Infineon Technologies AG (IFX.DE) - Deep Dive Research Report

Prepared 6 June 2026. All figures in EUR unless noted. Fiscal year ends 30 September; "FY2025" = the year to 30 September 2025.

Reporting calendar note

Infineon reports on a 30-September fiscal year. Working backwards from today (6 June 2026), the most recent reported period is Q2 FY2026 (quarter ended 31 March 2026), reported on 5-6 May 2026 - inside the 90-day window. The five concalls used in this report are:

1. What the company does

Infineon makes the semiconductors that move and manage electrical power, and the chips that let machines sense the physical world and run safely. If a current needs to be switched, converted, stepped up or down, measured, or controlled - inside a car, a wind turbine, an AI server rack, a phone charger, a heat pump, a factory robot, or a passport - there is a reasonable chance the chip doing it carries the Infineon name. This is not the glamorous, leading-edge logic that Nvidia or TSMC are famous for. It is the unglamorous, physics-heavy layer underneath: the transistors that handle hundreds of volts and hundreds of amps without melting, the microcontrollers that decide when a brake actuator fires, the sensors that tell an engine how much air is flowing.

The company was carved out of Siemens. Siemens spun off its semiconductor division as Infineon in 1999 and floated it on the Frankfurt and New York exchanges in 2000. That heritage matters: Infineon inherited Siemens' industrial and automotive DNA rather than the consumer-PC focus of its American peers, which is why even today the bulk of its revenue comes from cars and industrial equipment rather than computers and phones. Over the next two decades it made two transformational moves. It bought International Rectifier in 2015, which deepened its lead in power MOSFETs, and it bought Cypress Semiconductor in 2020 for roughly $9 billion, which bolted on microcontrollers, wireless connectivity, NOR flash memory and the PSoC programmable-chip family. The most recent move, completed in August 2025, was the $2.5 billion all-cash purchase of Marvell's automotive Ethernet business (the "Brightlane" portfolio), which gives Infineon the high-speed in-car networking piece it lacked.

The core value proposition is efficiency and reliability at the point where electricity does work. A power semiconductor that wastes 1% less energy when converting voltage saves a data-center operator millions in electricity and cooling over the life of the equipment, and extends an electric car's range. A microcontroller that has been qualified to automotive safety standards lets a carmaker ship a brake-by-wire system without fear of a recall. These are products where being slightly better, and being trusted, compounds into market leadership.

What makes the product hard is a combination of materials physics and qualification. Power devices are built not on the leading-edge nanometre nodes but on specialised processes using silicon, and increasingly silicon carbide (SiC) and gallium nitride (GaN), where the engineering challenge is handling heat and high voltage rather than cramming in transistors. Infineon runs its own fabs to control this. And in automotive and industrial markets, a chip must survive a decade of vibration, temperature swings and electrical stress, then pass years of customer qualification testing before it ever ships - which is exactly what locks competitors out once Infineon is designed in.

A concrete example: when an electric vehicle brakes and recovers energy, the traction inverter has to flip the high-voltage battery's DC into the AC the motor wants, switching thousands of times a second at 400 or 800 volts. Infineon supplies the SiC power module that does this, the gate-driver chip that controls it, the current sensor that measures it, and the AURIX safety microcontroller that supervises the whole thing. One car, multiple Infineon chips, each one qualified and designed in years before the vehicle launches.

2. Business segments

Through FY2025 Infineon reported in four segments. Management announced on the Q2 FY2026 call (6 May 2026) that from Q4 FY2026 these four will be consolidated into three - Automotive (ATV), Power Systems (PS) and Edge Systems (ES). The four-segment view below reflects how the business has actually been reported and measured over the five concalls in this report; the reorganisation is covered in Section 7.

Automotive (ATV) - roughly 50% of group revenue

This is the heart of Infineon. It sells microcontrollers, power semiconductors and sensors into cars, and it is the largest automotive semiconductor supplier in the world - around 13.5% of the automotive chip market, ahead of NXP, with roughly 32% of the automotive microcontroller market (ahead of Renesas) and a similar lead in automotive power.

