAM expansion bet. Current GaN manufacturing uses 200mm wafers; 300mm offers 2.25x the die area per wafer, dramatically improving economics. Aixtron's Hyperion platform (showerhead architecture for 300mm Si fab compatibility) is in pilot production at several leading customers. If 300mm GaN becomes the standard, it opens the entire 300mm silicon foundry ecosystem to Aixtron. Imec's October 2025 announcement of a 300mm GaN Power Electronics Program featured Aixtron's Hyperion as the foundational equipment.
Valuation is cheap relative to peers. Aixtron trades at 40-70% discount to semiconductor equipment peers on every metric. Kerrisdale Capital (long, Nov 2025) argues applying peer-group 33x P/E to projected 2028 EPS of ~EUR 2.00 implies 270%+ upside to ~EUR 66/share. The stock surged 15% on that disclosure alone.
G10 batch tool adoption locks in customers for 5-7 year cycles. Tool qualification takes 6-18 months, costs millions, and creates reactor-specific recipe libraries. Customers do not switch casually. This is a genuine self-reinforcing monopoly.

Bear Case

SiC overcapacity persists beyond 2028. Chinese SiC substrate prices have collapsed from ~$1,500 to ~$400-500 per 6-inch wafer. Upstream utilization is ~50%. If overcapacity takes until 2028+ to resolve, revenue stays depressed and the stock sits dead money.
Chinese MOCVD competition (AMEC). AMEC broke the Aixtron-Veeco LED duopoly and now holds >70% of Chinese LED MOCVD. China has a discreet 50% domestic sourcing mandate for wafer fab equipment. AMEC sources 80% of components domestically. If AMEC achieves GaN/SiC MOCVD parity, it could replicate its LED market capture in power -- at least within China (which is 60% of Aixtron's revenue geography).
Extended equipment cycle trough. FY2026 is guided down another 7%. If 2027 also disappoints, the market will grind the stock lower. Consensus expects EUR 544M revenue in 2026 (-2% YoY) before recovery in 2027 (+22%). Any miss on that trajectory is painful.
EV adoption slowdown delays SiC recovery. Temporary, but could stretch the trough by another 12-18 months.
China export control risk. MOCVD is not currently targeted by US/EU BIS export controls (those focus on sub-16nm logic/memory), but future restrictions on defense-grade GaN are possible.
Margin compression is real. EBIT margins have fallen from 25% (FY2023 peak) to 18% (FY2025), with FY2026 guided at 16-19%. R&D spending is sticky at 15-17% of revenue.

Nvidia 800V HVDC Supply Chain Thesis -- The Unique Angle

This is the piece most analysts are still missing.

The physics are unforgiving. When rack-level power consumption reaches 600kW and individual GPUs consume 3,600W, traditional power delivery infrastructure fails. Nvidia's solution is an industry-wide transition to 800V high-voltage DC (HVDC) architecture -- converting grid power directly to 800V at the facility perimeter, then stepping down through intermediate stages to point-of-load voltages. This architecture delivers 5% end-to-end efficiency improvement, 45% copper reduction, and enables megawatt-scale racks.

GaN is irreplaceable in this stack. While SiC handles the high-voltage AC-to-800V conversion, GaN's superior Baliga Figure of Merit (2,000x silicon) makes it essential for the 800V-to-48V and 48V-to-point-of-load stages where switching frequency and power density matter most. Traditional silicon MOSFETs simply cannot achieve the required efficiency at MHz switching frequencies.

Aixtron's technological moat in GaN is even deeper than the headline 77% MOCVD number suggests. The company commands approximately 90% market share at 200mm wafer sizes specifically for GaN power. The G10-GaN platform delivers a 25% cost-per-wafer advantage versus all competitors through its proprietary Planetary batch reactor architecture. This is not incremental -- it is a structural advantage that compounds with production scale.

Competitive Metric	Aixtron G10-GaN	Competitors
Wafer uniformity	3x improvement vs. prior gen	Standard
Equipment uptime	>=90%	Lower
Cost per wafer	25% below all competitors	Baseline
Cleanroom footprint	50% reduction	Standard
In-situ cleaning	Fully automated	Manual offline

The technological differentiation extends beyond hardware specifications. Aixtron's five-flow injector system provides individual control of thickness and composition uniformity, while Topside Temperature Control enables closed-loop individual wafer temperature management. These capabilities are essential for the defect densities required in power semiconductor manufacturing.

Veeco, Aixtron's primary Western competitor, has struggled to maintain relevance in GaN power. After losing LED market share to Chinese competitor AMEC, Veeco pivoted toward specialty applications. In GaN power specifically, Veeco's market presence is minimal. At Veeco's own 2024 investor conference, management sized Infineon's 300mm GaN opportunity at over 100 reactors -- a tacit acknowledgment that Aixtron, not Veeco, will capture most of this demand.

The full power delivery architecture -- where GaN and SiC fit:

In Nvidia's 800V HVDC architecture, power conversion happens in multiple stages from grid to GPU:

Grid AC → [SiC: AC-to-800V DC rectification] → 800V DC distribution → 
[GaN: 800V-to-48V conversion] → 48V distribution to rack → 
[GaN: 48V-to-point-of-load conversion] → GPU/accelerator

SiC handles the high-voltage, high-power first stage (AC-to-800V). GaN handles the subsequent stages where switching frequency and power density matter most. A single megawatt-scale rack could require dozens of GaN power conversion modules. Across a hyperscale data center with thousands of racks, the GaN content adds up to billions of dollars in devices -- and hundreds of millions in MOCVD tool demand.

The timing alignment is critical. Commercial 800V HVDC deployments are expected around 2027, coinciding precisely with Nvidia's Rubin Ultra/Kyber rack-scale systems. Aixtron's 300mm GaN Hyperion platform is in pilot production, with volume shipments expected 2026-2027 -- perfectly synchronized with this demand cycle.

The investment case crystallizes around a simple observation: you cannot build megawatt-scale AI infrastructure without GaN power semiconductors, and you cannot manufacture GaN power semiconductors at scale without Aixtron's MOCVD equipment. This is not derivative exposure to AI -- it is direct exposure to the most acute physical constraint facing AI infrastructure deployment.

Analysts estimate this GaN power ramp for AI could add EUR 400M+ in incremental tool revenue by 2029. For context, Aixtron's entire FY2025 revenue was EUR 557M. If even half of that incremental GaN demand materializes, it fundamentally changes the earnings trajectory.

GaN Market Inflection -- The Numbers

The GaN power device market is no longer a story. It reached USD 355M in 2024 according to Yole Group, representing a tenfold expansion from 2020. The projections:

Segment	2024	2030E	CAGR
GaN power devices (total)	$355M	$2.9-4.4B	42-49%
Data center + telecom segment	Subset	Growing fastest	53%
Consumer (chargers)	Largest today	Maturing	Decelerating
Automotive (OBC, DC-DC)	Emerging	Significant	High

Aixtron's revenue exposure to this market expansion is substantial. Power electronics (SiC + GaN combined) represented 55-57% of equipment revenue in FY2024-2025. In Q1 2023, GaN orders alone comprised over one-third of total equipment order intake. Management has explicitly stated that AI "has the potential to become the largest single application for GaN power semiconductors."

Current adoption remains early. This is not a mature market where Aixtron must fight for share -- it is a nascent market where technological leadership translates directly into order capture. The company's ~90% share at 200mm and first-mover position at 300mm give it the pole position.

Valuation Disconnect

Aixtron trades at a 40-70% discount to semiconductor equipment peers despite its dominant market position. Current metrics:

P/E (TTM): 23-30x versus semi equipment peer median of ~29x
P/S multiple: Contracted 51% since start of 2024
Market cap: ~EUR 1.8-2.6B range during 2025-2026 (varies with stock price at EUR 17-23)

Kerrisdale Capital, which disclosed a long position in November 2025, argues Aixtron is "the most underappreciated AI beneficiary in the market." Their analysis suggests applying peer-group 33x P/E to projected 2028 EPS of ~EUR 2.00 implies 270%+ upside. BofA upgraded to Buy (from Underperform), raising its target to EUR 25.10 and calling it their "SMID top pick for 2026," citing AI reaching ~50% of sales by 2027. Barclays upgraded to Overweight with a EUR 20 target and an upside scenario of EUR 30.

The near-term financial picture shows cyclical weakness. FY2025 guidance was narrowed to EUR 530-565M revenue, down from FY2024's EUR 633M. But free cash flow improved dramatically to EUR 182M from negative EUR 72M, demonstrating cost discipline through the downcycle.

Business

Business Model

Aixtron is a capital goods manufacturer. Revenue is primarily one-time equipment sales (80%) plus recurring aftermarket service, spare parts, and consumables (20%). Equipment orders are lumpy, negotiated, and tied to customer capex cycles. Revenue is recognized in two stages: on delivery and on final installation/acceptance.

Revenue model: Project-based capital equipment sales (~80%) + spares, service, upgrades (~20%). Blended ASP estimated at EUR 3.5-5.5M per system. The service/aftermarket business is growing with the installed base, offers higher margins than equipment, and provides modest cyclical dampening. Think of it like the razor/blade model -- equipment sales are the razor, and spares, consumables, and service contracts are the blades. Except in this case, the razor costs EUR 5M and the customer cannot easily switch to a competitor's blades because the process recipes are reactor-specific.

How the economics work for customers: A single Aixtron G10-GaN system can process 15 x 200mm wafers per run, with runs taking 3-5 hours for GaN HEMT structures. That translates to 30-90 wafers per day per tool. At several thousand dollars per epiwafer (depending on the device structure), a single MOCVD tool can generate tens of millions of dollars in annual production value for the customer. The EUR 5-10M tool price pays for itself within the first year of production -- which is why customers care far more about yield, uniformity, and uptime than they do about the sticker price.

Margin structure: Gross margins run 40-44% through the cycle, EBIT margins 16-25%. R&D intensity is maintained at ~15% of revenue regardless of cycle position, which means opex is sticky. This creates high operating leverage -- powerful on the upside, painful during downturns. The cost base has a high fixed component: R&D runs at EUR 80-95M/year whether revenue is EUR 450M or EUR 630M. SG&A is similarly sticky. So when revenue drops 12% (as in FY2025), EBIT drops nearly 24%. But the flipside is what makes this stock interesting as a cycle play: when revenue recovers 22% (as consensus expects for FY2027), EBIT should expand by 40-50%.

How the Economics Work -- Unit Economics Detail

The economics of Aixtron's business deserve a closer look because they explain why customers tolerate high ASPs and why the business generates attractive returns even in downturns.

Equipment unit economics (illustrative for G10-GaN):

Tool ASP: EUR 5-10M per cluster system
Annual output: 30-90 wafers/day x 300 production days = 9,000-27,000 wafers/year per tool
Epiwafer value: several thousand USD per wafer (varies by device structure)
Annual production value per tool: potentially $30-100M+
Tool payback period: typically <12 months from first wafer out

This payback math is why customers care about yield, uniformity, and uptime orders of magnitude more than they care about a 10% price premium over a competitor. A 1% improvement in wafer uniformity can mean millions in additional downstream revenue for the customer. A 5% improvement in tool uptime means 5% more production. The Aixtron G10-GaN's 25% cost-per-wafer advantage versus competitors is not just a selling point -- it translates directly into customer profitability at scale.

Aftermarket unit economics:

Installed base: ~77% of all MOCVD tools globally run Aixtron equipment
Service/spares revenue: EUR 112M/year (FY2025), growing with the installed base
Service margins: higher than equipment margins (likely 50%+ gross)
Razor/blade dynamic: every equipment sale creates 10-20 years of aftermarket revenue
The installed base is a compounding asset that grows with every equipment cycle and provides revenue stability through downturns

R&D as competitive investment:

EUR 81-95M/year regardless of revenue cycle (15-17% of revenue)
Innovation Center: EUR 100M one-time investment for 300mm capability
306 patent families (up from 265 in FY2024; 22 new applications in 2025)
R&D compounding: each generation of tools builds on the accumulated know-how from prior generations. This creates a widening gap between Aixtron and potential new entrants.
A new competitor starting from scratch would need to spend EUR 90M+/year for potentially 5-10 years before generating meaningful revenue. The cumulative R&D investment barrier is measured in billions.

Revenue Mix (FY2025)

By application (% of equipment revenue):

Power Electronics (GaN + SiC) -- 57% (EUR 254.9M). MOCVD/CVD for EV inverters, fast chargers, data center PSUs, renewables.
Optoelectronics (Lasers, 3D Sensing) -- 23% (EUR 101.4M). AI data center lasers (InP), VCSELs for 3D sensing. Surged in FY2025 -- the fastest-growing segment.
LED / Micro LED -- 15% (EUR 65.4M). Specialty LEDs, Micro LED displays. Declined significantly as high R&D investment cycle ended.
Service / spare parts / consumables -- 20% of total revenue (EUR 112.0M). Grew slightly (+1% YoY) even as equipment revenue fell 15%.

By type:

Equipment revenues: EUR 444.6M (80%)
Service / spare parts: EUR 112.0M (20%)

Geographic mix (FY2025):

Region	Revenue (EUR M)	% of Total	YoY Change
Asia	332.6	60%	-20%
Americas	110.4	20%	+66%
Europe	113.6	20%	-24%
Total	556.6	100%	-12%

Asia dominates, driven by China (optoelectronics, SiC). The Americas surged 66% in FY2025, likely reflecting US CHIPS Act-related investments.

Why It Matters -- End Markets in Detail

The world is shifting from traditional silicon semiconductors to compound semiconductors for applications where silicon simply cannot perform. These materials handle higher voltages, switch faster, operate at higher temperatures, and emit/detect light -- capabilities that silicon lacks. Every EV inverter, every data center laser, every fast charger, every LiDAR sensor depends on compound semiconductors. And every one of those chips starts life on an Aixtron deposition system.

End-use applications:

Power electronics (GaN) -- Fast chargers (the GaN charger market exploded in 2022-2024), data center power supply units, EV on-board chargers, motor controllers, renewable energy microinverters, AI server power supplies. GaN devices are displacing silicon in low-to-medium voltage applications (up to ~650V today, pushing higher). The killer app emerging now is AI data center power conversion, where GaN's MHz switching frequency enables the power density required for megawatt-scale racks.
Power electronics (SiC) -- EV traction inverters (this is where 80% of SiC content goes today), DC fast-charging stations, industrial high-voltage converters, rail power systems, wind and solar inverters. SiC is the material of choice for high-voltage, high-power applications (900V+). The 800V EV battery architecture (Tesla Model 3/Y, Porsche Taycan, Hyundai Ioniq 5) structurally increases SiC content per vehicle.
Optical data communications -- InP and GaAs lasers for fiber-optic data transmission in AI data centers, telecom networks, and 5G/6G infrastructure. This is the fastest-growing segment right now. AI training clusters require moving massive amounts of data at the speed of light through fiber optics -- every 800G+ transceiver needs compound semiconductor lasers grown by MOCVD.
3D sensing -- VCSEL arrays (GaAs-based) for facial recognition (every iPhone uses them), LiDAR for autonomous vehicles and robotics, gesture control for AR/VR. Growing in automotive and industrial applications.
LED / Micro LED -- Specialty LEDs (ROY colors), UV LEDs for sterilization, and next-gen Micro LED displays for AR/VR headsets, automotive displays, and potentially large-format TVs. The Micro LED opportunity is real but timing is uncertain -- mass transfer yield remains the bottleneck (achieving six-nines, 99.9999%, at scale is still unsolved).
RF / 5G / Defense -- GaN amplifiers for 5G base stations, AESA radar for military applications, electronic warfare systems. The defense GaN market is projected to reach $14.5B by 2030, driven by surging global defense spending.

Aixtron sells the tool that makes all of this possible. Without epitaxial deposition, none of these devices can be manufactured. The company sits at the very top of the compound semiconductor value chain -- the classic picks-and-shovels play.

MOCVD Technology -- How It Actually Works

The Problem Being Solved

Silicon dominates >95% of the semiconductor market. But silicon has fundamental physical limitations rooted in atomic physics -- not engineering failures that better process nodes can fix:

Band gap too narrow (1.1 eV): Silicon cannot efficiently emit light (indirect band gap) and breaks down at relatively low electric fields.
Low critical electric field (~0.3 MV/cm): Limits voltage blocking per unit thickness.
Moderate thermal conductivity (150 W/mK): Constrains power density before thermal runaway.
Low electron saturation velocity: Limits high-frequency performance.

