AT&S Austria Technologie & Systemtechnik AG (ATS.VI) - Deep Dive Research Report

Prepared 2026-06-17. Fiscal year ends March 31. Most recent published reporting period: Q3 FY2025/26 (nine months to December 31, 2025), reported February 2, 2026. Full-year FY2025/26 results are scheduled for June 29, 2026 and are not yet public as of this writing.

Section 1: What the Company Does

AT&S makes the thin, multi-layered boards that connect the chips inside almost everything electronic, and it makes the even smaller, far more precise boards that sit directly underneath the most advanced processors on the planet. There are two things to understand. The first is the printed circuit board (PCB): the green (or other-coloured) board you would see if you opened a smartphone, a hearing aid, or a car's control unit. It is the nervous system of a device, routing power and signals between components. The second, and the more important one for AT&S's future, is the IC substrate: a miniature, ultra-high-precision board that bridges a bare silicon chip and the main PCB. As chips get faster and hotter, the substrate becomes a bottleneck. It has to carry thousands of electrical connections at spacings measured in microns, dissipate heat, and keep signals clean at very high data rates. AT&S is one of only five companies in the world that can make the most advanced version of these substrates at scale.

The company was founded in 1987 in Leoben, in the Austrian province of Styria, where it is still headquartered. The pivotal moment in its history was 1994, when Hannes Androsch (a former Austrian finance minister and vice chancellor), together with Willibald Dörflinger and Helmut Zoidl, led a management buyout of AT&S out of the Austrian state holding company ÖIAG. At the time AT&S had around 940 employees. It listed on the Vienna Stock Exchange at the end of the 1990s and spent the next three decades expanding from a European PCB maker into a global supplier with around 13,500 to 14,000 employees, building large plants in China (Shanghai and Chongqing), India (Nanjangud), and most recently a multi-billion-euro IC substrate complex in Kulim, Malaysia. Androsch remained the controlling shareholder and chairman of the supervisory board until his death in December 2024, an event that matters enormously to this story (see Sections 9 and 11).

The core value proposition is precision at scale. Anyone can etch a simple circuit board. Very few can build a 20-plus-layer substrate, with copper traces finer than a human hair, that will survive being soldered to a $30,000 AI accelerator and run reliably for years in a data centre. The hard part is yield: getting a high enough fraction of these intricate boards to come out defect-free, consistently, in volume. That capability takes years and billions of euros of clean-room investment to build, and customers will not move a qualified product to an unproven supplier casually.

A concrete example: when a company like AMD designs a new server GPU, the silicon die cannot talk to the rest of the server on its own. It needs a substrate that fans out the chip's dense connections to the larger pitch of the motherboard, delivers clean power to billions of transistors, and handles the heat. AMD (or its packaging partner) gives AT&S the electrical and mechanical specification. AT&S develops the substrate, qualifies it through months of reliability testing, and then manufactures it in Chongqing or Kulim. The substrate ships to an assembly house where the die is mounted on top, and the finished package goes into an accelerator card. AT&S earns money on every one of those packages built.

"AI has definitely changed the speed of progress." - Peter Griehsnig, CTO, Q3 FY2025/26 call (February 2, 2026)

Section 2: Business Segments

AT&S reports through two business units. They share a manufacturing philosophy but serve different customers, sit at different points of the technology curve, and carry very different strategic weight.

Microelectronics (the growth bet)

This is the unit that makes IC substrates and the most advanced substrate-like PCBs. Its products go underneath high-performance processors: server and AI accelerator chips, high-performance computing (HPC) silicon, and advanced packaging architectures. The core capability is the ability to build ABF (Ajinomoto Build-up Film) substrates, the flip-chip ball-grid-array (FC-BGA) packages used for the largest, hottest chips. This is a genuinely scarce skill. Globally only five companies do it at the leading edge, and the qualification barriers are extreme.

Microelectronics is manufactured chiefly in Chongqing, China (the established high-volume substrate plant), in Leoben, Austria (R&D, the new IC substrate competence centre opened June 2025, and high-end production), and in the new Kulim, Malaysia megafab. This is where AT&S has placed its largest capital bet, several billion euros across Kulim and Leoben, on the thesis that AI and HPC will drive a structural shortage of advanced substrates.