The core capability is the marriage of three things almost no rival has under one roof: the AURIX safety microcontroller family that runs critical functions, the power devices (silicon and SiC) that drive motors and manage the high-voltage battery, and the sensor portfolio that feeds the car data. The newest piece, from the Marvell acquisition, is automotive Ethernet - the high-speed data backbone of the "software-defined vehicle." Building a chip that a carmaker will trust to fire an airbag or steer a car takes years of qualification, which is why this segment's customer relationships are measured in vehicle programme lifetimes, not quarters. It exists as its own unit because automotive has its own safety standards (ISO 26262), its own decade-long design cycles, and its own customer base of OEMs and tier-one suppliers. Within the segment Infineon wins on breadth (it can supply the whole electrified powertrain) and on its installed base of design-ins; it competes against NXP, Renesas, STMicroelectronics, Texas Instruments and onsemi. Management treats this as the strategic anchor - the place where electrification and the software-defined vehicle create a structurally rising chip content per car.

Power & Sensor Systems (PSS) - roughly 32% of group revenue

PSS is the second-largest segment and, right now, the most exciting one because it owns the AI data-center power story. It makes the power-conversion semiconductors that take grid electricity down through the voltage steps to feed an AI server's GPUs, plus RF chips, MEMS microphones, and power devices for chargers and consumer electronics. The core capability is high-efficiency power conversion using cutting-edge silicon MOSFETs, SiC and GaN. As AI racks have jumped from a few kilowatts to tens of kilowatts and now toward 800-volt DC architectures, the power-delivery problem has become acute, and Infineon's content per rack has risen sharply. It competes here with Monolithic Power Systems, Texas Instruments, Analog Devices, Renesas, Vicor and Delta. Management talks about PSS as the near-term growth engine - this is where the "extremely robust" AI demand and the raised guidance show up.

Green Industrial Power (GIP) - roughly 9% of group revenue

GIP sells the heavy-duty power modules - largely IGBTs (insulated-gate bipolar transistors) and SiC modules - that go into solar inverters, wind turbines, electricity grids, industrial motor drives, trains and EV charging infrastructure. The capability is handling very high power and voltage reliably for 20-plus years in harsh environments. It exists separately because its customers (renewable-energy and industrial equipment makers) and its product physics differ from consumer and automotive. It is cyclical and was the weakest segment through the FY2025 industrial downturn, but it is geared to the energy transition. Competitors include STMicroelectronics, onsemi, Mitsubishi Electric, Fuji Electric and Wolfspeed (in SiC).

Connected Secure Systems (CSS) - roughly 9% of group revenue

CSS is the security-and-connectivity business, much of it inherited from Cypress. It makes secured microcontrollers for payment cards, passports, SIM cards and IoT devices, plus the PSoC programmable microcontrollers and Wi-Fi/Bluetooth connectivity chips. The core capability is hardware security - chips that resist physical and electronic attack and carry government and payment-industry certifications (Common Criteria, EMVCo). It exists separately because the security-certification world is its own regulatory universe. Infineon is a leader in secured elements; competitors include NXP, STMicroelectronics and Thales. Management frames CSS as a steadier, IoT-and-security-leveraged business rather than a primary growth bet.

3. Products and business detail

The power semiconductor portfolio is the spine. It spans discrete MOSFETs and IGBTs, integrated power stages, gate drivers, and complete power modules. The flagship lines include OptiMOS (silicon MOSFETs), CoolMOS (high-voltage superjunction MOSFETs), CoolSiC (silicon-carbide MOSFETs and modules) and CoolGaN (gallium-nitride devices). These go everywhere a voltage needs converting: AI server power-supply units, EV traction inverters, solar inverters, fast chargers, motor drives. Infineon holds roughly 17-19.5% of the global power semiconductor market - more than double the next competitor - and around 20% of the SiC market and roughly 10% of the GaN market, the two fast-growing wide-bandgap materials.

Microcontrollers are the second pillar. AURIX is the automotive safety microcontroller used for powertrain, braking, steering and domain control. PSoC and the Cypress-derived MCU families serve industrial and IoT applications. AURIX in particular is deeply entrenched: it is qualified into vehicle platforms that ship for years.

Sensors and RF: MEMS microphones (XENSIV), magnetic and current sensors, radar chips (including 60GHz radar for presence detection and automotive), and RF front-ends. Security: secured elements for payment, government ID and SIM, plus TPM security controllers. Connectivity: Wi-Fi and Bluetooth from the Cypress portfolio. Automotive Ethernet: the newly acquired Brightlane PHYs and switches, expected to contribute roughly $225-250 million of revenue in FY2026.