Compound semiconductors solve all of these:

Property	Silicon	SiC (4H)	GaN	GaAs	InP
Band gap (eV)	1.1	3.3	3.4	1.4	1.3
Critical electric field (MV/cm)	0.3	2.8	3.3	0.4	0.5
Electron mobility (cm2/Vs)	1,400	900	2,000 (2DEG)	8,500	5,400
Thermal conductivity (W/mK)	150	490	130	46	68
Saturation velocity (10^7 cm/s)	1.0	2.0	2.5	1.0	1.0
Band gap type	Indirect	Indirect	Direct	Direct	Direct

SiC blocks 10x the voltage of silicon per unit thickness. GaN switches 10x faster. GaAs and InP emit light efficiently -- silicon physically cannot. These are order-of-magnitude advantages, not incremental engineering improvements. They are rooted in the fundamental physics of atomic bonding and crystal structure.

Crystal Structures

Compound semiconductors crystallize in two main structures that determine their electronic properties:

Zinc blende (cubic): GaAs, InP, GaP. Atoms arranged in two interpenetrating face-centered cubic lattices, one for each element. The tetrahedral bonding geometry is identical to diamond (silicon's structure), which is why these materials share some processing compatibility with silicon. Zinc blende materials are the workhorses of optoelectronics and RF applications.
Wurtzite (hexagonal): GaN, AlN, SiC. Atoms in a hexagonal close-packed arrangement with two interpenetrating sublattices. The asymmetry of the wurtzite structure creates strong internal piezoelectric and spontaneous polarization fields. This is critically important for GaN HEMT device operation -- these polarization fields induce a Two-Dimensional Electron Gas (2DEG) at the AlGaN/GaN interface without any intentional doping. The 2DEG is a sheet of high-mobility electrons confined to a few nanometers at the interface, enabling the high-frequency, high-power switching that makes GaN so valuable.

Band Gap Engineering

The band gap is the single most important parameter determining what a semiconductor can do:

Wider band gap -> higher voltage blocking (critical for power electronics -- a wider gap means the device can withstand higher electric fields before breakdown)
Direct band gap -> efficient light emission (critical for LEDs and lasers -- direct gap materials can emit photons when electrons recombine across the gap; indirect gap materials like silicon cannot do this efficiently)
Tunable band gap -> engineered wavelength (by alloying different III-V materials: InxGa1-xN covers the UV to green spectrum; InGaAsP covers the telecom wavelengths at 1310nm and 1550nm)

The III-V compound semiconductor system offers an extraordinary design space: by mixing different alloys (InGaAlN, InGaAsP), engineers can continuously tune the band gap from 0.7 eV (InN, infrared) to 6.2 eV (AlN, deep UV). This tunability is physically impossible with silicon, which has a fixed 1.1 eV indirect band gap.

This is what makes MOCVD so valuable -- it is the only manufacturing-scale technique that can precisely control the composition of these multi-element alloys layer by layer, with nanometer-scale thickness control, across an entire 200mm wafer.

For investors, the band gap engineering capability translates directly into market opportunity. Every new material combination that device engineers dream up requires an MOCVD reactor to grow it. As the compound semiconductor design space expands (new alloys, new device architectures, new applications), the demand for MOCVD equipment expands with it. Aixtron does not need any single application to succeed -- it benefits from the entire design space being explored.

What disappears without compound semiconductors:

No blue/white LEDs (the entire LED lighting revolution depends on GaN)
No efficient EV inverters (silicon IGBTs waste 2-3x more energy than SiC MOSFETs)
No 5G base stations (GaN RF amplifiers enable the required power density)
No fiber-optic communications (InP and GaAs lasers carry all internet traffic)
No modern phased-array radar (GaN enables AESA systems in defense)
No compact fast chargers (GaN power ICs enable efficient charging)

Why GaN and SiC Beat Silicon for Power -- Baliga's Figure of Merit

BFOM = epsilon x mu x E_c^3 (permittivity x mobility x critical field cubed).

GaN BFOM: ~2,000x silicon
SiC BFOM: ~500x silicon

A GaN or SiC device can be 10x smaller than silicon for the same voltage rating. Smaller device means lower capacitance, faster switching, and lower switching losses. The result: 50-80% reduction in power conversion losses, enabling smaller, lighter, more efficient systems.

The MOCVD Breakthrough

Metalorganic Chemical Vapor Deposition (MOCVD) -- also called MOVPE -- was pioneered by Harold Manasevit at Rockwell International in 1968. The breakthrough: using metalorganic compounds as gas-phase precursors that decompose on a hot substrate to deposit crystalline thin films.

Before MOCVD, compound semiconductor epitaxy used methods that could not scale:

Method	Era	Problem
Liquid Phase Epitaxy (LPE)	1960s	Poor uniformity, no quantum well capability; still used for some infrared detectors
Molecular Beam Epitaxy (MBE)	1970s	Excellent interface control but single wafer, hours per run, ultra-high vacuum (~10^-10 Torr)
Hydride Vapor Phase Epitaxy (HVPE)	1970s	Fast growth but poor thin-layer control; now used primarily for bulk GaN boule growth

MOCVD's key advantage over MBE: it works at moderate pressures (50-760 Torr vs 10^-10 Torr for MBE), enabling batch processing of multiple wafers simultaneously. This made compound semiconductors manufacturable at scale for the first time.

Aixtron was founded in 1983 to commercialize MOCVD technology. They licensed the Planetary Reactor design from Philips in 1989, which became the foundation of their equipment portfolio.

Think of MOCVD like spray-painting, except the "paint" is vaporized metal-organic molecules, the "surface" is a red-hot crystal spinning at high speed, and the thickness of each coat must be accurate to the width of a few atoms.

Epitaxy and the Lattice Matching Problem

"Epitaxy" means growing a crystalline film whose atomic structure is aligned with the underlying substrate crystal. For high-quality devices, the epitaxial film must be nearly perfect -- a single misplaced atom per billion is enough to degrade device performance.

The central challenge is lattice matching. Mismatch creates strain; large mismatch generates threading dislocations -- line defects that propagate through the film and destroy device performance.

System	Lattice Mismatch	Difficulty
GaAs on GaAs	0% (homoepitaxy)	Easy
AlGaAs on GaAs	<0.1%	Easy
SiC on SiC	0% (homoepitaxy)	Moderate (high temp)
GaN on SiC	~3.5%	Hard
GaN on sapphire	~16%	Very hard
GaN on Si(111)	~17% + thermal mismatch	Extremely hard

That GaN epitaxy works at all on sapphire (16% mismatch) is remarkable -- it required decades of nucleation layer innovation to reduce dislocation densities from ~10^10 to ~10^8 cm^-2.

The MOCVD Process Flow

[1]  SUBSTRATE LOADING     -> Wafers loaded onto susceptor in clean environment
[2]  REACTOR PURGE         -> Establish carrier gas (H2 or N2) flow
[3]  HEAT-UP               -> Ramp susceptor to growth temperature (500-1650C)
[4]  NUCLEATION LAYER      -> Low-temperature buffer layer (450-600C for GaN)
[5]  ANNEAL/RECRYSTALLIZE  -> Ramp to high temp; islands coalesce, improve crystal quality
[6]  BUFFER LAYER GROWTH   -> High-temperature growth; lateral coalescence reduces dislocations
[7]  ACTIVE LAYER GROWTH   -> Quantum wells, channel layers, barrier layers (device-specific)
[8]  DOPING LAYERS         -> N-type (silane) and p-type (Cp2Mg) doping
[9]  CAPPING & COOL-DOWN   -> Controlled cooling to manage thermal stress
[10] UNLOAD & CHARACTERIZE -> PL mapping, XRD, bow, surface inspection

Critical engineering challenges:

Susceptor design determines temperature uniformity. Aixtron's Planetary Reactor rotates individual wafers on a rotating platter -- a "planet and sun" motion that averages out temperature and gas flow non-uniformities.
Gas mixing and metering: tiny variations in gas flow translate directly to composition variations in the film.
Crystal quality depends on temperature, pressure, gas ratios, and substrate condition. Defects propagate through every subsequent layer.
Some device structures require 50+ distinct layers with abrupt interfaces. Switching gas compositions rapidly and cleanly is a major challenge.
In-situ monitoring (measuring the wafer during growth) allows real-time feedback and is a key Aixtron differentiator.

Metalorganic Precursors

Precursor	Formula	Role	State	Notes
TMGa (Trimethylgallium)	Ga(CH3)3	Gallium source	Liquid bubbler	~$300-600/kg
TMAl (Trimethylaluminum)	Al(CH3)3	Aluminum source	Liquid bubbler	Extremely pyrophoric
TMIn (Trimethylindium)	In(CH3)3	Indium source	Solid sublimer	>$1,000/kg
NH3 (Ammonia)	NH3	Nitrogen source	Compressed gas	Cheap but large volumes
AsH3 (Arsine)	AsH3	Arsenic source	Compressed gas	Highly toxic
PH3 (Phosphine)	PH3	Phosphorus source	Compressed gas	Highly toxic
SiH4 (Silane)	SiH4	N-type dopant	Dilute gas	Silicon donor atoms
Cp2Mg	Mg(C5H5)2	P-type dopant	Solid sublimer	Mg acceptors in GaN

Pyrolysis mechanism: Ga(CH3)3 -> Ga(CH3)2 + CH3 -> GaCH3 + CH3 -> Ga + CH3. Free gallium atoms adsorb on the surface and incorporate into the growing crystal. Methyl radicals recombine as methane and are exhausted.

Reactor Architectures -- Planetary vs. Showerhead

Both architectures are Aixtron's, serving different purposes. This dual-architecture strategy is unique in the industry.

Planetary Reactor (production flagship):

Large main susceptor ("platter") rotates around central axis
Multiple satellite disks, each holding one wafer, rotate individually via Gas Foil Rotation (GFR) -- contactless, frictionless spinning from gas flowing through channels beneath each disk
Precursors injected from center, flow radially outward in horizontal laminar flow
Dual rotation (orbital + spin) produces exceptional uniformity -- analogous to planets orbiting the sun while spinning on their axes
Pros: High throughput (5-15 x 200mm per run), excellent uniformity, high precursor utilization
Cons: Larger footprint, more complex flow dynamics

Close Coupled Showerhead (CCS) (R&D + 300mm development):

Water-cooled plate with thousands of holes positioned millimeters above the susceptor
Group-III and Group-V precursors through separate holes to prevent pre-mixing
Vertical stagnation-point flow creates uniform boundary layer
Pros: Inherently uniform gas distribution, scalable to different wafer sizes, lower gas-phase residence time
Cons: Lower batch throughput, showerhead deposits require cleaning

Strategic significance: Aixtron's 300mm GaN-on-Si development uses single-wafer showerhead architecture (not planetary) because 300mm silicon fabs are built around single-wafer processing flows. This is a deliberate architectural pivot for future growth.

Key Technical Metrics

Metric	State of the Art (2025)	Why It Matters
Thickness uniformity	<0.5-1.0% sigma	Quantum well energy, HEMT threshold voltage
Composition uniformity	<0.5-1.0% sigma	LED wavelength binning yield
TDD (GaN/sapphire)	~1-5 x 10^8 cm^-2	Down from 10^10 in the 1990s
TDD (GaN homoepitaxy)	~10^4-10^6 cm^-2	Required for laser diodes, vertical power
Throughput	15 x 200mm per run (G10-GaN)	Determines cost per epiwafer
Run duration	2-8 hours	GaN LED ~4h; HEMT ~3-5h; SiC ~2-4h
Temperature uniformity	+/-0.5-1.0C across susceptor	1C variation -> ~2nm wavelength shift in InGaN
Run-to-run reproducibility	<0.5-1.0% sigma	Manufacturing consistency
Precursor utilization	10-40% (group-III)	Cost control for expensive metalorganics
Effective wafers/day	30-90 x 200mm	Production capacity per tool

Historical improvement -- roughly one order of magnitude per decade:

Era	Wafer Size	TDD (GaN/sapphire)	Wafers/Run	Uniformity
1990s	2"	~10^10 cm^-2	1-3	~5-10%
2000s	2"-4"	~10^9 cm^-2	6-12 x 2"	~2-5%
2010s	4"-6"	~5 x 10^8 cm^-2	8-14 x 4"	~1-2%
2020s	6"-8"	~1-5 x 10^8 cm^-2	5-15 x 8"	<1%

This trajectory is driven by reactor engineering, not just materials science -- which is why the equipment maker captures value. Each decade of improvement in defect density, throughput, and uniformity represents decades of accumulated empirical know-how that new entrants cannot leapfrog. The MOCVD equipment business rewards continuous investment in incremental improvement -- exactly the kind of long-duration competitive advantage that compounds over time.

The implication for investors: Aixtron's current technology lead represents 40 years of continuous improvement. A competitor starting today, even with unlimited capital, would need at least 5-10 years to reach parity in advanced applications. This timeline is what gives Aixtron its competitive moat -- not any single patent or product feature, but the accumulated depth of process knowledge that only comes from four decades of customer-facing reactor engineering.

Process Failure Modes

Failure Mode	Cause	Impact	Mitigation
Threading dislocations	Lattice mismatch at interface	Non-radiative recombination, leakage	Optimized nucleation, buffer engineering
Composition non-uniformity	Temperature or flow variations	LED wavelength shift, HEMT threshold shift	Dual rotation, multi-zone heating
Wafer cracking	Thermal expansion mismatch (GaN-on-Si)	Catastrophic yield loss	AlN interlayers, strain-compensating superlattices
Parasitic gas-phase reactions	TMAl:NH3 adduct formation	Precursor waste, particles	Separate injection, reduced pressure
Carbon contamination	Incomplete precursor decomposition	Unintentional doping	Optimized temperature, V/III ratio
Memory effects	Mg on reactor walls	Cross-contamination between runs	Reactor conditioning, dedicated chambers

Key Terminology

Term	Definition
MOCVD/MOVPE	Metalorganic Chemical Vapor Deposition / Metalorganic Vapor Phase Epitaxy
Epitaxy / Epiwafer	Growth of crystalline film on substrate; the finished wafer with epitaxial layers
HEMT	High Electron Mobility Transistor -- GaN transistor using 2DEG for high-frequency switching
2DEG	Two-Dimensional Electron Gas -- sheet of electrons at AlGaN/GaN interface, enabling HEMTs
MQW	Multiple Quantum Well -- ultra-thin alternating layers (2-3nm) for LEDs and lasers
TDD	Threading Dislocation Density -- crystal defects per cm2, lower is better
V/III ratio	Ratio of group-V to group-III precursor flow; typically 1,000-10,000:1 for nitrides
Susceptor	Heated platform on which wafers sit in the reactor
In-situ	Measurement performed during growth, inside the reactor
Wafer bow	Curvature of wafer due to stress; must be <50um for lithographic processing

Product Portfolio

G10-GaN -- GaN Power Electronics MOCVD

Grows GaN epitaxial layers on silicon or SiC substrates for power switching devices (HEMTs, diodes) used in chargers, data center power, automotive, and industrial applications.

3-chamber cluster tool, 15 x 200mm total capacity
50% smaller footprint than prior generation
3x uniformity improvement over competition (company claim)
Particle density ~0.1/cm2 (production-grade cleanliness)
=90% equipment uptime
25% lower epitaxy cost than any competing platform (company claim)
Five-flow injector system with individual control of thickness and composition uniformity
Topside Temperature Control enables closed-loop individual wafer temperature management
Market share: 70-90% of GaN power MOCVD globally (Kerrisdale estimate; ~90% at 200mm)
ASP estimate: EUR 5-10M per cluster system

Also in installed base: AIX G5+ C (predecessor, still widely deployed)

G10-SiC -- SiC Epitaxy

Grows SiC epitaxial layers on SiC substrates at 1,500-1,650C for power MOSFETs and diodes used in EV inverters, onboard chargers, industrial motor drives, and grid infrastructure.

Warm-wall planetary reactor design for SiC-CVD
Dual wafer size: 150mm and 200mm capable
Auto Feed-Forward individual on-wafer temperature control
In-situ UV pyrometry (LaytTec InSide P400)
Automated in-situ cleaning
100th system shipped Q3 2025 (just 3 years after launch)
Market share: ~35-40% (targeting 50% by 2029); winning 50-60% of new 200mm business
ASP estimate: EUR 3-6M per tool

G10-AsP -- Optoelectronics & RF

Grows GaAs, InP, and GaN-on-SiC epitaxial layers for VCSELs, datacom/telecom lasers, 5G RF amplifiers, and photonic ICs.

Next-gen cluster for arsenide/phosphide optoelectronics
Growing rapidly on AI-driven optical interconnect demand (800G+ transceivers)
Described as "the leading solution" for AI data center laser applications
Secured "repeat orders from blue-chip customers" in FY2025
ASP estimate: EUR 2-4M per tool
Also in portfolio: AIX 2800G4-TM -- planetary reactor, 15x4" or 8x6", all-graphite chamber, triple gas injection. Workhorse for VCSELs and RF.