It is clearly the unit management is steering the company toward. In the first nine months of FY2025/26 it grew from roughly €463 million to €621 million in revenue, up around 34% year over year, with a 65% quarter-on-quarter jump in Q3 as the Kulim and Leoben lines ramped and pricing and mix improved (Q3 FY2025/26 call). It is also the unit that inflicted the most pain: the start-up costs of Kulim and Leoben, and a "key customer" with volatile ordering behaviour (widely understood to be the anchor customer for the Malaysian substrate plant), were the direct cause of the FY2024/25 profit warning.

Electronic Solutions (the cash base)

This unit makes high-density interconnect (HDI) PCBs, modules, and system integration for end devices rather than for bare processors. Its biggest end market is mobile devices (AT&S has long been an Apple supplier and was named among Apple's top-200 suppliers), followed by automotive, industrial, medical, and aerospace/defence electronics. The core capability here is miniaturisation and integration: packing a complete functional module (sensing, connectivity, power) into the tiny, reliability-critical footprint a hearing aid, a medical implant, or a smartphone camera demands.

Electronic Solutions is made in Shanghai, China (mobile/HDI volume), Nanjangud, India, Fehring, Austria, and the Hinterberg/Leoben sites in Austria (module assembly, medical, and system integration). It exists as a separate unit because its economics and customers differ from Microelectronics: shorter product cycles, more seasonality (it swings with the mobile device calendar), lower capital intensity per euro of revenue, and a more diversified base of end markets. In the first nine months of FY2025/26 it was roughly flat at around €670 million, held back by mobile seasonality.

Within the group, Electronic Solutions is the steadier, more diversified base that funds the company while Microelectronics scales. Management talks about it far less on calls than the substrate ramp, which tells you where the strategic energy sits.

Section 3: Products and Business Detail

The product catalogue spans a ladder of complexity. At the top sit IC substrates / FC-BGA (ABF) substrates for high-performance computing, AI accelerators, servers, cloud, networking, and 5G base stations. These are the crown-jewel products: 20-plus layers, micron-scale features, the highest value per unit and the highest difficulty. Below them are substrate-like PCBs and high-end HDI (high-density interconnect) boards, used where space and signal integrity are tight but the extreme density of a CPU substrate is not required. Then come modules and embedded component packaging (ECP), where AT&S buries components inside the board itself to shrink the footprint, used heavily in medical, automotive, and miniaturised mobile applications. Finally there are rigid, flexible, and rigid-flex PCBs plus full module/system integration for medical devices, hearing aids, automotive control units, and industrial equipment.

What makes these hard to make is process knowledge plus qualification. An advanced substrate requires control of layer-to-layer alignment at the micron level, copper plating uniformity, laser drilling of tens of thousands of microvias, and warpage control on a thin organic board that wants to bow when heated. The output is judged on yield, and yield is the accumulated result of years of process tuning that does not transfer to a new entrant by buying the same equipment. On top of that, every advanced product must be qualified by the customer through months of reliability testing before it ships in volume, and medical and automotive products carry their own certification regimes (ISO 13485 for medical devices, IATF 16949 for automotive).

Geographically, manufacturing is concentrated in Austria (Leoben, Fehring, Hinterberg) for R&D and the highest-end production, China (Shanghai for HDI/mobile, Chongqing for substrates) for established high-volume work, India (Nanjangud), and Malaysia (Kulim) for the new substrate megafab. The Kulim plant is the single most important physical asset to understand: it is a multi-billion-euro greenfield IC substrate site in Kulim Hi-Tech Park, built to serve high-performance computing, AI, edge computing, IoT, servers, cloud, networking, and 5G. Its ramp, originally timed to an anchor customer's needs, slipped one to two quarters in 2024 and was the proximate cause of the profit warning, but in FY2025/26 it became a revenue contributor as AI demand filled capacity.