Manufacturing is largely in-house and concentrated in the West, which Infineon presents as a geopolitical-resilience advantage. Front-end fabs sit in Germany (Dresden, Regensburg, Warstein), Austria (Villach, home to a 300mm thin-wafer power fab and SiC), and Malaysia (Kulim, a 200mm SiC fab). The headline capacity project is the Dresden "Smart Power Fab," a roughly €5 billion 300mm facility for power and analog/mixed-signal chips, whose opening and production ramp management has been pulling forward (guided to open in summer 2026). The combination of 300mm power manufacturing (which lowers unit cost versus the 200mm industry norm) and in-house SiC/GaN is a structural cost-and-supply edge.

Geographies: Infineon sells globally, with major exposure to European and Asian automotive and industrial customers and a fast-growing North American AI data-center customer base (Amazon, Google, Cisco, Dell, HP and others have been cited as power-delivery customers). Distribution runs through both direct OEM/tier-one relationships and a large distribution channel for the long tail of industrial and consumer customers.

4. Customers

Infineon's customers fall into a few distinct camps, each buying for different reasons and on different timelines.

Automotive OEMs and tier-one suppliers (Bosch, Continental, ZF, plus carmakers directly) are the largest. Inside these accounts the buyer is an engineering and platform team, not a procurement clerk: a chip is selected because it has been qualified into a specific vehicle programme, and that decision is made two to four years before the car ships. The criteria are functional safety certification, automotive-grade reliability, supply security, and the breadth to source many parts from one vendor. The sales cycle is long and the switching cost is high - once an AURIX microcontroller or a SiC traction module is designed into a platform, replacing it means re-qualifying the whole subsystem, which carriers rarely do mid-cycle. That installed-base lock-in is why Infineon's automotive share is sticky.

AI data-center and cloud customers are the fastest-growing camp. Here the buyers are the hyperscalers and the server/power-supply makers (named on calls and in coverage: Amazon, Google, Cisco, Dell, HP). They buy on power-conversion efficiency, density (watts per cubic inch), and the ability to keep pace with rapidly rising rack power. The design cycle is faster than automotive but the qualification bar is still real, and Infineon's content per rack rises with each generation.

Industrial and renewable-energy customers (solar/wind inverter makers, grid equipment, industrial automation) buy GIP modules on reliability and power density, with multi-year design lives. Payment, government and IoT customers buy CSS security chips on certification and trust.

Concentration is low at the customer level - Infineon serves thousands of customers across cars, industry and consumer, with a large distribution tail - which makes revenue more diversified than a single-end-market chip company. The risk is end-market concentration (automotive cyclicality) rather than single-customer concentration. Contract structure is a mix of design-win-driven multi-year programme business (automotive, industrial), framework supply agreements, and shorter-cycle distribution and consumer business. The automotive design-win backbone gives multi-year revenue visibility; the consumer and parts of the AI business are more order-driven.

5. Competitive landscape

Infineon sits at the top of the power-semiconductor world and near the top of the automotive-semiconductor world, but it competes against a different set of rivals in each segment.

In power semiconductors broadly it is the clear number one (~17-19.5% share, roughly double the runner-up) and is widening the gap, helped by its 300mm power manufacturing and in-house SiC/GaN. In automotive it leads on overall share and microcontrollers, with NXP and Renesas as the closest microcontroller rivals and STMicroelectronics, onsemi and Texas Instruments competing across power and analog. In AI data-center power delivery the field is more contested and newer: Monolithic Power Systems is the incumbent in on-board voltage regulation, while Texas Instruments, Analog Devices, Renesas, Vicor and Delta all compete for the rack power tree; Infineon's edge is the breadth of its power-device portfolio (silicon, SiC, GaN) feeding the high-power stages. In SiC specifically it competes with STMicroelectronics, onsemi, Wolfspeed and ROHM; in GaN with onsemi, Navitas and Power Integrations.

The barriers to entry are high but not absolute. They come from (1) manufacturing - owning advanced power fabs, especially 300mm and SiC, requires billions of capital and years of process learning; (2) qualification lock-in - once designed into a car or industrial platform, a part is rarely swapped; and (3) portfolio breadth - the ability to sell the whole power-and-control content of a system. The main structural risk to these barriers is wide-bandgap disruption: SiC and GaN reset some of the process advantages and have drawn in well-funded specialists (Wolfspeed, Navitas) and Chinese entrants subsidised by Beijing's push for domestic power-chip self-sufficiency.