Hyperion (300mm GaN) -- Next Generation

Single-wafer showerhead architecture (not planetary) for 300mm Si fab compatibility
First prototype systems shipped to pilot lines in 2025
Targets integration into existing 300mm CMOS fabs -- massive TAM expansion
Volume production expected 2027-2028
Veeco competing with Propel 300; customer commitments will be the differentiator
Imec's October 2025 300mm GaN Power Electronics Program featured Aixtron's Hyperion as foundational equipment
This is Aixtron's long-term bet on GaN power going mainstream.

R&D / Close Coupled Showerhead (CCS) Systems

Smaller-scale systems for research institutions and universities
Used for developing next-generation devices and materials (2D nanostructures, memristors for neuromorphic computing)
Lower ASPs but strategically important: customers who develop processes on CCS tools often scale to Planetary Reactor tools for volume production
Provides early visibility into emerging applications -- the CCS installed base at research institutions is a leading indicator of future production tool demand
The Innovation Center in Herzogenrath uses CCS architecture for its 300mm GaN development program

LED MOCVD

CRIUS II-XL (CCS architecture, up to 19x4") and AIX R6 (next-gen, 12x6") for GaN-on-sapphire
Lost significant China market share to AMEC post-2017; retains position outside China
Micro LED for AR/VR headsets and automotive displays is real optionality but mass production is years away
LED is now 15% of systems revenue and declining; this is a mature, commoditizing market where AMEC has the cost advantage in China

Service / Aftermarket (EUR 112.0M, 20% of revenue)

Spare parts, consumables, upgrades, service contracts
Growing with installed base (77% MOCVD share = massive installed base)
Higher margins than equipment sales
More stable revenue stream that partially cushions cyclicality
Grew slightly (+1% YoY) in FY2025 even as equipment revenue fell 15%

Competitive Moat -- Multi-Layered

Aixtron's moat is one of the strongest in the semiconductor equipment industry. It rests on several reinforcing factors:

Switching costs (strongest moat). MOCVD tool qualification at customer fabs takes 6-18 months, costs millions, and creates reactor-specific recipe libraries. Customers develop proprietary process recipes on Aixtron tools. Switching requires re-qualifying everything from scratch -- a multi-million dollar, multi-year effort. This is why 77% market share has been stable for years.
Process IP (306 patent families, expiring 2026-2045). The Planetary Reactor architecture -- Gas Foil Rotation dual-rotation system -- delivers unmatched batch throughput and uniformity. 40+ years of reactor engineering, gas dynamics, and epitaxial growth know-how cannot be reverse-engineered from papers or patents alone.
Installed base network effect. 77% share means most process engineers worldwide are trained on Aixtron equipment. New hires at customer fabs already know the platform. New fabs default to the system their staff knows. This creates a self-reinforcing preference cycle.
Dual architecture. Planetary (production) + Close Coupled Showerhead (R&D, 300mm) -- unique capability to address both current production and future market transitions.
R&D intensity compounding. 15% of revenue invested in R&D continuously (EUR 81M in FY2025). The EUR 100M Innovation Center provides 300mm development capability that competitors lack.
Market structure protection. The MOCVD market (~$1-2B total) is too small to attract the silicon equipment giants (AMAT, Lam Research, TEL). They have $15-27B revenue bases -- MOCVD is a rounding error. This structural ceiling permanently limits the competitive field to three players.

Value Chain Position

Substrates -> Epitaxy Equipment -> Epiwafer Growth -> Device Fab -> Modules -> End Customer
(SiC, GaN,    [AIXTRON HERE]     Foundries/IDMs     Front/back    Power, RF,   EVs, data centers,
sapphire,                        grow crystal        end fab       optical      5G, defense,
Si(111))                         layers                                         consumer

Layer	Key Players	Revenue Pool	Gross Margins	Barrier to Entry
Substrates	Wolfspeed, Coherent, SICC, TanKeBlue, Sumitomo	~$1-2B (SiC)	20-40%	Very high
Epitaxy Equipment	Aixtron, Veeco, AMEC	~$1-2B	40-45%	Extremely high
Epiwafer Growth	IDMs + foundries (IQE, EpiWorld, Landmark)	~$3-5B	30-50%	High
Device Fabrication	Infineon, STMicro, onsemi, Wolfspeed, Rohm	~$5-10B	40-55%	Very high ($1-5B+ fab cost)
Modules & Systems	Infineon, Danfoss, Semikron, Coherent	~$10-20B	25-40%	Moderate

The epitaxy equipment layer captures the highest margins with the most defensible position. Aixtron sits at the classic "picks and shovels" position in the compound semiconductor revolution.

Key customers: Integrated Device Manufacturers (Infineon, STMicro, onsemi, Wolfspeed, Rohm, TI), epitaxial wafer foundries (IQE, WIN Semiconductors, VPEC), Chinese compound semi companies (Innoscience, Sanan Optoelectronics), and research institutions/universities worldwide. No single customer disclosed above 10% of revenue. Asia represents 60% of revenue (geographic concentration risk).

Innoscience -- Aixtron's largest GaN customer and a window into the China ecosystem:

Innoscience (Suzhou, China) is the world's largest GaN IDM with 31% device market share (Yole Group 2024). They run multiple 8-inch GaN production lines using Aixtron MOCVD tools. Sanan Optoelectronics is another major Chinese customer with significant compound semi capacity. These customers represent meaningful revenue for Aixtron today. The China GaN ecosystem is a paradox: simultaneously Aixtron's biggest growth market (demand growing faster in China than anywhere else) and its biggest competitive risk (AMEC and domestic sourcing mandates threaten to displace Aixtron tools over time). The most likely resolution is a gradual shift in revenue mix rather than a cliff -- China's share of Aixtron revenue declines from ~25-35% today to ~15-20% by 2030 as AMEC gains share in commodity applications, while absolute China revenue holds roughly flat because the overall market is growing fast enough to offset share loss.

Pricing power: Moderate-to-strong. Once a customer is qualified on an Aixtron tool, price sensitivity decreases dramatically -- switching to save 10% on tool cost makes no economic sense when requalification costs millions and takes 18 months.

Barriers to entry in detail:

Decades of process IP: MOCVD recipe development is highly empirical -- cannot be reverse-engineered from published literature.
Switching costs: Tool qualification takes 6-18 months, costs millions, and creates reactor-specific recipe libraries.
Chicken-and-egg problem: New entrants cannot get qualification without an installed base, cannot build installed base without qualification.
R&D intensity: EUR 90M+/year is table stakes. A new entrant would need years of spending before generating revenue.
Customer co-development: Joint development with Infineon, STMicro on next-gen processes builds relationships over decades.
Market concentration: Top 3 players hold ~80%+ share; top 10 CVD players hold ~96% of global revenue.

TAM & Market Sizing

Market	2024-2025	2030 Projected	CAGR
MOCVD equipment (global)	~$1.0-1.9B	~$2.4-2.8B	8-10%
SiC epitaxy equipment	~$1.3B	~$4.9B	~16%
Overall epitaxy equipment	~$5.7B	~$11.5B	~12%
GaN power devices	~$355M (2024)	~$2.9-4.4B	~42-49%
SiC power devices	~$3.4B (2024)	~$10.4B	~20%
AI optical interconnects	~$19.4B	~$36.0B	~20%
Semiconductor lasers (InP)	~$3.1B (2023)	~$5.2B (2029)	~8.7%
Micro LED panels	Early stage	~$2.2B (2030)	High double-digit

Aixtron's directly addressable market is the MOCVD + SiC CVD equipment market, which totaled roughly USD 1.05B in 2024 (USD 550M MOCVD + USD 500M SiC CVD).

Market share:

MOCVD for compound semiconductors: ~77% global (Yole Group 2024). Veeco is #2 at 8%, TNS at 5%, AMEC at 4%.
SiC CVD systems: ~27-40% (#2 behind ASM International at 32%, ahead of TEL at 16%).

Competitive Landscape

Company	Ticker	MOCVD Share	Key Strength	Threat to Aixtron
Veeco	VECO	~25-28%	Broader portfolio (MBE + MOCVD); micro-LED; Epiluvac SiC entry	Low-moderate (weaker in GaN power)
AMEC	688012.SS	~20-25%	>70% China LED; government backing; 50% domestic sourcing mandate	High for China market
ASM International	ASM	-- (SiC CVD)	32% SiC CVD share; ALD market leader	Moderate in SiC only
Tokyo Electron	8035.T	-- (SiC CVD)	~16% SiC CVD; scale + customer relationships	Moderate in SiC only
TNS	4091.T	~5%	Gas supply + equipment bundling	Low

Veeco in detail: Veeco is the #2 Western MOCVD player with a broader product portfolio (MBE, ion beam, wet processing alongside MOCVD). After losing LED market share to AMEC in China, Veeco pivoted toward specialty applications. In GaN power specifically, Veeco's market presence is minimal. At Veeco's own 2024 investor conference, management sized Infineon's 300mm GaN opportunity at over 100 reactors -- a tacit acknowledgment that Aixtron, not Veeco, will capture most of this demand. Veeco is competing directly with Aixtron in the 300mm space with its Propel 300 system and has secured at least one order from a leading GaN-on-Si IDM. Worth watching, but Aixtron's installed base advantage and qualification head start make it the favorite.

AMEC in detail: The single most important competitive risk. AMEC broke the Aixtron-Veeco LED MOCVD duopoly after 2017 and now holds >70% of Chinese LED MOCVD. China's 50% domestic sourcing mandate is policy-level structural support. AMEC sources 80% of MOCVD components domestically. The technology gap in SiC/GaN power MOCVD is real but narrowing. Chinese tools are competitive in commodity LED but not yet in high-performance power or optoelectronics. Several other Chinese startups (Beijing NAURA, JSG, TOPEC, Agnitron) are working to qualify MOCVD/CVD systems with reportedly 50-75% price discounts. The technology gap buys years of protection, but the China market faces structural long-term pressure.

ASM International in SiC: Market leader in SiC CVD with ~32% share (ahead of Aixtron at ~27-40%, then TEL at ~16%). They leveraged existing silicon equipment customer relationships to build their SiC position. ASM is primarily a silicon equipment company ($2.3B revenue, ALD market leader), so SiC is a small part of their revenue -- but they are a serious competitor in that specific segment. The difference: ASM does not compete in MOCVD at all. Aixtron's MOCVD monopoly is entirely separate from the SiC CVD competition. In SiC CVD, Aixtron is the challenger, not the incumbent.

The SiC CVD competitive dynamic: Unlike MOCVD where Aixtron is dominant, SiC CVD is a three-way fight (ASM 32%, Aixtron 27-40%, TEL 16%). Aixtron is gaining share -- winning 50-60% of new 200mm business with the G10-SiC -- but it is not yet the clear leader. The 200mm transition is the key battleground: whoever establishes the most installed base at 200mm will likely hold share for the next 5-7 years. Aixtron's milestone of 100 G10-SiC systems in 3 years is impressive and suggests momentum is on their side.

Why AMAT, Lam, and TEL don't enter MOCVD: The MOCVD market (~$1-2B) is too small relative to their $15-27B revenue bases. Applied Materials alone generates more revenue in a single quarter than the entire MOCVD market in a year. For a company with $27B in revenue, investing $100M in R&D to chase a $200M revenue opportunity (their addressable share of a $1B MOCVD market) makes no economic sense. The return on R&D investment is better deployed in lithography, etch, or deposition for silicon. This structural ceiling permanently protects the MOCVD incumbents. The compound semi equipment market is big enough to support three players profitably but too small to attract the silicon giants. Aixtron benefits from this market structure dynamic indefinitely.

The Veeco Competitive Dynamic -- 300mm Battleground

Veeco deserves deeper analysis because it is the only Western competitor with a credible challenge to Aixtron in the 300mm GaN transition.

Current state: Veeco's Propel 300 is a single-wafer showerhead MOCVD system targeting 300mm GaN-on-Si. In November 2025, Veeco announced receiving a Propel 300 order from "a leading GaN-on-Si power semiconductor IDM" -- believed to be Infineon or Texas Instruments. This is a real competitive win.

Veeco's strengths:

Broader product portfolio (MBE, ion beam, wet processing + MOCVD)
Strong relationships in GaAs/InP optoelectronics
Epiluvac acquisition (2022) gives SiC CVD entry
US-based, which may provide advantage in CHIPS Act-funded fabs
The Propel 300 is a purpose-built 300mm tool, not an adaptation of an existing platform

Veeco's weaknesses:

Minimal installed base in GaN power (Aixtron has 70-90% of the existing tools)
Process engineers at customer fabs are trained on Aixtron equipment
Aixtron's batch Planetary architecture provides a cost-per-wafer advantage that single-wafer tools cannot match in 200mm production
Aixtron's Innovation Center gives it a development advantage (Veeco does not have an equivalent 300mm GaN cleanroom facility)

The 300mm competitive dynamic: In 300mm GaN, both companies are using single-wafer showerhead architectures (not Aixtron's traditional Planetary). This levels the playing field somewhat because Aixtron's batch throughput advantage does not apply at 300mm. The competition will be decided by:

Who gets qualified first at major customers (Aixtron has a head start with Hyperion already at pilot lines)
Whose process yields are better (data not yet publicly available)
Customer relationships and switching costs from 200mm tools (strongly favors Aixtron)
Fab integration and compatibility (Aixtron designed Hyperion specifically for 300mm Si fab flows)

My assessment: Aixtron is favored to win the majority of 300mm GaN business because of its installed base advantage, head start in qualifications, and the natural tendency for customers to stick with the platform they know. But Veeco will likely win a meaningful minority share (20-30%) because customers want a second source, and Veeco has demonstrated it can deliver competitive 300mm tools. A Veeco share of 20-30% would be consistent with historical MOCVD market dynamics where Aixtron leads but does not take 100%.

GaN Market Adoption -- From Promise to Production Reality

The GaN power story has been "two years away" for a decade. That is finally changing. The GaN power device market reached USD 355M in 2024 (Yole Group), representing a tenfold expansion from 2020. More importantly, the growth is accelerating -- driven by real production deployments, not lab demonstrations:

Consumer chargers: GaN chargers went from exotic to mainstream in 2022-2024. Anker, Apple (MagSafe), Samsung, and dozens of others now ship GaN chargers. This market is maturing but established the manufacturing base.
Data center PSUs: This is where the growth is shifting. TrendForce projects SiC and GaN combined will represent just 17% of data center power systems by 2026, but exceed 30% by 2030. The transition has barely started.
Automotive: GaN for on-board chargers (OBC) and DC-DC converters is emerging. The automotive qualification cycle is 3-5 years, so design wins happening now will drive production volume in 2027-2029.
AI-specific: Management has explicitly stated that AI data centers have the potential to become the largest single application for GaN power semiconductors by 2027.

The 300mm wafer transition amplifies the opportunity. 300mm GaN-on-Si can be processed in existing 300mm silicon fabs -- potentially opening the massive CMOS foundry capacity to GaN production. This is why the Hyperion system is such an important bet: it is not just about wafer size, it is about making GaN accessible to the silicon fab ecosystem.

The AMEC Threat -- Detailed Assessment

AMEC deserves its own section because it is the single most important competitive risk to the Aixtron thesis.

What happened in LED: AMEC entered the MOCVD market around 2013-2014, initially selling basic LED MOCVD tools at steep discounts. The Aixtron-Veeco LED duopoly dismissed the threat. By 2017, AMEC had captured meaningful share in Chinese LED fabs. Today AMEC holds >70% of Chinese LED MOCVD. Aixtron lost a major market segment in China within 5 years.

Why it happened: China's LED industry was the perfect storm for disruption. LED epitaxy has relatively relaxed process requirements compared to power or opto (wider tolerances, fewer layers, less sensitive to defect density). Chinese LED fabs were price-sensitive and willing to accept "good enough" quality at 50-75% lower tool cost. Government domestic sourcing mandates provided policy tailwinds. And AMEC benefited from sourcing 80% of components domestically.

Why power and opto are harder to crack: The requirements for power semiconductor epitaxy are fundamentally more demanding than LED. Power devices (SiC MOSFETs, GaN HEMTs) require:

Tighter thickness and composition uniformity (<0.5% vs 2-3% for LED)
Lower defect density (reliability requirements for automotive/industrial applications)
Higher temperature processing (1,500-1,650C for SiC)
More stringent particle control (power device yield sensitivity)
Longer qualification cycles (automotive = 3-5 years)

These requirements create a quality gap that Chinese tools have not yet bridged. The empirical process know-how required -- developed over 40 years at Aixtron -- cannot be acquired through capital expenditure or government mandate alone.

The timeline to worry: Based on LED market dynamics, a reasonable estimate for AMEC to achieve competitive parity in GaN power MOCVD for commodity applications is 3-5 years (2028-2030). For advanced applications (300mm, high-reliability automotive, cutting-edge SiC), the timeline extends to 7-10 years. Aixtron's non-China revenue should be insulated for the foreseeable future. The China-facing revenue (~35-40% of the Asia geography, or ~20-25% of total) faces structural long-term pressure.