Notable milestones that reshaped the business: the 1994 buyout out of state ownership; expansion into China (Shanghai, then Chongqing); the building of Kulim and the Leoben substrate expansion (the largest capex cycle in the company's history); the opening of Europe's first IC substrate competence centre in Leoben in June 2025; and the September 2024 sale of the Ansan, South Korea plant (a medical-focused facility) to Italy's SOMACIS for roughly €405 million, a portfolio-sharpening move that also raised cash without diluting shareholders.

Section 4: Customers

The customer base splits along the two units. In Microelectronics, customers are chip designers and their packaging partners. AMD has been named publicly as an AI substrate customer, and management has referenced "three additional US technology customers focused on artificial intelligence." The anchor customer for the Kulim substrate plant has not been named by the company, but is widely understood in the market to be a large US processor maker whose own roadmap delays in 2024 fed directly into AT&S's "volatile ordering behaviour of a key customer" language. In Electronic Solutions, the marquee customer is Apple (mobile HDI), alongside automotive OEMs and tier-1s, industrial electronics makers, and medical device companies.

The buying decision differs by segment. For an advanced substrate, the decision is made by the chip designer's packaging and supply-chain engineering teams, on criteria of yield, reliability, capacity commitment, and the ability to co-develop next-generation designs. The sales cycle is long, often well over a year from design engagement to volume, because the part must be qualified. For mobile and consumer HDI, the cycle tracks the customer's annual product cadence and is more price-competitive.

Why customers choose AT&S: it is one of a handful of companies that can make leading-edge substrates at all, it is the only one with a significant European manufacturing base (which matters for customers wanting supply outside the Taiwan/Japan/Korea cluster), and it co-develops with customers rather than just building to print. Switching costs are high once a product is qualified: moving a substrate to another supplier means re-qualifying, re-testing reliability, and risking yield on a part that sits under a very expensive chip. That lock-in cuts both ways, as the next point shows.

Concentration is the central customer risk. In Microelectronics, a small number of large customers drive the substrate plants, and the FY2024/25 episode demonstrated the danger vividly: when one key customer pulled back orders, AT&S could not simply redirect a half-built megafab to other buyers overnight. The flip side is that the AI demand surge in FY2025/26, led by AMD and the unnamed US AI customers, refilled that capacity. Contract structure is a mix of long-term capacity and supply arrangements for the substrate business (necessary to underwrite multi-billion-euro plants) and more transactional volume in mobile/consumer, which makes the substrate side more predictable in principle but exposed to a single customer's roadmap in practice.

Section 5: Competitive Landscape

The industry that matters most for AT&S's future is advanced IC substrates, specifically ABF (Ajinomoto Build-up Film) FC-BGA substrates. This is a concentrated industry. Five suppliers (Unimicron, Ibiden, Nan Ya PCB, Shinko Electric Industries, and AT&S) controlled roughly 60% of global capacity in 2025, with no single player exceeding around an 18% share, which leaves it concentrated but competitive rather than a monopoly. The other leading-edge skill, the very highest layer-count AI/HPC substrates, is where the technology race is sharpest, and where players like Kinsus, Samsung Electro-Mechanics, and Zhen Ding are also pushing.

AT&S's structural edge is that it is the only one of the leaders with a substantial European manufacturing footprint and a new, modern megafab (Kulim) outside the East Asian cluster. For customers worried about geographic concentration in Taiwan and Japan, that is a real differentiator. Where AT&S is exposed: it is smaller than Ibiden and Unimicron, it entered leading-edge ABF later than the Japanese incumbents, and it carries the heaviest relative capex burden because it is building capacity that the incumbents already have.

Barriers to entry are genuinely high. Leading-edge substrate capability requires years of yield learning, billions in clean-room capex, customer qualification, and access to the specialised ABF material itself (Ajinomoto's build-up film is effectively a single-source input for the whole industry). A new entrant cannot buy its way in quickly. The clearest evidence of how the landscape is shifting is the take-private of Shinko Electric Industries by a Japan Investment Corporation-led consortium (completed in 2025), which removed one of the five leaders from public markets and signals consolidation and strategic re-shuffling at the top of the industry.

In Electronic Solutions, the competitive set is broader and more commoditised: a long tail of Asian HDI and PCB makers compete on price and scale, and AT&S wins on miniaturisation, reliability, and module integration rather than on cost.