Market caps are approximate peer-size references as of June 2026 and move daily.

6. Industry

Demand for Infineon's products is driven by three secular forces and one cycle. The secular forces are electrification (electric vehicles and the rising chip content of all vehicles), the energy transition (solar, wind, grid, industrial efficiency), and now AI data-center power (the electricity appetite of GPU clusters). The cycle is the broader semiconductor and industrial/automotive inventory cycle, which ran through a downturn in 2024-2025 and, on Infineon's most recent commentary, is recovering.

The power semiconductor market was valued at roughly $55.7 billion in 2025 and is forecast to grow to around $77 billion by 2031 (industry-research estimates vary; one forecast puts the market at $101.7 billion by 2035). Within it, the wide-bandgap materials are the high-growth slices: SiC growing well over 20% annually and GaN faster still. Infineon's stated AI-power addressable market alone is €8-12 billion by the end of the decade, against the ~€1.5 billion of AI revenue it guides for FY2026.

Infineon sits at the device layer of the supply chain - it designs and largely manufactures its own chips, then sells to module makers, tier-ones, OEMs and hyperscalers. It depends on wafer materials (including SiC substrates), some outsourced foundry capacity for non-power logic, and back-end assembly partly in Asia.

The regulatory and policy environment cuts both ways. Tighter automotive safety and emissions rules and decarbonisation mandates lift chip content; export controls, tariffs and the US-China technology conflict create uncertainty (management repeatedly flagged tariff and US-dollar headwinds through FY2025); and China's drive for domestic power-semiconductor self-sufficiency is a long-term competitive threat. Government chip subsidies (EU Chips Act, US CHIPS Act) partly fund Western fab expansion including Dresden.

Cyclicality is real: automotive and industrial demand swings with the economy and with customer inventory, and Infineon went through a clear soft patch in 2024-2025. The offsetting feature is that the structural drivers (chip content per car, energy transition, AI power) raise the trend line through the cycle. The current tailwind is the AI-power surge and the cyclical recovery in autos/industrial; the lingering headwind is a weak US dollar (which Infineon flags as a multi-point drag on reported revenue) and trade/tariff uncertainty.

7. Growth triggers

Drawn strictly from the five concalls in this report. Forward-looking only.

• AI data-center power revenue scaling to ~€1.5 billion in FY2026 and ~€2.5 billion in FY2027 - described by the CEO as a tenfold increase in AI sales within three years. (Raised on Q4 FY25 call, 12 Nov 2025; reaffirmed and extended to the FY27 target on Q1 FY26 call, 4 Feb 2026 - repeated trigger.)

  "a tenfold increase in our AI sales within just three years" - Jochen Hanebeck (Q1 FY26 call, 4 Feb 2026)

• AI-power addressable market of €8-12 billion by the end of the decade - the longer-runway framing behind the near-term target. (Q4 FY25 call, 12 Nov 2025.)

• €500 million of AI-related capital expenditure pulled forward into FY2026, lifting planned FY26 investment to around €2.7 billion to meet surging demand. (Q1 FY26 call, 4 Feb 2026.)

• Dresden "Smart Power Fab" opening in summer 2026 with immediate ramp of the advanced MOSFET generations needed for data-center power. (Q1 FY26 call, 4 Feb 2026; reiterated Q2 FY26 call, 6 May 2026 - repeated trigger.)

• Marvell automotive-Ethernet (Brightlane) integration, completing Infineon's software-defined-vehicle content with high-speed in-car networking; expected to add roughly $225-250 million of revenue in FY2026. (Flagged Q3 FY25 call, 5 Aug 2025; deal closed 14 Aug 2025.)

• Automotive design wins named on the call: a 48-volt-based control system at a major American automaker and a steer-by-wire system at a German premium manufacturer, with continued AURIX microcontroller platform adoption. (Q1 FY26 call, 4 Feb 2026.)

• Reorganisation into three business units (Automotive, Power Systems, Edge Systems) from Q4 FY2026, presented as a structural simplification to sharpen focus on power and edge growth. (Q2 FY26 call, 6 May 2026.)