What to watch: AMEC shipping production-qualified SiC or GaN power MOCVD tools to a customer outside China -- particularly a Western IDM -- would be a serious escalation of the competitive threat. Until that happens, the risk is contained to the Chinese domestic market.

Regulatory & Geopolitical Landscape

The China revenue equation: Asia represents 60% of revenue, with China a significant portion of that (likely 25-35% of total, though the company does not break out China specifically). This creates a complex geopolitical calculus:

Near-term: China is still buying Aixtron tools. Chinese compound semi companies (Innoscience, Sanan Optoelectronics) are major customers. Innoscience is the world's largest GaN IDM with 31% device market share and is expanding aggressively.
Medium-term: Domestic sourcing pressure is building. China's discreet 50% domestic sourcing mandate for wafer fab equipment is structural policy, not cyclical. As AMEC improves its GaN/SiC capability, Chinese fabs will face increasing pressure to switch.
Long-term: The non-China market is growing faster. US CHIPS Act, EU Chips Act, and reshoring trends are driving investment in Western fabs. Aixtron's Americas revenue grew 66% in FY2025. The geographic mix is naturally diversifying.

The China risk is real but manageable. Even if China revenue declines 20-30% over the next 3-5 years (as AMEC gains share in commodity applications), it can be more than offset by growth in the Americas and Europe. The critical risk is AMEC achieving parity in advanced applications (300mm GaN, cutting-edge SiC) -- which would signal a broader competitive challenge beyond just the China market.

Aixtron's regulatory framework: The company is governed by the German Corporate Governance Code (DCGK), the German Stock Corporation Act (AktG), EU Market Abuse Regulation (MAR), and the German Securities Trading Act (WpHG). It files with BaFin/Deutsche Borse, not the SEC. KPMG AG serves as external auditor (unqualified opinion on FY2025).

Export controls: MOCVD equipment is NOT currently targeted by US/EU BIS export control packages (Oct 2022, Apr 2024, Dec 2024). Those focus on sub-16nm logic/memory. Future restrictions are possible if defense-grade GaN applications attract attention. This creates both risk (China revenue loss) and near-term opportunity (stockpiling). Management notes that semiconductor equipment is "currently not affected by U.S. tariffs" and that the EU Dual-Use Regulation applies, requiring export licenses for certain destinations.

EU Chips Act: EUR 43B investment to raise Europe's semiconductor share from <10% to 20%. Germany earmarked ~EUR 20B. Aixtron, as Europe's compound semi equipment champion, benefits directly from Infineon, STMicro, and ZF/Wolfspeed European fab investments.

SiC substrate geopolitics:

Supplier	2024 Share	Country	Notes
Wolfspeed	33.7%	USA	Filed Ch.11 June 2025; emerged Sept 2025
TanKeBlue	17.3%	China	Rapid share gains; pricing pressure
SICC	17.1%	China	First 300mm SiC substrate; 200mm leader
Coherent	13.9%	USA	Declining share

Chinese SiC substrate makers have collapsed 6-inch wafer prices from ~$1,500 to ~$400-500, restructuring the economics of the entire SiC value chain.

Technology Roadmap

200mm SiC (underway now): Industry migration from 150mm to 200mm SiC increases chip output ~85% per substrate. STMicro began 200mm production Q4 2025 in Catania (scaling to 15,000 wafers/week by 2033); Infineon's EUR 2B Kulim fab entered 200mm SiC mass production in 2025. Aixtron's G10-SiC is designed for both 150mm and 200mm, winning 50-60% of new 200mm business.

300mm GaN-on-Si (Hyperion, 2027-2028): Prototype shipped to pilot lines 2025. Single-wafer showerhead for 300mm Si fab compatibility. If successful, enables GaN power integration into existing 300mm foundries -- massive TAM expansion. High-risk, high-reward. Volume contribution 2028+ at earliest. Veeco competing with Propel 300.

Vertical GaN: onsemi unveiled vGaN (Oct 2025) for 1200V+ applications -- ~3x smaller than lateral GaN. Could eventually challenge SiC at higher voltages. Commercialization at scale: 2028+ at earliest. Watch closely but not an imminent SiC threat.

300mm SiC (5+ years away): SICC (China) exhibited first 300mm SiC substrate at Electronica 2024; Wolfspeed announced successful single-crystal wafer. Volume production: 2030+ at earliest. Not investable near-term.

Ultra-wide bandgap (Ga2O3, AlN, diamond):

Material	Status	Timeline to Commercial	Threat to SiC/GaN
Ga2O3 (Gallium Oxide)	Most advanced UWBG; lab devices exist	2028-2030 (niche applications)	Low this decade
AlN (Aluminum Nitride)	Early R&D	2030+	None near-term
Diamond	Early R&D	2030+	None near-term

Not a threat to SiC/GaN this decade. Importantly, MOCVD equipment for these materials would represent future Aixtron TAM, not competition. Aixtron's reactor expertise would likely extend to UWBG materials as they mature, creating a new growth vector rather than an obsolescence risk.

Industry Structure & Cycle Dynamics

The compound semiconductor equipment market is highly consolidated in MOCVD (effectively a monopoly at 77% share) and moderately concentrated in SiC CVD (three players hold 75%).

Cyclicality: This industry is highly cyclical. Equipment demand is driven by customer capex, which swings dramatically based on:

End-market demand signals (EV production, data center buildout, consumer electronics cycles)
Technology transitions (new wafer sizes, new materials)
Government incentive programs (Chips Act, EU Chips Act)
Inventory cycles at the device level

However, compound semi equipment is less cyclical than memory or logic equipment because end markets (EVs, 5G, defense) are tied to multi-year infrastructure transitions, not consumer device replacement cycles. The customer base is smaller and orders are lumpier -- a few large orders can swing quarterly results.

Historical cycle data: Aixtron's revenue has ranged from EUR 210M (2016 trough) to EUR 634M (2024 peak) over the past decade. Cycles typically last 2-4 years from trough to peak. The LED-driven boom/bust of 2010-2013 was more severe (EUR 1B+ peak to EUR 240M trough) because LED was a single end-market. Today's diversification across power, opto, and LED/Micro LED should make cycles shallower.

Consolidation history:

2017: AMEC breaks the Aixtron-Veeco LED MOCVD duopoly in China
2022: Veeco acquires Epiluvac (Swedish SiC CVD startup) to enter SiC market
2024: Aixtron acquires production site near Turin, Italy

The MOCVD market is too small (~$1-2B) to consolidate further among the big three, and too small to attract the silicon equipment giants. This is a stable oligopoly that benefits incumbents.

Emerging threats and disruptors:

Chinese competition is the most credible long-term threat. Beyond AMEC, Beijing NAURA, JSG, TOPEC, and Agnitron are all working to qualify MOCVD/CVD systems. Chinese competitors reportedly offer prices 50-75% below Western equivalents. But the technology gap remains wide, particularly in advanced applications (data center lasers, 300mm GaN, advanced SiC power). Chinese tools are competitive in commodity LED but not yet in high-performance applications. Qualification cycles give Aixtron years of protection.

No credible alternative to MOCVD/CVD for volume manufacturing of compound semiconductors exists today. MBE remains too slow. This is not like lithography where a fundamentally different technology (EUV vs. DUV) could disrupt the incumbent.

Adjacent entrants: ASM International (already #1 in SiC CVD) and Tokyo Electron (already #3 in SiC CVD) are the most credible. Both are massive companies with deep customer relationships in silicon equipment. Their SiC CVD positions were built on those relationships, not MOCVD expertise. They do not compete in MOCVD.

Porter's Five Forces

Force	Assessment	Detail
Supplier power	Low-Moderate	Components sourced broadly; no single supplier has outsized leverage. Aixtron assembles from externally sourced modules.
Buyer power	Moderate-High pre-qualification; Low post-qualification	Large IDMs negotiate hard on initial purchases. But once qualified, switching costs dramatically reduce buyer power -- price sensitivity drops when requalification costs millions and takes 18 months.
Threat of new entrants	Low	Extremely high technical barriers (40 years of empirical process know-how), 6-18 month qualification cycles, chicken-and-egg installed base problem, EUR 90M+/year R&D table stakes.
Substitutes	Low	MOCVD is the established method for volume manufacturing of III-V compound semiconductors. MBE is orders of magnitude too slow for production. HVPE is used only for bulk substrates.
Competitive rivalry	Moderate	Aixtron dominates MOCVD (77%) but faces real competition in SiC CVD from ASM (32%) and TEL (16%). AMEC is the disruptive competitor but confined to China/LED for now.

Net assessment: Industry structure is highly favorable for Aixtron. The combination of high switching costs, high entry barriers, low substitution risk, and a market too small for silicon equipment giants creates a rare structural moat.

EU Research Consortia & Strategic Partnerships

ALL2GaN: 45-partner European consortium developing a GaN integration toolbox for power electronics across 12 countries. Aixtron's participation positions it at the center of European GaN ecosystem development and provides early access to next-generation process requirements.

NeuroSys: Aachen-region neuromorphic computing cluster where Aixtron develops 2D materials (memristors) for neuromorphic computing applications. Speculative but shows forward-thinking R&D portfolio extending beyond current compound semiconductor applications.

APEVA Group: 87% stake in APEVA Holdings Ltd. (UK) / APEVA Co. Ltd. (South Korea) focused on OLED deposition technology. Small, non-core. APEVA SE i.L. in Germany being wound down. The OLED bet did not pay off, but the financial impact is immaterial.

No material JVs or partnerships drive the investment thesis directly. Aixtron's most important "partnerships" are its deep customer relationships with major IDMs and foundries -- these are not formal JVs but are critical to the business through the qualification and co-development process.

Assets & Operations Footprint

Location	Function	Notes
Herzogenrath, Germany	HQ, R&D, Manufacturing, Sales, Service	Innovation Center (1,000+ m2 cleanroom, 300mm development); EUR 100M investment, fully commissioned early 2025
Cambridge, UK	R&D, Manufacturing, Service	AIXTRON Ltd.
Turin, Italy	Production	AIXTRON S.R.L. (acquired 2024)
Heerlen, Netherlands	R&D	AIXTRON B.V.
Santa Clara, USA	Sales, Service	AIXTRON Inc.
Hwaseong, South Korea	Sales, Service
Shanghai, China	Sales, Service
Hsinchu, Taiwan	Sales, Service
Tokyo, Japan	Sales, Service
Kulim, Malaysia	Service

Aixtron is moderately asset-light. It assembles systems from sourced components rather than running fabs. PP&E is EUR 242M against EUR 557M revenue.

Corporate Structure

AIXTRON SE (Germany) -- Parent Company
    |
    |-- AIXTRON Ltd. (UK) -------------- 100% -- R&D, Manufacturing, Service
    |-- AIXTRON Korea Co. Ltd. --------- 100% -- Sales, Service
    |-- AIXTRON K.K. (Japan) ----------- 100% -- Sales, Service
    |-- AIXTRON China Ltd. ------------- 100% -- Sales, Service
    |-- AIXTRON Taiwan Co. Ltd. -------- 100% -- Sales, Service
    |-- AIXTRON Inc. (USA) ------------- 100% -- Sales, Service
    |-- AIXCell Ltd. (UK) -------------- 100%
    |-- AIXTRON Malaysia Sdn. Bhd. ----- 100% -- Service
    |-- AIXTRON S.R.L. (Italy) --------- 100% -- Production
    |-- AIXTRON B.V. (Netherlands) ----- 100% -- R&D
    |
    |-- APEVA Holdings Ltd. (UK) ------- 87% -- OLED deposition technology (non-core)
         |-- APEVA Co. Ltd. (Korea) ----- 87%
    |
    [APEVA SE i.L. (Germany) -- excluded from consolidation May 2025, in liquidation]

Clean, simple structure. All operating subsidiaries 100% owned. No shell entities, no layered holding structures.

Adjacent Industries & Convergence Themes

AI infrastructure convergence: This is the most powerful theme. AI is simultaneously driving demand across three of Aixtron's end markets:

SiC + GaN for data center power conversion (NVIDIA 800V HVDC)
InP/GaAs for optical transceivers connecting GPUs (800G+ photonics)
GaN RF for potential edge AI communications infrastructure

No other company in the semiconductor equipment industry has this kind of triple exposure to AI infrastructure. ASML is exposed through lithography for AI chips themselves. Aixtron is exposed through the power and connectivity layers that make those chips usable.

Photonics for compute: Silicon photonics is converging with III-V materials. Intel, TSMC, and GlobalFoundries are integrating InP and GaAs lasers with silicon photonic circuits for co-packaged optics (CPO). CPO moves the optical transceiver from the front panel directly onto the ASIC package, dramatically reducing power consumption and latency for AI training clusters. This architecture requires MOCVD-grown III-V epitaxial layers -- expanding Aixtron's addressable market into the silicon foundry ecosystem for the first time.

Electrification mega-trend: Beyond EVs, the broader electrification of everything -- grid modernization, industrial motor drives, renewable energy inverters, heat pumps, data center power -- structurally increases demand for power semiconductors. SiC and GaN are taking share from silicon in every high-performance power application. This is a multi-decade structural shift.

Photonics for compute -- the silicon photonics convergence: This deserves special attention because it represents a potential TAM expansion that few equipment investors appreciate. The problem: as AI training clusters scale to tens of thousands of GPUs, the electrical copper interconnects between chips become the bottleneck. You cannot move 400Gbps+ between racks using electrical cables over meaningful distances without unacceptable latency and power consumption. The solution: optical interconnects that use light instead of electricity.

Currently, optical transceivers sit on the front panel of network switches. The emerging architecture -- co-packaged optics (CPO) -- moves the transceiver directly onto the ASIC package, reducing power consumption and latency by 30-50%. Intel, TSMC, and GlobalFoundries are integrating InP and GaAs lasers with silicon photonic circuits to enable CPO. This heterogeneous integration requires MOCVD-grown III-V epitaxial layers bonded or hetero-integrated with silicon photonic circuits.

For Aixtron, CPO represents an entirely new customer category: silicon foundries that have never bought MOCVD equipment before. If TSMC or GlobalFoundries adds III-V epitaxy capability to their silicon photonics lines, they will need Aixtron (or Veeco) tools. This is TAM expansion into an ecosystem worth hundreds of billions of dollars -- even a tiny penetration rate would be meaningful for a company with EUR 557M in revenue.

The timeline is uncertain (CPO at scale is probably 2028-2030), but the direction of travel is clear. Every major semiconductor company is investing in silicon photonics, and all of them will eventually need III-V epitaxy capability. Aixtron's G10-AsP platform is already the "leading solution" for datacenter laser epitaxy -- it is well-positioned to capture this convergence.

Defense and sovereignty: Global defense spending is surging. GaN enables phased-array radar (AESA) systems that are the backbone of modern military capabilities. The AESA radar market is growing to $14.5B by 2030. Semiconductor sovereignty concerns (EU Chips Act, US CHIPS Act) are driving reshoring of compound semiconductor manufacturing to Europe and the US, creating new greenfield fab investments that directly benefit Aixtron as Europe's equipment champion.

The 300mm GaN Opportunity -- Why It Could Be Transformative

The 300mm wafer transition for GaN is not just about making bigger wafers. It is about unlocking a completely different manufacturing ecosystem.

Today, GaN power devices are made on 200mm (8-inch) wafers in dedicated compound semiconductor fabs. These are relatively small facilities with limited capacity. The entire GaN power MOCVD equipment market is a subset of a ~$1B MOCVD market.

300mm (12-inch) silicon fabs are a completely different scale. The world's leading foundries (TSMC, GlobalFoundries, Samsung) and IDMs (Intel, TI, Infineon) operate hundreds of 300mm tools. These fabs represent trillions of dollars of installed infrastructure. If GaN can be grown on 300mm silicon wafers and processed in those existing fabs, three things happen:

Manufacturing cost drops dramatically. 300mm wafers have 2.25x the die area of 200mm. Combined with the higher throughput and yield of 300mm fabs, the cost per GaN chip could fall 40-60%.
Capacity becomes virtually unlimited. Instead of building dedicated GaN fabs ($1B+ each), device makers can add MOCVD tools to existing 300mm silicon fabs. The incremental capacity is enormous.
Aixtron's addressable market expands into the silicon fab ecosystem. Instead of selling tools only to compound semi specialists, Aixtron could sell to every major silicon fab that wants to add GaN capability. The TAM expansion potential is measured in multiples, not percentages.

This is why the Hyperion platform matters so much. It uses single-wafer showerhead architecture (not Aixtron's traditional Planetary) specifically because 300mm silicon fabs are built around single-wafer processing flows. Aixtron deliberately sacrificed its batch throughput advantage to gain compatibility with the existing 300mm fab infrastructure.

Current status: Prototype Hyperion systems are at customer pilot lines. Imec's 300mm GaN Power Electronics Program (announced October 2025) uses Aixtron's Hyperion as the foundational equipment. Volume production expected 2027-2028.