Market caps are approximate orders of magnitude as of June 2026 and move continuously; treat them as peer-size references only.

Section 6: Industry

Demand for AT&S's most important products is driven by the AI and high-performance computing build-out. Every AI accelerator and server CPU needs an advanced substrate, and the newest architectures need bigger, denser, more demanding substrates. As packages grow beyond the silicon reticle limit and data rates and power delivery climb, more substrate layers and more substrate area are required per chip, so substrate content per system is rising faster than unit volumes alone. Secondary drivers include 5G infrastructure, networking, automotive electrification (more electronic content per car), and medical electronics.

The advanced IC substrate / ABF (FC-BGA) market is sized by multiple research houses in the multi-billion-dollar range and is forecast to grow at a solid double-digit pace through the end of the decade, with AI/HPC the dominant growth vector (Valuates, Mordor Intelligence, Yole Group). AT&S sits in the middle of the global semiconductor supply chain: downstream of the chip designers and the ABF material maker (Ajinomoto), and upstream of the assembly/packaging houses and the final OEMs.

There is a real geographic dynamic at play. Leading-edge substrate capacity has historically been concentrated in Japan, Taiwan, and Korea. Western chip customers and Western governments want supply diversified, which is the strategic logic behind AT&S's European base and its Malaysian plant outside the traditional cluster. This is less classic "import substitution" and more supply-chain de-risking, and it is a tailwind for the one leading supplier headquartered in Europe.

Regulation and certification shape the market through reliability and quality standards (automotive IATF 16949, medical ISO 13485) and increasingly through semiconductor industrial policy, as the EU and others subsidise local advanced electronics manufacturing. The industry is cyclical: it swings with the broader semiconductor cycle and with end-market demand (the mobile slump and weak server demand hurt AT&S in FY2023/24-FY2024/25), and it is capital-intensive, so periods of overbuild can compress pricing. The current tailwind is AI-driven substrate demand; the persistent headwinds are pricing pressure in lower-end products and foreign exchange (a weak US dollar against the euro hurts a company that sells in dollars and reports in euros).

Section 7: Growth Triggers

All triggers below are drawn from the six concalls. The most recent is the Q3 FY2025/26 call (February 2, 2026).

• Kulim (Malaysia) additional substrate lines ramping to support FY2026/27 revenue. Management said new Kulim capacity coming online underpins the €2.1-2.4 billion FY2026/27 revenue guidance. (Q3 FY2025/26 call, February 2, 2026; repeated theme from Q1 and Q2 FY2025/26.)

• Microelectronics revenue inflection driven by AI/HPC. The unit grew ~34% year-to-date and +65% quarter-on-quarter in Q3, on Kulim/Leoben ramps plus favourable product mix and pricing. (Q3 FY2025/26 call, February 2, 2026.)

• FY2026/27 revenue guidance of €2.1-2.4 billion reiterated. A step up from the ~€1.7 billion FY2025/26 base. (Reiterated across Q1, Q2, and Q3 FY2025/26 calls.)

• Cost-savings program delivering, expected to be sustainable. The €130 million target was hit by Q3, with management expecting to exceed €160 million by year-end and most of it to persist into future years, supporting margins as volume grows. (Q3 FY2025/26 call, February 2, 2026.)

• New AI customer wins beyond AMD. Management referenced AMD plus "three additional US technology customers focused on AI" filling substrate capacity. (FY2024/25 results, May 15, 2025; reinforced through FY2025/26 calls.)

• Leoben IC substrate competence centre (opened June 2025) feeding high-end development and European production. (FY2024/25 results, May 15, 2025.)

• Capex moderating as the build phase completes, from roughly €250 million toward €200 million, with the heavy investment cycle past its peak and free cash flow turning positive (operating free cash flow of €223 million year-to-date by Q3). (Q3 FY2025/26 call, February 2, 2026.)