• Raised full-year FY2026 guidance from "moderate" to "significant" revenue growth, to above €16 billion with a segment margin around 20%, with Q3 FY26 revenue guided to roughly €4.1 billion on broad-based recovery and AI strength. (Q2 FY26 call, 6 May 2026 - new.)

  "Entering the second half, we expect growth to significantly outpace previous expectations." - Jochen Hanebeck (Q2 FY26 call, 6 May 2026)

8. Key risks

Automotive cyclicality and EV-demand wobble. Roughly half of revenue rides on the car cycle, and a slower-than-expected EV transition or a recession in European/Chinese auto production hits Infineon directly through inventory destocking. The mechanism is well-rehearsed: when carmakers and tier-ones cut builds, they first run down chip inventory, so Infineon's orders fall faster than end demand. This is a high-probability, moderate-magnitude drag and was visible in the FY2024-2025 softness. It is structural, not catastrophic.

US-dollar weakness. Infineon reports in euros but sells heavily in dollars. Management has repeatedly quantified the currency drag - a weaker dollar shaved several percentage points off reported FY2025 growth, and the raised FY26 guidance is in spite of an adverse currency assumption. This is a high-probability, recurring headwind that depresses reported (not constant-currency) growth.

"ongoing tariff uncertainties and weaker US dollar" - flagged in the Q3 FY25 results (5 Aug 2025), where management noted currency alone reduced reported growth by roughly 6 percentage points.

Tariffs and the US-China technology conflict. Tariff uncertainty was a stated reason for trimming FY25 guidance on the Q2 FY25 call. The mechanism runs through both demand (customers delaying orders) and supply-chain cost. Moderate probability, hard to size.

Chinese power-semiconductor competition. Beijing's drive for domestic self-sufficiency in power chips is funding local champions in IGBTs and increasingly SiC. Over time this could erode Infineon's pricing and share in the large China market, the world's biggest for EVs and industrial power. Lower-probability near-term, but a genuine long-term threat to the moat.

AI-power execution and concentration risk. The bull case now leans heavily on AI data-center power scaling 10x in three years. If the AI capex cycle cools, or if competitors (MPS, TI, ADI, Navitas with Nvidia design-ins) win the next-generation 800VDC and vertical-power sockets, the most exciting growth line disappoints. The €500m capex pull-forward raises the cost of being wrong. Moderate probability, high relevance because it is the swing factor in the current narrative.

Wide-bandgap technology disruption. SiC and GaN reset some of Infineon's silicon process advantages and invite specialists. Wolfspeed (financially stressed) and Navitas show both the opportunity and the volatility of the pure-plays; the risk to Infineon is that it loses the materials race in a category it does not yet dominate as it does silicon. Moderate probability, moderate magnitude.

Capex and fab-ramp timing. Dresden is a multi-billion-euro bet whose returns depend on filling 300mm capacity at the right time in the cycle. A ramp into a soft market would pressure margins through under-utilisation - the classic semiconductor capacity trap.

9. Walk the talk

Concalls used: Q2 FY25 (8 May 2025), Q3 FY25 (5 Aug 2025), Q4 FY25 (12 Nov 2025), Q1 FY26 (4 Feb 2026), Q2 FY26 (5-6 May 2026). The most recent is within 30 days of this report.

Start with Q2 FY25 (May 2025). Management had to play defense. Revenue came in at €3.591 billion, up sequentially, but the team trimmed the full-year outlook to "slightly down" versus the prior year, blaming tariff uncertainty and a weaker dollar. This was an honest, conservative move at a moment of macro fog - and crucially, they did not abandon the AI and automotive growth narrative, they just dialed back the near-term numbers. They guided Q3 to about €3.7 billion.

"revenue is expected to reach €3.7 billion" for Q3 FY25 (Q2 FY25 results, 8 May 2025)

Move to Q3 FY25 (Aug 2025). Revenue landed at €3.704 billion - essentially bang on the €3.7 billion guide - and the segment margin of 18.0% came in above the company's own forecast. They guided Q4 to "around €3.9 billion." So far, two consecutive quarters of delivering what they said, with margin beating. This is the profile of a conservative-but-accurate management team.

Q4 FY25 (Nov 2025). Revenue was €3.943 billion, slightly ahead of the ~€3.9 billion guide, and the strongest quarter of the fiscal year. Full-year revenue of €14.662 billion matched the "around €14.6 billion / slightly down" framing management had carried all year. The "slightly down" guidance given back in May was delivered almost exactly (down 2%). They then did the thing that separates credible management from cheerleaders: rather than over-claim, they raised the concrete AI-power target to ~€1.5 billion for FY26 and pointed to an €8-12 billion addressable market - a forward bet, but a numbered one. They guided Q1 FY26 to about €3.6 billion.