Risk: This is high-risk, high-reward. Veeco is competing with its Propel 300 and has secured at least one order. The technology is not yet proven in production. If 300mm GaN takes longer to qualify than expected (2029+), the near-term impact on Aixtron's revenue is minimal. But if it succeeds on the expected timeline, it could be the most significant inflection point in Aixtron's history.

Secular Tailwinds & Headwinds

Tailwinds:

Tailwind	Mechanism	Magnitude	Durability
AI data center power	NVIDIA 800V HVDC (2027); GaN+SiC at every power conversion stage	Very high	5-10+ years
GaN power adoption	Device market 6x by 2030; fastest-growing compound semi segment	Very high	5-10+ years
AI optical interconnects	800G+ transceivers need InP/GaAs lasers; demand 6x by 2030	Very high	3-5+ years
EV 800V architecture	SiC MOSFETs required; ~$200-500 SiC content per EV	High	5-10+ years
Defense GaN	AESA radar to $14.5B by 2030; global defense spending surging	High	5-10+ years
5G/6G RF	GaN amplifiers for base stations; 6G commercial ~2030	High	5-10 years
EU Chips Act	EUR 43B European fab investment; Aixtron is EU's compound semi champion	Moderate	3-5 years
Micro-LED	AR/VR headsets, large signage	Moderate	Timing uncertain (2028+)

Headwinds:

Headwind	Probability	Impact
China MOCVD self-sufficiency (AMEC)	High (structural)	High for China revenue
SiC overcapacity (~50% upstream utilization)	High (ongoing)	High near-term
Equipment cycle lumpiness	High (inherent)	Moderate
EV adoption cyclicality	Moderate	Moderate (temporary)
Vertical GaN disruption (onsemi)	Low-moderate	Low near-term (2028+)

Management

Executive Board (Vorstand)

Name	Title	Since	Term Ends	Background
Dr. Felix Grawert	CEO (Chairman of Executive Board)	Aug 2017	Aug 2030	Engineering (university of applied sciences). Responsible for strategy, sales, marketing, R&D, operations. Has led Aixtron through a full cycle -- navigated the 2019-2020 trough, drove the upcycle to EUR 634M peak revenue, launched the G10 platform, oversaw the Innovation Center (on schedule, on budget). Now managing the downcycle with cost cuts. The 13-year contract (2017-2030) signals high board confidence.
Dr. Christian Danninger	CFO (Member of Executive Board)	May 2021	Apr 2029	Responsible for finance, IR, compliance, IT, HR, legal, ESG. Managed the balance sheet through the heavy capex cycle while maintaining zero bank debt. Delivered EUR 182M FCF in FY2025 through working capital optimization.

The two-member Executive Board is lean -- standard for German mid-cap SEs but creates concentration risk. Mitigated by a 6-member Executive Committee of senior managers, though none have statutory Executive Board authority.

CEO assessment: Grawert is a credible through-cycle operator. He invested heavily during the upcycle (Innovation Center, G10 platform R&D) and is cutting costs during the downturn. His track record -- launching the G10 family, delivering the 100th G10-SiC, completing the Innovation Center on time and budget -- demonstrates execution credibility.

Supervisory Board (Aufsichtsrat)

Name	Role	Independent?	Since	Background	Key Credential
Alexander Everke	Chairman	Yes	2024	Former CEO, ams-OSRAM	Also on ASML Supervisory Board -- deep visibility into semi equipment industry from #1 player
Frits van Hout	Deputy Chair	Yes	2019	Dutch entrepreneur; physics background	Deep photonics ecosystem (Kendrion, PhotonDelta, InvestNL, SmartPhotonics)
Ingo Bank	Member	Yes	2025	CFO of City Football Group	Financial accounting + auditing expertise
Karen Florschutz	Member	Yes	2024	EVP Connected Intelligence, Airbus Defence & Space	Defense/industrial tech perspective
Dr. Stefan Traeger	Member	Yes	2022	Chairman of JENOPTIK AG (optics/photonics)	Complementary optics/photonics industry insight
Prof. Dr. Anna Weber	Audit Chair	Yes	2019	Professor of Business Administration; certified public auditor	Also on Wacker Chemie Supervisory Board

All 6 members are independent. No former Executive Board members on the Supervisory Board. Kim Schindelhauer, the longtime Chairman and co-founder figure, stepped down at the May 2025 AGM after 30+ years.

Board quality assessment: This is a high-quality board. Everke's ASML Supervisory Board seat is the standout credential -- direct visibility into semi equipment from the #1 player. Traeger running JENOPTIK provides complementary optics/photonics insight. Van Hout's deep involvement in the Dutch photonics ecosystem gives line-of-sight into the InP supply chain feeding Aixtron's opto segment. Prof. Dr. Weber's certified public auditor credentials make the Audit Committee genuinely competent.

Committee structure:

Audit Committee: Weber (Chair), Bank, Everke
Capital Markets Committee: Everke (Chair), Bank, Traeger
Nomination Committee: Everke (Chair), Florschutz, Traeger, van Hout
Remuneration Committee: van Hout (Chair), Everke, Florschutz

Governance structure: Single class of shares, one vote per share. No dual-class structure. No poison pills. Full compliance with German Corporate Governance Code (DCGK). KPMG AG as external auditor (unqualified opinion on FY2025). 99% free float, no controlling shareholder, no anti-takeover provisions. Declaration of full DCGK conformity as of February 2026, no deviations. External ICS review by Deloitte in FY2025 found no inadequacies.

Insider Ownership & Skin in the Game

Name	Role	Shares Owned	% of Outstanding	Est. Value	Source
Dr. Felix Grawert	CEO	Not disclosed above threshold	<0.1% (est.)	<EUR 2M (est.)	Primarily LTI grants
Dr. Christian Danninger	CFO	Not disclosed above threshold	<0.1% (est.)	<EUR 2M (est.)	Primarily LTI grants
Alexander Everke	SB Chair	Not disclosed	Negligible	Unknown	Unknown

The yellow flag: Insiders are not buying their own stock. At EUR 8.45 (the 2025 low), neither the CEO nor the CFO purchased shares on the open market, even though they had long-term contracts implying they believed in the company's future. This may reflect European cultural norms, but it still signals limited personal conviction through capital deployment.

No material directors' dealings flagged under EU MAR reporting in FY2025. The prior stock option program (Conditional Capital II 2012) expired December 31, 2024. No new stock option program was launched.

Compensation & Alignment

FY2025 Executive Board compensation:

Component	FY2025 (EUR K)	FY2024 (EUR K)
Short-term employee benefits	3,258	3,433
Share-based payments (fair value)	3,234	4,406
Total	6,492	7,839

Total comp of EUR 6.5M is reasonable for a EUR 2B market cap European semi equipment company. SBC as % of revenue is 0.6% -- negligible dilution.

LTI structure: Includes sustainability targets over a 3-year reference period (energy consumption reduction, diversity/inclusion) alongside financial performance conditions. The sustainability KPIs dilute the direct pay-for-performance link somewhat, though they align with EU regulatory trends and the financial component likely dominates.

Change-of-control: Executives can resign with 3 months' notice. Severance equals anticipated compensation for remaining contract term, capped at 2 years' remuneration (~EUR 6.5-7M per executive max). Standard European provision.

No unusual perks. Supervisory Board total comp: EUR 737K for all six members. Modest.

Capital Allocation Track Record

Dimension	Grade	Evidence
M&A	A-	No acquisitions. Organic growth through R&D. APEVA venture being wound down but financially immaterial.
R&D investment	A	Consistent 14-15% of revenue. Innovation Center (EUR 100M) is bold, well-timed. G10 platform: 100+ systems in 3 years.
Buybacks	C	Authorized but unused. 670K treasury shares (0.6%). Stock at EUR 8.45 was a perfect buyback opportunity -- they did nothing.
Dividends	C+	Token EUR 0.15/share (0.9% yield, ~20% payout). Not hostile, but not meaningful.
Capex discipline	A	Heavy investment during upcycle (2023-2024). Sharp reduction in FY2025 as Innovation Center completed.
Dilution	A	Share count flat at 113.5M for years. No equity raises. Minimal SBC dilution.

Overall capital allocation grade: B+. The main criticism is failure to deploy excess cash for buybacks at depressed prices. With EUR 225M liquidity and zero bank debt, Aixtron could have repurchased 5-10% of shares at EUR 8-12 and created significant value. Common European mid-cap conservatism.

Management DD Verdict

Dimension	Rating	Key Finding
Skin in the Game	Yellow	Minimal insider ownership. No open-market purchases even at multi-year lows.
Holdings Concentration	Yellow	Executive wealth not primarily in AIXA equity.
Shell / Cross-Holdings	Green	No related-party transactions. No shell entities. Simple structure.
Capital Allocation	Green	Disciplined R&D, no destructive M&A, minimal dilution. Loses points for buyback inaction.
Compensation Alignment	Green	Reasonable comp, negligible SBC dilution (0.6% of revenue).
Governance Quality	Green	All-independent board with strong credentials. Full DCGK compliance.
Litigation / Enforcement	Green	No material litigation, no regulatory actions, no sanctions.
Overall	B+	Clean, competent, well-governed. Main gap is low insider ownership.

Fiduciary Duty Assessment

Duty of Care: Strong. Board holds 4 regular + 2 extraordinary meetings per year. Strategy workshops, quarterly performance reviews, and detailed technology roadmap discussions are documented. External ICS review by Deloitte in FY2025 found no inadequacies. Risk management system in place with whistleblower mechanism.

Duty of Loyalty: Clean. No self-dealing, no related-party transactions, no corporate opportunities diverted. The annual report explicitly states: "No conflicts of interest of members of the Supervisory Board or the Executive Board were reported in the fiscal year under review."

Duty of Good Faith: No red flags. Compliance manual updated. Risk management system covers financial, operational, and strategic risks.

Management Green Flags

Zero related-party transactions in FY2025
All Supervisory Board members independent with relevant expertise
Full DCGK compliance with no deviations
Negligible dilution (0.6% SBC / revenue)
Fortress balance sheet managed conservatively through the downturn
Innovation Center delivered on schedule and on budget -- execution credibility
CEO contract extends to 2030, CFO to 2029 -- leadership stability
Simple, transparent corporate structure with no shell entities
100th G10-SiC delivered 3 years after launch -- product execution track record

Management Yellow Flags

Low insider ownership with no open-market purchases at multi-year lows
Buyback inaction despite EUR 225M cash and stock at depressed levels
Two-person Executive Board creates key-person and succession risk
LTI sustainability KPIs dilute the direct pay-for-performance link (though aligned with EU trends)

Management Red Flags

None identified.

Bottom line: These are clean, competent operators running a well-governed company. The governance is textbook for a German SE. I would trust these people with capital. The risk is not fraud or misalignment -- it is cyclical execution and the timing of the recovery. This is a "trust the professionals" team, not a "founder with 20% of net worth on the line" situation.

Holdings Concentration -- Where Is Management's Money?

Name	Role	AIXA Holdings	Other Interests	Where Is Majority Wealth?
Dr. Felix Grawert	CEO	Minimal (LTI grants)	None disclosed	Salary/savings, not AIXA equity
Dr. Christian Danninger	CFO	Minimal (LTI grants)	None disclosed	Salary/savings, not AIXA equity
Alexander Everke	SB Chair	Not disclosed	ASML (SB); Duagon (Deputy Chair)	Private ventures + other boards
Frits van Hout	SB Deputy Chair	Not disclosed	Bambi Belt, Kendrion (Chair), PhotonDelta, InvestNL, SmartPhotonics, Picterus	Diversified photonics/tech ecosystem
Ingo Bank	SB Member	Not disclosed	City Football Group (CFO)	City Football Group compensation
Karen Florschutz	SB Member	Not disclosed	Airbus Defence & Space (EVP)	Airbus compensation
Dr. Stefan Traeger	SB Member	Not disclosed	JENOPTIK AG (Chairman EB)	JENOPTIK equity/compensation
Prof. Dr. Anna Weber	SB Audit Chair	Not disclosed	Wacker Chemie AG (SB); University of Heilbronn	Academic/professional income

Executive Board members' wealth is not primarily in AIXA stock. At EUR 6.5M combined annual compensation, even years of LTI grants leave their exposure modest. The Supervisory Board members have appropriately diversified professional interests -- you want independent directors with active careers, not dependents who cannot afford to lose their board seat.

No concerning cross-holdings with customers, suppliers, or competitors. Van Hout's SmartPhotonics (InP foundry) board seat is worth noting -- SmartPhotonics could be an Aixtron customer -- but the relationship is fully disclosed and likely provides useful market intelligence rather than creating conflicts.

Related-Party Transactions & Red Flag Scan

Clean. The annual report states: "In the 2025 fiscal year, the company did not conclude or carry out any material transactions with related parties." No consulting contracts with management-affiliated entities, no IP licensing agreements, no related-party leases. KPMG reviewed and confirmed with an unqualified opinion. No litigation, no regulatory actions, no sanctions.

Financials

The Financial Story

Aixtron's financials tell a clear cyclical story with a structural growth overlay. Revenue roughly doubled from EUR 260M (2019 trough) to EUR 634M (2024 peak) -- a ~12% CAGR through the full cycle. The company then entered a downturn, with FY2025 revenue declining 12% and FY2026 guided down another 7%. But even in this downturn, the company is generating substantial free cash flow, the balance sheet is fortress-grade, and margins have compressed only modestly. This is a company that bends but does not break during downturns.

The most important financial development in recent years was the EUR 100M Innovation Center investment (2023-2024), which temporarily crushed free cash flow but is now the platform for 300mm GaN development.

P&L Summary (EUR millions)

Metric	2021	2022	2023	2024	2025	2026E	2027E
Revenue	429	463	630	633	557	520 (guidance)	666 (consensus)
YoY growth	+67%	+8%	+36%	+0.5%	-12%	-7%	+22%
Gross margin	42.3%	42.0%	~44%	41.4%	40%	41-42%	--
EBITDA	109	114	168	146	117	117	168
EBIT	99	105	157	131	100	83-99	149
EBIT margin	23.1%	22.6%	24.9%	20.7%	18.0%	16-19%	22.3%
Net income	96	100	145	106	85	68-80 (est.)	112
EPS (EUR)	0.85	0.89	1.29	0.94	0.76	0.60-0.70 (est.)	0.98
R&D (EUR M)	57	~70	88	91	81	--	--
R&D % rev	13.2%	~15%	13.9%	14.4%	15%	--	--

5-year margin ranges: Gross 40-44% | EBIT 18-25% | Net 15-23%.

Margins are compressing through the downturn but holding up well -- only 4-5pp of EBIT margin erosion despite an 18% peak-to-trough revenue decline. This reflects pricing power and cost discipline.

Free Cash Flow

Metric	2021	2022	2023	2024	2025
Operating CF	66	37	-47	26	208
CapEx	17	30	63	99	28
Free cash flow	49	8	-110	-72	182

Critical context: The -EUR 110M and -EUR 72M FCF in FY2023-2024 were driven by the Innovation Center construction (EUR 100M cumulative) plus inventory build. In FY2025, capex dropped to EUR 28M (investment complete) and inventory was reduced by EUR 85M. The EUR 182M FCF is not sustainable -- it includes one-time working capital release. Normalized FCF is probably EUR 80-100M at current revenue levels.

Balance Sheet

Virtually debt-free: Debt/equity 0.38%
Net cash: EUR 225M (end FY2025)
Total liquidity: EUR 225M cash + EUR 200M undrawn revolving credit facility = EUR 425M
Equity ratio: 88%
Total equity: EUR 910M
Inventory: EUR 351M (~230 days at FY2025 revenue) -- reduced from EUR 436M peak
No debt maturity risk. No bank liabilities.

This balance sheet can sustain a prolonged downturn. No financial stress risk whatsoever.

Recovery Scenario Modeling

The financial case for Aixtron on a 2-3 year view depends on the revenue recovery materializing. Here is how the math works across scenarios:

Scenario A: Base Case Recovery (FY2027 EUR 666M, consensus)

Revenue: EUR 666M (+22% from FY2026)
Gross margin: 42-43% (mix improvement as opto grows)
EBIT: EUR 140-155M (EBIT margin 21-23%)
Net income: EUR 105-120M
EPS: EUR 0.93-1.06
At 22-25x forward P/E: EUR 20-27 stock price
Return from EUR 17: +18% to +59%

Scenario B: GaN Inflection Bull Case (FY2027 EUR 800M)

Revenue: EUR 800M (+54% from FY2026, if GaN power ramp accelerates)
Gross margin: 43-44% (higher-margin GaN and opto mix)
EBIT: EUR 185-210M (EBIT margin 23-26%)
Net income: EUR 140-160M
EPS: EUR 1.24-1.41
At 25-30x P/E (re-rating as structural growth recognized): EUR 31-42
Return from EUR 17: +82% to +147%

Scenario C: Extended Trough (FY2027 EUR 550M, no recovery)

Revenue: EUR 550M (flat with FY2025)
Gross margin: 39-40% (continued mix pressure)
EBIT: EUR 75-90M (EBIT margin 14-16%)
Net income: EUR 55-70M
EPS: EUR 0.49-0.62
At 18-20x trough P/E: EUR 9-12
Return from EUR 17: -29% to -47%

The asymmetry is clear: the upside scenarios (+18% to +147%) are larger than the downside (-29% to -47%). This is because the operating leverage works in Aixtron's favor: each incremental EUR of revenue carries 35-45% incremental EBIT margin. The fixed cost base that crushes margins on the way down amplifies them on the way up.