"We are very confident that we will achieve, and it will be sure EUR 1.7+." - Peter Griehsnig, CTO, on the FY2025/26 revenue target, Q3 FY2025/26 call (February 2, 2026)

Section 8: Key Risks

Single-customer concentration in the substrate business. This is the defining risk and it has already materialised once. AT&S built a multi-billion-euro megafab in Kulim substantially around one anchor customer. When that customer's roadmap slipped and its ordering turned volatile in 2024, AT&S was left with a half-ramped plant generating start-up costs and no offsetting revenue, which forced a profit warning and a dividend cut. Management's own language is unusually blunt:

The company "assumes that the price pressure and volatile ordering behaviour of a key customer will continue," and its two new factories would start high-volume production "one to two quarters later than originally planned." - AT&S outlook adjustment, FY2024/25

This is a high-probability, high-impact risk that has partly resolved (AI demand from AMD and others is now filling capacity), but the structural exposure to a few large customers in a capital-intensive business remains.

Capex overhang and balance-sheet stress. AT&S spent enormously on Kulim and Leoben at a moment when end-market demand was weak. Net debt sat around €1.3 billion with net debt/EBITDA near 2.0 at Q3 FY2025/26. If the AI substrate ramp had failed, the company would have been carrying a giant fixed-cost base against soft revenue. The risk is now lower than it was in 2024, but it is the reason a dividend has not returned.

Foreign exchange. AT&S sells a large share of substrate output in US dollars and reports in euros. A weak dollar, which persisted into FY2025/26, directly compresses reported revenue and margin even when volumes grow. Management repeatedly flagged FX as a headwind across the FY2025/26 calls. This is a high-probability, moderate drag rather than a catastrophic risk.

Pricing pressure and cyclicality in lower-end products. The Electronic Solutions base and the lower end of the substrate market face persistent price competition from a long tail of Asian PCB makers, and the whole industry swings with the semiconductor cycle. A downturn in mobile or a general semis correction would hit the cash-generating base.

Governance and ownership instability. The 2024 power struggle (the controlling shareholder forcing out a long-tenured CEO over a failed capital-increase plan), followed by Androsch's death in December 2024, left the company with a substantially rebuilt management board and an unsettled core-shareholder situation. A management team less than two years old executing the largest capacity ramp in company history is an execution risk in its own right (see Section 9).

ABF material single-sourcing. The entire industry depends on Ajinomoto's build-up film. A supply disruption in that one input would hit AT&S and every competitor at once. Low probability, but worth naming because it is a genuine single point of failure for the product class.

Section 9: Walk the Talk

The six calls used for this assessment: Q2/H1 FY2024/25 (October 31, 2024), Q3 FY2024/25 (February 4, 2025), FY2024/25 annual (May 15, 2025), Q1 FY2025/26 (August 2025), Q2/H1 FY2025/26 (November 4, 2025), and Q3 FY2025/26 (February 2, 2026). The most recent is ~135 days before today; the company has not reported since, with FY2025/26 annual results due June 29, 2026.

The starting point matters because this is a management team in transition. The October 2024 H1 call took place amid open conflict between the long-serving CEO Andreas Gerstenmayer (CEO since 2010) and controlling shareholder Hannes Androsch over a proposed €450-500 million capital increase that would have brought the state holding ÖBAG in at around 25%. Androsch opposed it and prevailed; Gerstenmayer announced his departure effective October 1, 2024. Androsch then died on December 11, 2024. A new CEO, Michael Mertin, took over on May 1, 2025, with Peter Griehsnig as CTO and, from February 1, 2026, Gerrit Steen as CFO. So the team making promises in the second half of the period is largely not the team that made the original commitments.

The clearest broken promise is the medium-term revenue target, which has been cut repeatedly. The old guidance pointed to roughly €3.5 billion of revenue by FY2026/27, then was revised down to about €3.1 billion in 2024, and is now €2.1-2.4 billion. That is a roughly one-third reduction in the medium-term ambition over about two years. Anyone who underwrote the original substrate story has watched the destination move twice.

The second broken promise is nearer-term. In May 2024, FY2024/25 revenue was guided at €1.7-1.8 billion. The actual outcome reported on May 15, 2025 was about €1,590 million, a clear miss, with the headline EBITDA margin of 38.1% flattered by the one-off gain on the Ansan plant sale (the underlying adjusted margin was 17.7%). The dividend, €0.40 the prior year, was cut to zero and has stayed there.