Q1 FY26 (Feb 2026). Revenue was €3.662 billion, just above the ~€3.6 billion guide. They reaffirmed FY26 guidance, then went further by pulling forward €500 million of AI capex and extending the AI target to ~€2.5 billion in FY27 - the "tenfold in three years" framing. They guided Q2 to ~€3.8 billion. Note the pattern: every quarterly guide so far had been met or modestly beaten.

Q2 FY26 (May 2026). Revenue was €3.812 billion, essentially in line with the ~€3.8 billion guide. The headline was the upgrade: full-year guidance lifted from "moderate" to "significant" revenue growth, to above €16 billion with a segment margin around 20%, and Q3 guided to ~€4.1 billion. This is the first clear upside revision of the cycle, and it came only after five straight quarters of hitting the number - not before.

The assessment is clear: this is a management team that does what it says. Across five consecutive quarters every revenue guide was met or modestly beaten, the full-year outlook was delivered almost to the euro, and the one big revision was an upgrade that came after the evidence was in, not a hopeful pre-announcement. The style is conservative-accurate. The single thing to watch is the AI-power target, which is the one place where management has put a large, ambitious, forward number on the table; everything else has been credibly under-promised and delivered.

10. Shareholder friendliness index

Dividends. Infineon has paid €0.35 per share in each of the last three fiscal years (FY2023, FY2024 and FY2025), proposed again at €0.35 and approved at the 19 February 2026 AGM. The dividend has been held flat - it rose from €0.27 (FY2021) to €0.32 (FY2022) to €0.35 (FY2023) and has not moved since. The FY2025 payout is roughly €456 million. The flatness reflects a deliberate choice to keep the cash dividend modest and steady while funding heavy capacity investment (Dresden, SiC) rather than a sign of stress; the payout ratio remains comfortably below earnings.

Buybacks and dilution. Infineon does not run a share-repurchase program. MoatMap's database shows zero buybacks over the trailing three years, consistent with the company's policy of returning cash through dividends and reinvesting the rest. Share count has drifted slightly upward over the period from equity-based compensation (the share grants visible in the insider data), so net share issuance is mildly dilutive rather than the count shrinking. Capital that could fund buybacks is instead going into the €2.5 billion Marvell acquisition and the multi-billion-euro fab build-out.

Verdict: Neutral - a steady, unspectacular flat dividend with no buybacks and mild option dilution, because management is prioritising reinvestment (fabs and M&A) over capital return.

11. Insider activities

Venue: Frankfurt/Xetra; insider transactions are reported under EU Market Abuse Regulation Article 19 (PDMR) via BaFin's Directors' Dealings register. The data below uses MoatMap's database (market: DE) as the spine. The most recent filing in the block is dated 11 May 2026 - within the cross-check window; a check of the period found no materially newer BaFin filings beyond what is captured here.

The picture is dominated by a single corporate event. On 7 April 2026, every member of the Management Board (Vorstand) plus a Supervisory Board member received share allocations priced at €39.38 - a coordinated batch consistent with the vesting/settlement of a long-term incentive plan rather than open-market activity. These are classified "Other" (equity compensation), not directional buys or sells:

Buys. There are no open-market purchases in the last 12 months. No conviction-buy signal is present.

Sells. The only directional open-market trade is the 11 May 2026 sale of 10,001 shares by Supervisory Board member Peter Gruber for roughly €618,000 at €61.76 (BaFin Directors' Dealings, 2026-05-11). The reason is not disclosed in the filing. Coming roughly a month after Gruber received share grants on 7 April, the most likely read is a routine post-vesting monetisation/diversification rather than a view on the business, but the filing does not state a reason, so this is inference, not disclosure. The size - about €0.6 million by a non-executive director - is not large in the context of board-level holdings.

Net assessment. Activity is narrow and almost entirely non-directional: ten of eleven transactions are equity-compensation grants ("Other") settled in one coordinated April batch, and the only true open-market trade is one modest sell by a supervisory board member. There were no insider buys. This is neutral - the absence of open-market buying removes any bullish conviction signal, but a single small post-vesting board sale is too routine to count as a bearish signal. Worth noting only that the large April grants reset insiders' equity exposure upward, so executives remain meaningfully aligned even without open-market purchases.