Red Flags Review

No aggressive accounting identified. Revenue recognition is straightforward (delivery + acceptance).
No declining cash flow trends -- the FY2023-2024 negative FCF was planned capex, now resolved.
No rising leverage -- the company is deleveraging (net cash growing).
Unqualified KPMG audit opinion.
Inventory: EUR 351M (~230 days) is elevated but typical for lumpy, high-ASP capital equipment. Reduced from EUR 436M peak, showing management is actively managing working capital.
No material litigation, no auditor changes, no going concern language, no restatements.

Financial health verdict: Excellent. Fortress balance sheet, value-creating ROIC, resilient margins through the downturn. The only financial concern is the revenue trajectory -- but that is cyclical, not structural.

ROIC Analysis

With net income of EUR 85M on invested capital of approximately EUR 590M, FY2025 ROIC is roughly 14%. At the FY2023 peak (net income ~EUR 146M), ROIC was closer to 25%. Estimated WACC of 9-10% means Aixtron creates substantial value through the cycle.

Operating Leverage -- Incremental Margin Analysis

Q3 2025 vs Q3 2024:

Revenue: -EUR 36.7M (-23.5%)
EBIT: -EUR 22.1M (-59%)
Incremental EBIT margin on the downside: ~60%

High operating leverage cuts both ways. R&D (~EUR 90M+/year) and SG&A are semi-fixed. When revenue declined 23.5% in Q3 2025, EBIT fell 59%. The same leverage should work in reverse -- a 20%+ revenue rebound could drive 40-50%+ EBIT growth.

FY2025 vs FY2024 (annual incrementals):

Delta Revenue: -EUR 76.6M (-12%)
Delta Gross Profit: -EUR 40.1M
Incremental gross margin on lost revenue: 52%
Delta EBIT: -EUR 30.9M
Incremental EBIT margin on lost revenue: 40%

The lost revenue was high-margin business (power electronics equipment). The remaining revenue held up margins reasonably well. Revenue mix matters: as optoelectronics (AI datacom) grows as a share, margins should improve -- opto tools tend to be higher-margin.

FX headwind: 1pp EBIT margin drag from strong euro vs USD/CNY-denominated revenue noted in Q3 2025. The company set a budget rate of 1.20 USD/EUR for 2026 (conservative). FX forward contracts are used opportunistically but no systematic hedging program is in place. Structural exposure: revenue predominantly in USD/CNY, costs predominantly in EUR.

What the incrementals tell us about the recovery: Aixtron's operating leverage means the recovery should be powerful. If FY2027 consensus (+22% revenue to EUR 666M) materializes:

Incremental revenue: ~EUR 146M
At 40-45% incremental EBIT margin: ~EUR 58-66M additional EBIT
Implied FY2027 EBIT: EUR 140-165M (vs. EUR 100M in FY2025)
Implied EBIT margin: ~21-25% (returning to near-peak levels)
Implied EPS: EUR 0.95-1.15

This is why the stock is interesting as a cycle play. The operating leverage is real and works dramatically in both directions.

Revenue mix shift impact on margins: As optoelectronics (AI datacom) grows as a share of revenue, margins should improve. Opto tools tend to be higher-margin than power electronics tools because the device structures are more complex and the customer base is less price-sensitive (hyperscalers buying laser capacity care about speed of delivery, not saving 5% on tool cost). The FY2025 revenue mix was already shifting -- opto grew to 23% of systems revenue despite the overall downturn. If this continues, the margin floor during the next downturn should be higher than the current trough.

Quarterly Data

Quarter	Revenue (EUR M)	EBIT (EUR M)	EBIT Margin
Q3 2024	156.3	37.5	24.0%
Q4 2024	~168 (est.)	~37 (est.)	~22%
Q1 2025	~105 (est.)	~18 (est.)	~17%
Q2 2025	~145 (est.)	~30 (est.)	~21%
Q3 2025	119.6	15.4	12.9%
Q4 2025	186.4	--	--

Financial Health Summary

Revenue growth:

3-year CAGR (FY2022-2025): ~6% (includes FY2023 surge of +36% and FY2025 decline of -12%)
Through-cycle revenue trajectory: EUR 260M (2019) to EUR 557M (2025), roughly doubling in 6-7 years = ~12% CAGR through the full cycle
FY2026 guidance implies the trough at EUR 520M before recovery

Margin resilience through the cycle:

Gross margin: 42% (2022) -> 44% (2023 peak) -> 41% (2024) -> 40% (2025 trough) -> 41-42% (2026E)
EBIT margin: 22.6% (2022) -> 24.9% (2023 peak) -> 20.7% (2024) -> 18% (2025) -> 16-19% (2026E)
Key insight: only 4-5pp of EBIT margin erosion despite an 18% peak-to-trough revenue decline. This demonstrates real pricing power and cost discipline.

FCF trajectory:

FY2023: -EUR 110M (Innovation Center construction -- planned)
FY2024: -EUR 72M (Innovation Center + inventory build -- planned)
FY2025: +EUR 182M (capex complete, inventory released -- normalization)
Normalized FCF: ~EUR 80-100M at current revenue levels
FCF yield on normalized basis: 4-5% (on market cap) -- not cheap, not expensive for a dominant franchise

ROIC vs. WACC through the cycle:

ROIC (FY2025 trough): ~14%
ROIC (FY2023 peak): ~25%
Estimated WACC: 9-10%
The company creates substantial economic value even at the trough

Recent Developments (Last 90 Days as of April 2026)

FY2025 full-year results (Feb 25, 2026): Revenue EUR 557M (-12% YoY). EBIT margin 18%, gross margin 40%. Optoelectronics surged on AI datacenter laser demand. FCF swung to EUR 182M. Dividend proposed at EUR 0.15/share (unchanged).
FY2026 guidance: Revenue ~EUR 520M (+/- EUR 30M), implying -7% YoY at midpoint. Gross margin 41-42%, EBIT margin 16-19%. Q1 2026 revenue expected ~EUR 65M (+/- EUR 10M). Key drivers: SiC decline due to overcapacity, partially offset by opto and GaN growth.
Workforce reduction: Further staff cuts agreed January 2026 at Herzogenrath. Mid-single-digit million EUR restructuring charge. Savings factored into 2026 guidance.
Innovation Center fully commissioned: Operational since early 2025, producing 300mm wafers to spec.
100th G10-SiC system delivered in 2025 -- 3 years after launch.
US tariff policy: Semiconductor equipment "currently not affected by U.S. tariffs." EU Dual-Use Regulation applies.

Next earnings: Q1 2026 Report -- April 30, 2026. AGM -- May 13, 2026.

Capital Allocation

Year	Dividend/Share (EUR)	Payout Ratio	Notes
2021	0.30	~35%
2023	0.40	~31%	Peak
2024	0.15	~16%	Cut due to FCF investment cycle
2025	0.15 (proposed)	~20%	Unchanged
2026E	0.18	~28%	Recovery expected

No share buyback program. Shares outstanding flat at ~113.5M. Authorized capital of 41.5M shares and conditional capital of 15M shares exist but are unlikely to be used.

The buyback question: With EUR 225M cash and the stock at EUR 17 (April 2025 low was EUR 8.45), Aixtron could have repurchased 5-10% of its shares at EUR 8-12 and created enormous value for remaining shareholders. Instead, the cash just sits. This is a common European mid-cap problem -- boards are culturally conservative about buybacks and prefer to maintain financial optionality. The authorization exists (up to 10% of shares outstanding, expires May 2027), but management has shown no inclination to use it.

Is this a problem? Mildly. The cash provides a genuine comfort blanket during a cyclical downturn -- it means the company can maintain R&D investment (critical for competitive position) and make strategic moves (like the Innovation Center) without tapping equity markets. But the opportunity cost of holding EUR 200M+ in cash at 0-2% interest rates while the stock trades at 23x trough earnings is real. If management had bought back 5M shares at EUR 10 average in early 2025, those shares would be worth EUR 86M at current prices on an investment of EUR 50M -- a 72% return in under a year, far better than the return on cash. This is the kind of capital allocation miss that separates "competent stewards" from "exceptional allocators."

Dividend trajectory: The dividend has been a non-event. EUR 0.15/share yields 0.9% at EUR 17 -- barely worth the administrative overhead. If Aixtron wanted to be taken seriously as a dividend payer, it would need to commit to EUR 0.40-0.50/share (2-3% yield), which would require EUR 45-57M/year. At normalized FCF of EUR 80-100M, that is achievable but would constrain the company's ability to invest in growth (R&D, potential M&A). The current approach -- token dividend plus organic R&D investment -- is probably the right one, but it would be improved by adding a systematic buyback program.

Catalysts & Risks

Bull Catalysts

Near-Term (0-12 months):

Catalyst	Timing	Impact
Q1 2026 results	April 30, 2026	High -- will show if EUR 65M quarterly revenue is a bottom
Q1 2026 order intake	May 2026	High -- first clean signal on recovery trajectory
NVIDIA GTC 2026	March 2026	Medium-high -- 800V HVDC updates validate GaN/SiC TAM
G10-AsP AI datacom design wins	Ongoing	Medium -- opto revenue trajectory
SiC fab utilization recovery	H2 2026	Medium -- leading indicator for equipment orders
AGM management commentary	May 13, 2026	Medium -- 2026/2027 outlook signals

Medium-Term (1-3 years):

Catalyst	Timing	Impact
SiC equipment upcycle	Late 2026 / 2027	Very high -- largest single revenue driver
300mm GaN Hyperion volume orders	2027-2028	High -- new TAM into 300mm Si fabs
Commercial 800V HVDC deployments	~2027	High -- validates Nvidia supply chain thesis
GaN automotive OBC/inverter adoption	2027+	Medium-high -- new device market
EU Chips Act fab investments	2026-2028	Medium -- domestic demand driver
Micro-LED production viability	2028+	Medium -- optionality

Bear Risks

#	Risk	Probability	Impact	Mitigation
1	SiC overcapacity persists beyond 2028	Medium	Very high	Aixtron's 200mm technology advantage; GaN and opto diversification
2	AMEC achieves GaN/SiC MOCVD parity in China	Medium	High	Technology gap still real (5-10 years); non-China markets unaffected
3	Extended equipment cycle trough	Medium	Medium-high	Net cash EUR 225M; zero debt; FCF recovery underway
4	Customer concentration	Medium	Medium	No single customer >10%; but few large accounts dominate
5	EV adoption slowdown	Low-medium	Medium	SiC demand tied to broader electrification, not just EVs
6	300mm GaN execution risk	Medium	Medium	Prototype stage; Veeco competing with Propel 300
7	FX risk (strong euro)	Medium	Low-medium	1pp EBIT margin per significant EUR move

Thesis Invalidators -- Exit Triggers

AMEC ships production-qualified SiC MOCVD tools to a major Western IDM
NVIDIA abandons or delays 800V HVDC architecture
Aixtron loses a top-3 customer to Veeco or AMEC
Loss of MOCVD market share below 65%
Industry pivots to 300mm SiC before Aixtron has a qualified tool

Risk Assessment Matrix (Detailed)

Risk	Likelihood	Impact	Existing Mitigants	Mgmt De-risk Plan	Can It Be Closed?
Extended SiC overcapacity	High	Very high	Diversified end-markets (power, opto, LED) with low correlation; China SiC demand still solid	Cost reduction programs (voluntary + involuntary); working capital optimization; reduced capex	Partially -- cyclical, resolves as EV demand catches up. 2-3 year headwind.
Chinese competition (AMEC)	Medium (5+ year)	High	77% market share, 306 patents, 40 years process know-how, deep switching costs	Continuous R&D (15% of revenue); 300mm platform to maintain technology lead	No -- China will continue investing. But technology gap remains wide for advanced applications.
Revenue decline FY2026	High	Medium	Strong balance sheet (net cash EUR 225M, zero debt)	Workforce reduction; working capital optimization; FY2026 guidance already conservative	Partially -- guided, so already in the price. Risk is if actual results are below guidance.
FX exposure	Medium	Low-medium	FX forward contracts used opportunistically; budget rate 1.20 USD/EUR for 2026	Monitoring and hedging on case-by-case basis	No -- structural exposure. Revenue in USD/CNY, costs in EUR.
300mm qualification delay	Low-medium	Medium	Multiple pilot systems delivered; Innovation Center producing wafers to spec	Active customer co-development	Yes -- closes once 2-3 major customers qualify tools for production (expected 2026-2028)
Customer concentration	Medium	Medium	No single customer >10% disclosed; customer base spans IDMs, foundries, research	Geographic diversification (Americas +66% in FY2025); EU Chips Act	Partially -- structural to equipment business
Technology obsolescence	Very low	Very high	Platform strategy (G10) reduces one-off risk; extensive customer collaboration during development	Innovation Center enables customer co-development	Partially -- each new platform is a bet, but track record is strong

Bear Case Scenario (Full Downside)

SiC overcapacity persists through 2029. AMEC captures 30%+ of non-LED MOCVD in China. GaN data center adoption slower than projected. Revenue stuck below EUR 600M through 2027. EBIT margins compress to 12-14%.

Bear case EPS: EUR 0.40-0.50 -> at 15x trough earnings = EUR 6-7.50 (55-60% downside). The balance sheet means the company survives easily, but the stock would be painful.

What makes the thesis definitively wrong: A Chinese competitor winning a major advanced-application customer -- SiC power for a tier-1 auto OEM, or optical interconnect lasers for a hyperscaler -- demonstrating that the technology gap has closed. That would signal the beginning of secular market share erosion, not just cyclical weakness. If that happens, the 77% MOCVD monopoly thesis breaks and the stock deserves a permanently lower multiple.

Portfolio correlation risk: Aixtron is correlated to global semiconductor capex cycles, EV adoption trajectory, and Asian (especially Chinese) industrial investment. If already holding other semi equipment names (ASML, AMAT, ASMI), Aixtron adds concentration risk in the semi equipment theme. It is partially hedged by the compound semi specificity -- Aixtron does not correlate with memory or logic equipment cycles.

Downside protection: The EUR 225M cash, zero debt, and 88% equity ratio provide genuine downside protection. Even in the worst bear case, the company does not face existential risk. It can sustain years of depressed revenue without cutting into the bone. The R&D investment can be maintained through the downturn (critical for maintaining the technology lead). Dividends can be cut to zero without consequence. This is a company that can survive a 5-year trough and emerge with its competitive position intact -- which limits the permanent capital loss risk even if the stock price suffers.

Cycle Positioning

Where are we? Near the trough. Both cyclical and secular -- the secular trend (silicon -> compound semiconductors) is irreversible, but equipment buying is highly cyclical.

Signal	Reading	Interpretation
Revenue trajectory	FY2025 down 12% YoY	Downcycle confirmed
Book-to-bill	1.04x (Q3 2025)	Stabilizing / inflecting
Backlog	EUR 258-286M (flat-to-declining)	Not yet recovering
Order intake (FY2025)	EUR 544M (-9% YoY)	Trough territory
SiC utilization	~50% upstream, ~70% device	Overcapacity -- still digesting
Optoelectronics	Accelerating Q2 2025+	AI-driven bright spot
Analyst upgrades	Jefferies, Barclays, BofA upgraded	Sentiment shifting

Historical cycle duration: 2-4 years peak-to-peak. Last peak: 2023-2024. Trough: 2025-2026. Recovery expected late 2026 / early 2027.

Leading indicators to watch:

Aixtron order intake / book-to-bill (quarterly)
SiC fab utilization rates (Yole, TrendForce reports)
Wolfspeed, Infineon, STMicro capex announcements
NVIDIA 800V HVDC architecture adoption timeline
GaN power device revenue growth (Yole quarterly)
800G+ optical transceiver shipment volumes

Decision Log

Valuation -- The Core Question

The investment case boils down to a single question: is the current price adequately compensating you for 12-18 months of weak earnings before the next upcycle? At EUR 17, you are paying 23x trough earnings for a company that earned EUR 1.29 EPS at the peak (FY2023) and could plausibly earn EUR 1.00+ again by FY2027. If the cycle turns as expected, the stock is cheap. If the trough extends, you sit dead money.