Against those misses, the FY2025/26 promises have so far been kept or beaten. The €1.7 billion revenue target for FY2025/26 was reaffirmed each quarter and looked secure by Q3 ("it will be sure EUR 1.7+"). The €130 million cost-savings target was hit on schedule by Q3, with management raising the bar to €160 million-plus. Free cash flow turned sharply positive (operating free cash flow of €125 million at H1 and €223 million year-to-date by Q3, against deeply negative figures a year earlier). And the Microelectronics ramp that was a source of pain in 2024 became a source of growth in 2025/26.

The honest read: this is management that over-promised badly on the multi-year substrate vision under the old regime, missed its own near-term FY2024/25 number, and cut the dividend. But the rebuilt team has, in its first few quarters, delivered on the smaller, more concrete commitments it set, hitting cost-savings targets, turning cash flow positive, and reaffirming a revenue number it then tracked toward. The credibility verdict is therefore split: treat the grand multi-year targets with scepticism given the track record of cuts, but the recent quarter-to-quarter delivery has been accurate. The June 29, 2026 FY2025/26 results will be the first full-year scorecard for the new team.

Section 10: Shareholder Friendliness Index

Dividends. AT&S paid €0.90 per share for FY2021/22 and €0.40 for FY2022/23, then cut the dividend to zero for FY2023/24 and again for FY2024/25 (the AGM of July 3, 2025 resolved no distribution). So across the last three fiscal years the trajectory is €0.40, then nil, then nil. The cut was not a routine pause: it accompanied the FY2023/24 forecast adjustment, the decision to sell the Ansan plant, and the heavy Kulim/Leoben investment phase. Capital that would have gone to dividends went into the substrate megafab and into reducing the debt taken on to build it.

Buybacks and dilution. There has been no meaningful share buyback program over the last three years. The relevant capital-management story runs the other way: in 2024 management proposed a large €450-500 million capital increase (which would have diluted existing holders by bringing in the state holding ÖBAG at around 25%), and it was the controlling shareholder's veto of that dilution that triggered the boardroom upheaval. The increase did not happen, so shares outstanding have stayed broadly stable at around 38.85 million rather than expanding. AT&S raised cash instead through asset sales (the ~€405 million Ansan disposal) and free cash flow. Note: no MoatMap database block was injected for this venue, so the buyback picture above rests entirely on company filings and news for the full three years; there were no buybacks in any window I could find.

Verdict: Hoards capital. AT&S is in a capital-absorption phase, no dividend, no buyback, and a recently-contemplated dilutive raise, because every available euro has been directed at building leading-edge substrate capacity rather than returning cash.

Section 11: Insider Activities

AT&S is listed in Vienna and reports managers' transactions under EU Market Abuse Regulation Article 19, published as EQS directors'-dealings (DD) notices and on the company's own IR news feed. The following covers roughly the last 12 months. AT&S does not maintain a US-style consolidated insider tape, so transactions are pieced together from individual MAR Art. 19 DD notices.

Buys, the signal. The notable feature is a cluster of insider buying in December 2025, at depressed share prices in the low-€30s, by two different insiders: a sitting executive board member (Schröder, who ran the all-important Microelectronics unit) and a supervisory board member (Tumpel-Gugerell, a former ECB executive board member). Two insiders buying in the open market in the same window, at prices well below where the stock had traded in prior years, is a constructive signal. The absolute sums are modest (tens of thousands of euros each), so this is a confidence gesture rather than a fortune-betting conviction buy, but cluster buying by both a management-board and a supervisory-board member is the kind of pattern worth flagging.

Sells, the why. Schröder had earlier sold roughly €85,000 of stock in 2025, with no disclosed reason; for an executive selling a portion and later buying back a smaller amount, the most likely explanations are personal diversification or tax, and there is no indication it reflected a business view. Note that Schröder subsequently stepped down from the executive board (effective with the February 2026 board streamlining) while remaining with the company to lead Microelectronics as EVP, so some of his trading should be read against that role change rather than as a market call.