12. Scenarios

Bull case. The AI-power surge proves to be the start of a multi-year content cycle, not a spike. Rack power keeps climbing toward 800-volt DC architectures, and Infineon's breadth across silicon, SiC and GaN lets it win not just the discrete power devices but larger slices of the rack power tree, holding off Monolithic Power Systems and the GaN pure-plays. AI revenue hits the €2.5 billion FY27 mark and keeps going toward the €8-12 billion addressable market. Meanwhile the automotive cycle turns up, the software-defined vehicle pushes chip content per car higher, and the Marvell Ethernet acquisition lets Infineon sell the full in-car networking-and-power stack. Dresden opens on time and fills quickly, lowering unit costs and lifting margins toward the high-20s. The dollar stabilises, removing the reported-growth drag. Infineon ends up as the dominant Western power-semiconductor supplier across the three biggest secular demand pools - cars, the grid, and AI compute - at once.

Base case. Management does roughly what it has guided. FY2026 revenue clears €16 billion with a segment margin near 20%, helped by the AI-power ramp and a gradual automotive/industrial recovery, partly offset by the weak dollar. AI revenue reaches around €1.5 billion this year and stays on track toward €2.5 billion next year. The reorganisation into three units lands without disruption. Dresden opens in summer and ramps through the year. The dividend stays at €0.35, no buyback appears, and capex stays elevated to feed the AI and SiC build-out. Infineon remains the clear power-semiconductor leader, growing modestly through the cycle with the structural drivers doing the heavy lifting and no single line dominating the outcome.

Bear case. The AI-power story disappoints - either the hyperscaler capex cycle cools faster than expected, or competitors win the next-generation power sockets, and the €2.5 billion FY27 target slips. At the same time the automotive recovery stalls on weak European and Chinese demand and a slower EV transition, leaving half the business in a prolonged destock. The weak dollar keeps eating reported growth. Chinese power-chip competitors, subsidised by Beijing, undercut Infineon on IGBTs and emerging SiC in the large China market, pressuring price and share. Dresden ramps into a soft market, so the new 300mm capacity sits underutilised and drags margins down through fixed-cost absorption - the classic capacity trap. The €2.5 billion Marvell deal underdelivers as software-defined-vehicle adoption proves slower than hoped. None of this is existential for a company with Infineon's installed base, but it would turn the current "significant growth" narrative back into the "slightly down, conserve and reinvest" mode of FY2025.

13. Further reading

• Energizing AI: Power Delivery Competition Heats Up - Vicor, MPS, Delta, ADI, Renesas, Infineon - SemiAnalysis, 2023-08-01 [paid]

A note on sources and limitations

• All five required concalls were located and used (Q2 FY25 through Q2 FY26); the most recent (Q2 FY26, 6 May 2026) is within the 90-day window.
• Segment revenue percentages are approximate; Infineon's primary-source PDFs (press releases and investor presentations) did not parse as machine-readable text, so segment splits are drawn from the aggregated commentary in the EQS/press summaries and the independent Rijnberk Invest Insights deep dive. Treat the 50/32/9/9 split as a close approximation rather than an exact FY2025 figure.
• Insider data is from MoatMap's BaFin-sourced database (market: DE), cross-checked for the recent window; no newer material filings were found.
• The only independent-analyst coverage located from the three target sources where Infineon is a named central player was a SemiAnalysis power-delivery piece that is both paywalled and older than 24 months; it is listed for reference only, metadata-only, with no paraphrase.

Sources:

• Infineon Q2 FY26 press release - raises full-year guidance | Gasgoo coverage
• Infineon Q1 FY26 results - successful start to FY2026 | Investing.com Q1 FY26 transcript
• Infineon FY2025 results press release | EQS-News FY2025 summary
• Infineon Q3 FY25 results - EQS-News
• Infineon Q2 FY25 results
• Marvell to sell Automotive Ethernet business to Infineon for $2.5bn
• Power semiconductor market size - Mordor Intelligence | GMInsights
• Rijnberk Invest Insights - Infineon deep dive
• Infineon dividend page | StockAnalysis dividend history

Would you like me to save this report to a file in the project, or generate any additional charts?