The variant perception is the GaN power inflection. Most analysts model Aixtron as a cyclical SiC equipment story with some GaN upside. Kerrisdale and a few sell-side bulls are modeling it as a structural AI beneficiary where GaN power becomes 50%+ of revenue by 2027-2028. The truth is probably in between, but the skew is favorable.

Current multiples (EUR 17.31, Dec 2025):

Metric	FY2025 (TTM)	FY2026E	FY2027E	5Y Range
P/E	23x	~26-29x	~17-19x	15-55x
EV/EBITDA	~15x	~17-19x	~13-14x	10-35x
EV/Revenue	3.1x	3.3x	~2.9x	2-8x
EV/EBIT	~17x	~18-25x	~12-17x	12-40x
P/FCF	10.8x (distorted)	N/A	~19-22x (normalized)	--

Note: At ~EUR 23 (Feb 2026), multiples are higher: P/E 30x, EV/EBITDA 22x. Stock has traded EUR 17-23 during analysis period.

Peer comparison:

Company	EV/EBITDA (NTM)	P/E (NTM)	Revenue Growth (NTM)
Aixtron	15-22x	23-35x	-2% to -7% (trough)
Veeco (VECO)	~15x	~22x	+5%
ASM International	~28x	~35x	+12%
ASML	~30x	~35x	+15%
Applied Materials	~18x	~22x	+8%

Aixtron trades at a significant discount to quality European semi equipment peers (ASML, ASMI). The discount undervalues the 77% MOCVD monopoly and structural AI exposure.

Scenario analysis:

Scenario	2027E EPS	Multiple	Implied Price	Return from EUR 17
Bear	EUR 0.70	20x	EUR 14	-18%
Base	EUR 1.00	28x	EUR 28	+65%
Bull	EUR 1.30	33x	EUR 43	+153%
Kerrisdale bull (2028)	EUR 2.00	33x	EUR 66	+288%

DCF sanity check: Normalized FCF of EUR 90M, growing at 8% long-term, discounted at 9%. Present value suggests fair value of EUR 22-28 per share.

What the Market Is Pricing In -- and What It Is Not

At EUR 17-23 / 23-35x trough P/E, the market prices in:

Continued trough through FY2026
Modest recovery starting FY2027
Aixtron as a cyclical equipment company with near-term headwinds

What the market is NOT pricing in:

Nvidia's 800V HVDC architecture as a structural GaN demand catalyst
Kerrisdale's scenario: EUR 1.4B+ revenue by 2029, EUR 2.00 EPS
BofA's scenario: AI-related applications reaching ~50% of revenue by 2027
300mm GaN Hyperion as a TAM expansion event
Micro-LED optionality
Optoelectronics (AI datacom lasers) as structural, not cyclical

The asymmetry is what makes this interesting. Bear case: EUR 10-14 (painful but not catastrophic with fortress balance sheet). Bull case: EUR 28-43 in the base-to-bull range. Risk/reward skews favorably for patient capital.

Pre-Buy Scorecard

Question	Answer
Can I state the thesis clearly?	Yes -- cyclical monopolist at trough earnings with structural tailwinds
Do I understand the business?	Yes -- MOCVD deposition equipment for compound semiconductors
Are the financials healthy?	Yes -- net cash, zero debt, 88% equity ratio, value-creating ROIC
Is the valuation reasonable?	Yes -- on 2-3 year recovered earnings; expensive on trough
Am I falling into a behavioral trap?	No -- narrative supported by fundamentals; bear case modeled (EUR 6-7.50)
Do the technicals support buying now?	Neutral -- basing pattern EUR 12-20; not oversold; scale-in appropriate
Have I sized the position appropriately?	Yes -- 2-4% with scaling plan
Do I have a clear exit plan?	Yes -- target EUR 25-28, reassess below EUR 11, thesis broken below EUR 9

Recommendation: BUY (Scale-In)

Technical Buy Check

Trend:

At EUR 17.31, likely near or slightly above the 50-day MA
The stock made a higher low (EUR 8.45 in April vs. prior cycle ~EUR 8 in 2022) and a higher high (EUR 20.14 vs. prior levels), then pulled back
Primary trend is ambiguous -- potentially forming a base

Support & Resistance:

Nearest support: EUR 14-15 (summer 2025 consolidation zone)
Nearest resistance: EUR 20 (November 2025 high)
Major support: EUR 8-9 (April 2025 low, only re-tests in severe bear scenario)

Chart Pattern: Broad base between EUR 12-20 during 2025. Consistent with cyclical bottoming. Break above EUR 20 with volume confirms the base; break below EUR 12 signals further downside.

Technical Verdict: Neutral-to-constructive. Potential basing pattern. Optimal entry: scale in at EUR 15-17 with willingness to add at EUR 12-14, or wait for breakout above EUR 20 with volume.

Position Sizing & Entry Plan

Conviction level: Medium (high structural quality, but cyclical timing uncertain)

Target position size: 2-4% of portfolio

Entry strategy: Scale in over 2-3 tranches:

Tranche	Price	% of Target Position	Trigger
Tranche 1	EUR 15-17	40%	Current trough levels
Tranche 2	EUR 12-14	30%	Further weakness on FY2026 results
Tranche 3	EUR 20-25+	30%	Confirmed order recovery (B/B >1.1x)

Maximum loss tolerance: 40% from average entry. If average entry is EUR 15, stop-loss at EUR 9 (roughly April 2025 low).

Re-evaluation triggers:

Upgrade to full position: Book-to-bill sustained >1.1x for 2 quarters; SiC utilization crosses 70%
Reduce position: AMEC wins production qualification at a Western SiC IDM; Aixtron loses >10pp share
Exit: Revenue fails to recover above EUR 600M by FY2027; EBIT margin drops below 15% structurally; any thesis invalidator

Investment Timing Framework

The hardest part of this investment is not the thesis -- it is the timing. Buying a cyclical monopolist at the trough sounds great in theory. In practice, troughs are psychologically difficult because:

Near-term results are ugly. FY2026 is guided down another 7%. Q1 2026 is guided at ~EUR 65M -- barely above a quarter-billion run rate. Every quarterly report for the next 12 months will look weak.
Analyst estimates may still be too high. Consensus for FY2026 is EUR 544M. The company guided EUR 520M (+/- EUR 30M). If the low end (EUR 490M) materializes and the market extrapolates further weakness, the stock could re-test EUR 12-14.
The recovery catalyst is 12-18 months away. The SiC equipment upcycle is expected in late 2026 / 2027. Commercial 800V HVDC deployments are expected ~2027. 300mm GaN volume is 2027-2028. That is a long time to hold a position that is not working.
Alternative deployment. Capital allocated to Aixtron at EUR 17 in April 2026 might be dead money for 6-12 months. The opportunity cost depends on what else is available.

The counter-argument: Trying to time the exact trough is a fool's errand. The stock went from EUR 8.45 (April 2025 low) to EUR 20.14 (November 2025 high) -- a 138% move -- in 7 months. By the time the trough is "confirmed" through multiple quarters of improving order intake, the stock has already re-rated. The 2019-2020 trough saw a similar pattern: the stock bottomed before the financials turned.

My preferred approach: Start a position now at 1.5% of portfolio. Add another 1.5% if the stock pulls back to EUR 12-14 (which is possible on weak FY2026 results). Add the final 1% on confirmation of cycle turn (order intake inflection). This way you have exposure to the recovery if it comes earlier than expected, but you retain dry powder if the trough extends.

The "if I had to buy at one price" answer: EUR 14-15 would be a strong entry with meaningful margin of safety. EUR 17 is a reasonable entry. EUR 20+ requires higher conviction in the near-term catalyst timeline.

Behavioral Traps Audit

FOMO: No. The stock is in a downturn, not a momentum run.
Confirmation bias: Possible. The structural story is compelling and easy to over-weight vs. cyclical risks. Counter: bear case modeled at EUR 6-7.50.
Narrative seduction: Possible. "77% MOCVD monopolist riding AI and EVs" is seductive. Counter: the financials confirm the narrative -- real margins, real market share, not a story stock.
Recency bias: Possible in reverse -- the downturn may be making me too bearish while the cycle could turn faster.
Verdict: Low risk. Main trap is narrative seduction, but data supports the narrative.

How to Invest in This Industry -- Peer Rankings

Rank	Company	Ticker	Why	Risk	Timeframe
1	Aixtron	AIXA:GR	Purest compound semi equipment play; 70-90% GaN power share; 35-40% SiC; trough entry	Equipment cycle; China risk	2-5 years
2	Infineon	IFX:GR	Largest SiC+GaN device portfolio; diversified; first to scale 300mm GaN	Not pure-play; auto cyclicality	2-5 years
3	Veeco	VECO	Diversified compound semi equipment; micro-LED optionality; SiC entry via Epiluvac	Less pure-play; smaller SiC position	2-4 years
4	Navitas	NVTS	Pure-play WBG IC company; GaN + SiC	Small; unprofitable; competitive risk	3-5 years (speculative)
5	onsemi	ON	Strong SiC position; vertical GaN R&D; auto exposure	Auto cyclicality; not pure-play	2-4 years

Avoid -- Wolfspeed (WOLF): Emerged from Chapter 11 in Sept 2025 but fundamental challenges remain. The debt is gone but the business model is still unproven at scale.

The case for Aixtron over direct device makers: It is the picks-and-shovels play. Aixtron wins regardless of which device maker succeeds. If Infineon wins in SiC, they buy Aixtron tools. If STMicro wins, they buy Aixtron tools. If a Chinese startup wins, they buy Aixtron tools (until AMEC catches up). The equipment maker captures the entire growth of the underlying market without bearing customer-specific risk.

Sell-Side Coverage

Analyst	Rating	Price Target	Key Thesis
BofA Securities	Buy (from Underperform)	EUR 25.10	SMID top pick 2026; AI ~50% of sales by 2027
Barclays	Overweight	EUR 20.00	AI as upside driver; GaN ~EUR 100M/year opportunity; upside EUR 30
Jefferies	Buy	EUR 27.00	AI data center growth
J.P. Morgan	Buy	--	Maintained post FY2025 results
Consensus (10-18 analysts)	Moderate Buy	EUR 17-20 avg	5-6 Buy / 5-11 Hold / 0-1 Sell

Kerrisdale Capital (Nov 2025): Long, EUR 66 target by 2028. "Most underappreciated AI beneficiary." Models EUR 1.4B+ revenue by 2029, EUR 2.00 EPS. Stock surged 15% on publication. Core argument: Aixtron addresses "two largest data center challenges from AI explosion -- power and connectivity."

Contrarian bear (Swiss Transparent Portfolio): "Secular Growth Trap with -30% Profit Collapse." Worth reading for balance, though the "obsolete tech" claim misunderstands MOCVD's value chain position.

Kerrisdale Revenue Model -- Stress-Testing the Bull Case

Kerrisdale's November 2025 long report is the most detailed public bull case for Aixtron. Their revenue model deserves examination:

Revenue trajectory (Kerrisdale model):

FY2025: ~EUR 550M (actual: EUR 557M -- close)
FY2026: ~EUR 600M (vs. company guidance EUR 520M -- more optimistic)
FY2027: ~EUR 800M (recovery + GaN ramp)
FY2028: ~EUR 1.1B (SiC upcycle + GaN acceleration)
FY2029: ~EUR 1.4B+ (full cycle + 300mm GaN volume)

Key assumptions:

GaN power equipment market triples from current levels
SiC equipment market doubles through 2029
Aixtron maintains or expands market share
300mm GaN reaches volume production by 2028
AI data center power conversion becomes the largest GaN application

What must go right for Kerrisdale to work:

GaN power adoption must accelerate on schedule (not delayed 2-3 years)
800V HVDC architecture must be adopted at scale (not just a few hyperscalers)
Aixtron must win the 300mm qualification race (not lose to Veeco)
SiC overcapacity must resolve by 2027-2028 (not persist to 2029)
AMEC must not achieve production parity in GaN power (not before 2030)

My assessment: Kerrisdale's FY2027-2028 estimates are aggressive but not implausible. The EUR 1.4B+ by 2029 requires everything to go right simultaneously, which is unlikely. A more realistic bull case is EUR 800M-1.0B by 2028-2029, which still implies significant upside from current prices. The value of the Kerrisdale thesis is not the specific price target (EUR 66) but the framework: Aixtron is being priced as a cyclical equipment company when it could be a structural AI infrastructure beneficiary. That mispricing creates the opportunity.

BofA's more conservative path: BofA models AI-related applications reaching ~50% of revenue by 2027, which implies revenue of EUR 600-700M with a structurally higher margin profile. Their EUR 25.10 target is achievable even without the full Kerrisdale revenue ramp. This is closer to the base case.

Independent Research Coverage

Source	Stance	Key Takeaway
Irrational Analysis (Jul 2025)	Cautiously bullish	"Traded poorly for 5 years because of SiC" but sees optics opportunities
SemiAnalysis	Neutral/informative	Useful for competitive dynamics; covered Veeco transformation
Kerrisdale Capital (Nov 2025)	Aggressively long	EUR 66 by 2028; most detailed public bull case
Swiss Transparent Portfolio	Short/bearish	Value trap thesis; useful contrarian check

Ownership Structure

Category	% of Shares
Institutional investors	71%
Retail investors (mostly German)	25%
Unidentified	3%
Treasury shares	1%

Top holders (WpHG notifications, year-end 2025):

Holder	%	Type
Goldman Sachs	7.1%	Mixed (proprietary, prime brokerage, market-making)
UBS	4.9%	Wealth management + trading
BlackRock	4.8%	Predominantly passive/index (MDAX/TecDAX)
Morgan Stanley	4.6%	Trading + client positions

Regional split of free float: US 31%, Continental Europe 28%, UK 20%, Germany 17%, ROW 4%. Notably international for a German mid-cap.

Key Device Maker Customers

Company	Ticker	Compound Semi Revenue	Aixtron Relevance
Infineon	IFX:GR	~$2-3B (SiC+GaN)	Largest European customer. EUR 2B Kulim fab. Co-developing 800V systems with Nvidia. Targeting 300mm GaN.
STMicroelectronics	STMPA:FP	~$1-2B (SiC)	200mm SiC production in Catania. Major G10-SiC customer.
onsemi	ON	~$1-2B (SiC)	SiC-focused; vGaN R&D
Texas Instruments	TXN	Growing GaN	2022 Supplier Excellence Award to Aixtron. 4x GaN expansion. Nvidia 800V partner.
EPC	Private	~$50-80M (GaN)	Pioneer eGaN FETs. Nvidia 800V partner. 6kW 800V-to-12.5V converter.
Navitas	NVTS	~$100M (GaN+SiC)	Pure-play WBG IC company
Innoscience	Private	~$100M+ (GaN)	World's largest GaN IDM (31% device share). China-based.

Industry Map -- Key Players Beyond Equipment

Understanding the broader industry ecosystem matters because these companies are Aixtron's customers. Their investment cycles drive Aixtron's order intake.

Epitaxy Equipment (Aixtron's layer):

Company	Ticker	Revenue (est.)	Market Cap	Key Position
Aixtron	AIXA:GR	EUR 557M (2025)	EUR 1.96B	77% MOCVD, 27-40% SiC CVD
Veeco	VECO	~$700M	~$2.5B	~25-28% MOCVD; MBE + ion beam
AMEC	688012.SS	~RMB 6B	~$15B	~20-25% MOCVD (>70% China LED)

Device Makers (IDMs -- Aixtron's primary customers):

Company	Ticker	Compound Semi Revenue	Key Products	Aixtron Relevance
Infineon	IFX:GR	~$2-3B (SiC+GaN)	CoolSiC MOSFETs, CoolGaN HEMTs	EUR 2B Kulim fab; co-developing 800V with Nvidia; 300mm GaN target
STMicroelectronics	STMPA:FP	~$1-2B (SiC)	SiC MOSFETs (Tesla supplier)	200mm Catania fab; major G10-SiC customer
onsemi	ON	~$1-2B (SiC)	EliteSiC MOSFETs; vGaN R&D	SiC expansion; vertical GaN could be future demand
Rohm	6963:JP	~$1B (SiC)	SiC MOSFETs and diodes	Japanese automotive supply chain
Wolfspeed	WOLF	~$800M (all SiC)	SiC substrates + devices	Post-Ch.11; 200mm targeted but execution uncertain
Texas Instruments	TXN	Growing GaN	GaN power ICs	Awarded Aixtron Supplier Excellence (2022); 4x GaN expansion; Nvidia partner

Fabless / Design (emerging GaN specialists):

Company	Ticker	Focus	Revenue	Notes
Navitas	NVTS	GaN + SiC ICs	~$100M	Pure-play WBG IC company; Nvidia 800V partner
EPC	Private	GaN power	~$50-80M	Pioneer in eGaN FETs; CEO stated Nvidia collaboration for "gigawatt scale" AI factories
Innoscience	Private	GaN HEMTs	~$100M+	World's largest GaN IDM (31% device share); China-based
Transphorm	Acquired by Renesas (2024)	GaN power	N/A	Validates trend of majors acquiring GaN capabilities

Epiwafer Foundries (outsource epitaxy using Aixtron tools):

Company	Ticker	Focus	Notes
IQE	IQE.L	III-V epiwafers (GaAs, InP, GaN)	Largest independent epiwafer foundry
WIN Semiconductors	3105.TW	GaAs/GaN foundry	Major RF/5G foundry
VPEC	Private	InP epiwafers	Key for optical interconnects

The device maker capex cycle is the primary driver of Aixtron's order intake. When these companies invest in new fab capacity (as Infineon is doing at Kulim and STMicro at Catania), they order Aixtron tools. When they pause investment (as most are doing during the current SiC overcapacity), Aixtron's orders decline. Understanding the investment cadence of these customers is essential for timing the cycle.