Net assessment. Over the last 12 months the insider activity is modest in size and net constructive in direction: the most recent and most decisive signal is December 2025 open-market buying by both a management-board and a supervisory-board member at multi-year-low prices, against only one earlier, smaller, unexplained sale. This reads as a mild bullish signal: insiders putting personal money in at the bottom of a difficult cycle, though the amounts are too small to be a high-conviction tell on their own.

Section 12: Scenarios

Bull case. The AI substrate boom proves durable and AT&S's timing looks inspired in hindsight. Kulim's additional lines ramp on schedule through FY2026/27, the megafab fills with demand from AMD and the other US AI customers, and the European IC substrate base in Leoben becomes a strategic asset that Western chip customers pay a premium to secure as they de-risk away from East Asia. The cost-savings program sticks, so the incremental substrate revenue drops through at high margins. Free cash flow, already positive, compounds and the balance sheet de-levers. With the build phase behind it and the boardroom settled under the new Mertin-led team, AT&S eventually restores a dividend, and the company that nearly broke itself building capacity into a downturn ends up as one of only five places in the world to get a leading-edge substrate, with the only major European footprint, just as the world decides it wants exactly that.

Base case. Management delivers roughly what it has guided. FY2025/26 lands around the €1.7 billion revenue mark with a margin in the low-to-mid 20s, and FY2026/27 climbs toward the €2.1-2.4 billion range as Kulim ramps, though probably nearer the low end if FX and pricing stay soft. Microelectronics carries the growth while Electronic Solutions stays roughly flat and cyclical. Cost savings hold, cash flow stays positive, debt grinds down, and the dividend stays suspended for now while the company finishes paying for its capacity. The medium-term targets are met but not beaten, the new management team builds a modest track record of doing what it says, and AT&S is a steadier, more cash-generative version of itself, still hostage to a handful of big customers and the dollar.

Bear case. The AI substrate demand that refilled Kulim turns out to be lumpy or front-loaded, and the same single-customer dynamic that caused the 2024 profit warning recurs: a key customer's roadmap slips again, orders turn volatile, and the megafab runs below the utilisation its fixed costs require. A broader semiconductor or mobile downturn hits Electronic Solutions at the same time. A weak dollar grinds reported margins lower even when volumes hold. The medium-term revenue target gets cut a third time, repeating the credibility damage of the last two cuts. Net debt, still around €1.3 billion, becomes heavy against falling EBITDA, the dividend stays at zero indefinitely, and the unresolved core-shareholder situation following Androsch's death produces a destabilising stake sale or another boardroom fight while the youngest management team in the company's recent history is trying to execute the largest ramp in its history.

Sources: AT&S IR - financial reports; AT&S Q3 FY2025/26 earnings transcript (Investing.com); AT&S Q1-3 2025/26 slides (Investing.com); AT&S H1 2025/26 results (publicnow PDF); AT&S Q1 FY25/26 (finanzwire); AT&S 2024/25 on Growth Course Again (I-Connect007); AT&S adjusts outlook 2024/25 (Evertiq); AT&S forecast/dividend/Ansan adjustment (AT&S IR); Wikipedia: AT&S; AT&S mourns Hannes Androsch (AT&S); Georg Riedl succeeds Androsch (MarketScreener); Industriemagazin: Androsch forces out Gerstenmayer; AT&S appoints Gerrit Steen as CFO (AT&S); EQS-DD: Gertrude Tumpel-Gugerell, Kauf (AT&S); EQS-DD: Ingolf Schröder, Kauf (AT&S); Börse Express: Schröder verkauft Aktien; AT&S Kulim (MIDA); Intel delays Penang/Malaysia facility (The Edge Malaysia); ABF substrate market (Valuates); Advanced IC substrates market (Mordor Intelligence); Yole Group: advanced IC substrate; AT&S dividend history (Parqet).

A note on what I could not fully verify: the exact share counts in the December 2025 Tumpel-Gugerell purchase are reported slightly inconsistently across German-language sources (some figures appear to mix her AT&S and Vienna Insurance Group purchases), so I have given approximate ranges. The FY2025/26 full-year figures will only be confirmed at the June 29, 2026 results.