SiC Supply Chain Deep-Dive

Understanding the SiC ecosystem matters because it is Aixtron's largest near-term revenue headwind and its biggest medium-term recovery catalyst.

The SiC overcapacity problem: The industry massively over-invested in SiC capacity during 2021-2024, driven by EV enthusiasm. Then EV adoption slowed, Chinese SiC substrate prices collapsed, and Wolfspeed filed Chapter 11. The result: upstream SiC substrate utilization is ~50%, device-level utilization is ~70%, and nobody is ordering new epitaxy equipment.

SiC substrate market dynamics:

Chinese SiC substrate makers (TanKeBlue, SICC) have driven 6-inch wafer prices from ~$1,500 to ~$400-500. This is restructuring the entire SiC economics. The 200mm transition (which offers ~85% more chips per wafer) further complicates the capacity picture because existing 150mm equipment becomes semi-obsolete.

Western IDM investment trajectory (the bull case for recovery):

Infineon Kulim, Malaysia: EUR 2B fab for 200mm SiC mass production (entered 2025). Potential for EUR 750M+ in CVD tool orders as capacity ramps.
STMicro Catania, Italy: 200mm SiC production began Q4 2025. Scaling to 15,000 wafers/week by 2033.
Wolfspeed: $5B North Carolina fab (pre-bankruptcy plan). 200mm targeted. Emerged from Chapter 11 Sept 2025 but execution uncertain.
Rohm: Expanding SiC capacity for Japanese automotive supply chain.
onsemi: Investing in SiC fab capacity; also developing vertical GaN as a potential SiC alternative for higher voltages.

The recovery thesis: these Western IDM investments represent massive potential tool orders that have been deferred, not cancelled. When utilization recovers (expected 2027-2028 as EV demand catches up and 200mm migration accelerates), equipment orders should inflect upward. Aixtron's G10-SiC is winning 50-60% of new 200mm business.

The SiC/GaN interaction: SiC overcapacity is actually accelerating GaN interest. Device makers looking to fill capacity gaps and diversify away from SiC-only strategies are investigating GaN for medium-voltage applications. This is particularly true in China, where GaN production is less constrained by US export controls than advanced SiC. The near-term SiC headwind may be creating a medium-term GaN tailwind.

Cyclicality Pattern -- Historical Context

Aixtron's revenue has ranged from EUR 210M (2016 trough) to EUR 634M (2024 peak) over the past decade. The LED-driven boom/bust of 2010-2013 was far more severe (EUR 1B+ peak to EUR 240M trough) because LED was a single end-market.

Today's business is more diversified across power electronics (GaN + SiC), optoelectronics, and LED/Micro LED, with low correlation between end markets. This should make cycles shallower going forward. Historical cycle duration: 2-4 years peak-to-peak.

Revenue by cycle:

Period	Type	Revenue Range	Driver
2010-2011	Peak	EUR 750M-1.08B	LED investment boom
2012-2013	Trough	EUR 227-240M	LED bust (-77% peak-to-trough!)
2014-2016	Recovery	EUR 195-283M	Slow, grinding recovery
2017-2020	New cycle	EUR 260-310M	Base-building; compound semi starts
2021-2024	Upcycle	EUR 429-634M	SiC + GaN + opto boom
2025-2026	Downturn	EUR 520-557M	SiC overcapacity; -18% peak-to-trough
2027+	Expected recovery	EUR 600-800M+	SiC recovery + GaN inflection

The key difference between the current cycle and the 2010-2013 disaster: diversification. The LED bust wiped out 77% of peak revenue because LED was the only end market. The current downturn has produced only an 18% peak-to-trough decline because weakness in SiC is partially offset by strength in optoelectronics. The business is structurally more resilient than it was a decade ago.

The SiC overcapacity problem is the primary cyclical headwind. Upstream SiC utilization is ~50%, which means fabs have substantial idle capacity before they need to order new equipment. The recovery depends on end-market demand (primarily EVs and industrial power) absorbing that capacity, which most analysts expect by 2027-2028.

Dilution Risk Assessment

Minimal. This is one of the cleanest capital structures in European semi equipment:

Share count flat at 113.5M for years. No equity raises. No ATM programs.
The prior stock option program (Conditional Capital II 2012) expired December 31, 2024. Only 20,900 shares remain authorized under that program.
No new stock option program was launched in FY2025.
SBC totals EUR 3.6M or 0.6% of revenue -- negligible.
Authorized capital: 41.5M shares for potential convertible bonds (expires May 2027) and regular authorized capital -- but with EUR 225M cash and zero debt, these are extremely unlikely to be used.
Cash flow sufficiency: After the Innovation Center investment is complete, Aixtron is solidly FCF-positive and self-funding. No capital raise needed under any reasonable scenario.
No convertible bonds outstanding. No warrant overhang.

Key-Person Risk

Moderate. Dr. Felix Grawert (CEO since 2017, contract through 2030) is the key figure. He led the G10 platform development, the Innovation Center construction, and the company through a full cycle. Loss of Grawert would be meaningful.

Mitigants:

6-member Executive Committee provides bench depth
CFO Danninger is established (contract through 2029)
Supervisory Board has documented succession planning
The company's competitive position is primarily structural (IP, switching costs, installed base), not personality-driven
Both executives are on very long-term contracts, reducing near-term departure risk

Near-Term Event Calendar

Event	Date	Significance
Q1 2026 results	April 30, 2026	High -- first read on whether FY2026 guidance is achievable. Q1 guided at ~EUR 65M (+/- EUR 10M), which is weak.
AGM	May 13, 2026	Medium -- management commentary on recovery timeline. Dividend vote (EUR 0.15 proposed).
Q2 2026 results	Late July 2026	High -- sequential trend matters. Any positive surprise in orders would be significant.
Nvidia GTC (if held H2 2026)	TBD	Medium-high -- 800V HVDC deployment updates, partner ecosystem expansion
Any guidance revision	Ad hoc	Very high -- positive or negative guidance revision would move the stock materially

Any guidance revision is a binary event for the stock. A positive revision (raising revenue guidance) would likely trigger 10-15% upside. A negative revision would trigger 10-20% downside. The stock is a coiled spring waiting for direction.

Key Questions to Keep Tracking

AMEC's SiC/GaN MOCVD progress: Any customer qualifications? Production tool shipments to Western IDMs?
NVIDIA 800V HVDC timeline: Confirmed for 2027? Which GaN/SiC device makers are qualified?
SiC utilization recovery: When does upstream utilization cross 70%? (Triggers equipment orders)
Aixtron 300mm Hyperion: Pilot line results? Customer commitments? Timeline to volume?
GaN power automotive design wins: Which OEMs adopting GaN for OBC/DC-DC?
Optoelectronics revenue durability: Structural AI demand or one-cycle spike?
Micro-LED transfer yield: Has anyone achieved six-nines? When does MOCVD demand materialize?

Monitoring Sources

Source	Coverage	Frequency
Aixtron IR (aixtron.com/en/investors)	Earnings, order intake, presentations	Quarterly
Semiconductor Today	Compound semi news, earnings	Daily
Compound Semiconductor Magazine	Technical + industry	Monthly
Yole Group reports	Market sizing, technology trends	Quarterly
TrendForce	SiC/GaN market data, pricing	Monthly
Kerrisdale Capital	Aixtron thesis updates	Ad hoc
Jefferies / Barclays semi equipment research	Analyst coverage	Quarterly

Key conferences:

Event	Focus	Timing
SEMICON Europa	European semiconductor	November
Electronica (Munich)	Power electronics	Biennial (even years)
CS ManTech	Compound semiconductor manufacturing	May
PCIM Europe	Power conversion	May
IEDM	Semiconductor devices (academic)	December
NVIDIA GTC	AI infrastructure, 800V HVDC updates	March (typically)

Best independent monitoring sources for the compound semi equipment thesis:

Source	Why It Matters
Aixtron quarterly order intake	The single most important leading indicator. Book-to-bill >1.1x sustained = cycle turning.
SiC fab utilization data (Yole, TrendForce)	When upstream utilization crosses 70%, equipment orders follow within 6-12 months.
Wolfspeed/Infineon/STMicro capex announcements	These are the major SiC tool buyers. Their investment plans drive Aixtron's order book.
NVIDIA 800V partner ecosystem updates	Each new partner announcement validates the GaN thesis.
GaN power device revenue (Yole quarterly)	Device market growth drives equipment demand with a 6-12 month lag.
800G+ optical transceiver shipment data	Leading indicator for optoelectronics equipment demand.
AMEC quarterly reports (688012.SS)	Monitor for any SiC/GaN MOCVD customer qualification announcements.

Why Aixtron Over Direct Device Makers -- The Picks-and-Shovels Argument

This is the core strategic reason to prefer Aixtron over Infineon, STMicro, onsemi, or any individual device maker in the compound semiconductor theme.

Device makers face customer-specific risk. If Tesla switches SiC suppliers from STMicro to onsemi, STMicro loses that revenue. If Hyundai delays 800V EV adoption, Infineon's SiC revenue gets pushed out. If Chinese EV makers in-source SiC, Western device makers lose share.

Aixtron wins regardless of which device maker succeeds. If Infineon wins in SiC, they buy Aixtron tools. If STMicro wins, they buy Aixtron tools. If a Chinese startup wins, they buy Aixtron tools. If GaN displaces SiC in some applications, Aixtron sells GaN tools instead of SiC tools. The equipment maker captures the entire growth of the underlying market without bearing customer-specific, end-market-specific, or geography-specific risk.

The installed base creates a structural advantage. Every tool sold generates 10-20 years of aftermarket revenue. Every qualification cycle deepens customer dependency. The installed base grows with every equipment cycle and never contracts (tools have long useful lives). This compounding effect means Aixtron's competitive position strengthens over time even if individual equipment cycles are volatile.

The trade-off: You pay for this structural advantage with higher cyclicality. Equipment demand is more volatile than device demand because small changes in end-market growth translate to large changes in capacity addition plans. A device maker growing 10% might not need new equipment; a device maker growing 20% suddenly needs to double their tool orders. This amplifier effect creates the volatility that creates the entry opportunities.

Data Confidence Notes

Category	Confidence	Notes
Aixtron financials 2021-2025	Confirmed	Official IFRS filings
GaN power market share (70-90%)	Estimated	Kerrisdale; not independently audited
SiC market share (~27-40%)	Estimated	Industry estimates vary
Market sizing (MOCVD, SiC, GaN)	Estimated	Market research firms; wide ranges
ASP estimates	Estimated	Not publicly disclosed
AMEC capabilities assessment	Estimated	Based on available reporting
Technology roadmap timelines	Estimated	Company + industry consensus
Nvidia 800V HVDC supply chain	Confirmed (partnerships) / Estimated (revenue impact)	GTC 2025 + analyst models

Implied Expectations at Current Price

At EUR 23 / ~35x 2026E P/E, the market is pricing in:

Continued trough through 2026
Modest recovery starting 2027
No credit for NVIDIA 800V HVDC catalyst, micro-LED optionality, or 300mm GaN

What Kerrisdale models: EUR 1.4B+ revenue by 2029, EUR 2.00 EPS. At 33x P/E = EUR 66. This requires GaN power to triple and SiC equipment market to double through the cycle. The math works directionally if the AI data center thesis plays out, but it is aggressive on timing.

What BofA models: ~50% of revenue from AI-related applications by 2027, justifying a structural re-rating from cyclical to growth multiple. They raised FY2027 EBITDA estimates +12.8%.

What consensus shows: 4-6 buy ratings against 7-8 holds and zero sells. Price targets range EUR 10.90-27.00. Average target: EUR 17-20 depending on when you snapshot it. The consensus is cautious -- majority Hold with targets near the current price.

Historical Valuation Context

Aixtron has traded at 10-25x forward P/E through the cycle, with trough valuations around 10-15x and peak valuations around 20-25x. The 5-year ranges: P/E 15-55x, EV/EBITDA 10-35x, EV/Revenue 2-8x.

At EUR 17.31 and 3.1x EV/Revenue, the market is pricing in roughly flat-to-modest revenue growth from current trough levels. If revenue recovers to EUR 650M+ (above the FY2024 peak) by FY2028, the stock is meaningfully undervalued. If the trough extends and revenue stays below EUR 600M, the current price is fair to slightly expensive.

Simple DCF sanity check: Normalized FCF of EUR 90M, growing at 8% long-term, discounted at 9% = terminal value of ~EUR 9B. Present value of next 10 years suggests fair value of EUR 22-28 per share. At EUR 17.31, there is 25-60% upside to DCF fair value depending on assumptions. At EUR 23, the margin of safety narrows but still offers upside to the base case.

Would I buy at 10-15% higher (EUR 19-20)? Yes, but with lower conviction. The optimal entry is EUR 15-17; EUR 12-14 would be a strong entry.

Investment Thesis Summary -- The One-Paragraph Version

Aixtron is the dominant supplier (77% MOCVD, 70-90% GaN power) of the deposition equipment required to manufacture every compound semiconductor chip. The stock is in a cyclical trough (FY2025 revenue -12%, FY2026 guided -7%) driven by SiC overcapacity and EV weakness, but the structural growth drivers are intact and accelerating: AI data center power conversion (Nvidia 800V HVDC), optical interconnects (800G+ transceivers), the 300mm GaN wafer transition, and the eventual SiC recovery. The balance sheet is fortress-grade (net cash EUR 225M, zero debt), governance is clean, and the competitive moat is deep (switching costs, installed base, process IP, market structure). At EUR 17-23, the market is pricing a cyclical equipment company at trough earnings and giving no credit for the structural AI thesis. If the GaN power inflection materializes as expected (2027-2028), the stock is dramatically mispriced. The risk is timing -- FY2026 will be ugly, and the cycle could extend. Scale in, do not go all-in.

Filing Review Note

Aixtron is a German SE that does not file with the SEC. Equivalent filings reviewed:

AIXTRON SE Annual Report 2025 (IFRS consolidated, management report, corporate governance statement)
KPMG AG unqualified audit opinion (FY2025)
WpHG voting rights notifications
No material litigation, no auditor changes, no going concern language, no restatements
No insider buying or selling above MAR thresholds in FY2025

Sources

AIXTRON SE Annual Report 2025 (290 pages, published February 25, 2026)
AIXTRON FY2025 Results Presentation
AIXTRON IR website
Kerrisdale Capital -- Aixtron Long Report (Nov 2025)
BofA Securities, Barclays, Jefferies, J.P. Morgan analyst reports
NVIDIA 800V HVDC Architecture developer blog
NVIDIA GTC 2025 partner announcements
Yole Group -- GaN power market sizing, SiC overcapacity analysis
TrendForce -- SiC/GaN market data
Irrational Analysis (Substack) -- independent coverage
Swiss Transparent Portfolio (Substack) -- contrarian bear case
Compound Semiconductor Magazine, Semiconductor Today
Imec 300mm GaN Power Electronics Program (Oct 2025)
Veeco Propel 300 order announcement (Nov 2025)
Texas Instruments 2022 Supplier Excellence Award
EPC CEO Alex Lidow statements on Nvidia collaboration

Topics

compound-semiconductors
mocvd
gan-power
silicon-carbide
ai-infrastructure
optical-interconnects
semi-equipment
ev-power-electronics
nvidia-800v-hvdc
300mm-wafer-transition
china-semiconductor-competition

Pages that link here (15)

IFX — Infineon Technologies NVTS — Navitas Semiconductor ON — onsemi (ON Semiconductor)TXN — Texas Instruments AI Infrastructure & Datacenter Optical Networking & Photonics Power Semiconductors & Power Electronics Semiconductor Capital Equipment & Test Foundry & Semiconductor Materials Semiconductors (cross-cutting / geopolitics)AIXTRON SE (AIXA) - Comprehensive Stock Pitch AIXTRON SE (AIXA) — Deep Research Stock Pitch The Aixtron Series | Part 1: Power Electronics The Aixtron Series | Part 2: The Optoelectronics Supercycle Aixtron (Frankfurt:AIXA) - Long Thesis

Open in Obsidian