AT&S (VIE:ATS): When a Cyclical Cash Generator Bets on the Future of AI Hardware
A detailed internal analysis of AT&S’s cash-generating PCB business and high-stakes substrate ramp, with bottom-up valuation, downside scenario modeling, and a focus on margin of safety.
This document is not financial advice. It is a personal research thesis written for internal use, intended to clarify my own thinking, uncover blind spots, and document my valuation framework for AT&S (VIE:ATS). While it has been structured to be readable by other value-oriented investors, it does not constitute a recommendation to buy, sell, or hold any security.
Readers should conduct their own due diligence and consult a financial advisor before making any investment decisions. Markets are risky, and this analysis—while thorough—is inherently based on assumptions, estimates, and forward-looking scenarios that may not materialize.
Note: As of publishing on July 20th, 2025, the stock is already up 11.05% to €21.80.
Thesis Summary
AT&S (Austria Technologie & Systemtechnik AG) is an Austrian electronics manufacturer operating two distinct businesses:
1. A legacy Printed Circuit Board (PCB) segment, which produces the layered platforms that connect and route electrical signals between components inside devices like smartphones, medical equipment, and cars. This segment is profitable, mature, and cash-generative.
2. A newer IC Substrate segment, which produces the ultra-thin, high-precision layers that sit between a semiconductor chip (like a CPU or AI accelerator) and the rest of the circuit board. These substrates are essential for enabling high-speed, high-density chips—especially in AI data centers, advanced PCs, and high-end consumer electronics.
Think of PCBs as the roads of an electronic city, and IC substrates as the on-ramps that connect the chip’s delicate internal structure to the rest of the system. The latter is where the real bottleneck lies in today’s high-performance computing landscape—and AT&S is one of the very few non-Asian companies with capacity in this space.
What This Thesis Does
This document was written primarily for myself—to pressure-test my thinking and expose any blind spots—then cleaned and structured so it could also be readable by other disciplined value investors. It assumes some comfort with accounting and business analysis, but not necessarily prior knowledge of electronics manufacturing or semiconductors.
Where relevant, technical terms are used directly (e.g. ABF substrates, BT materials, FO-WLP) and should be searched or queried via ChatGPT if unclear—I haven’t explained everything in-line for the sake of brevity.
The thesis includes:
A clear breakdown of how the two businesses generate value,
A deep dive into segment economics, including revenue models, margins, capex intensity, and cost structure,
Bottom-up valuation using conservative assumptions (Owner Earnings, SOTP, and DCF logic),
Monte Carlo simulations to explore probabilistic value outcomes,
A rigorous section on risks—including leverage, customer concentration, dilution, execution failure, and technology disruption.
Conclusion & Valuation Takeaway
At a market cap of ~€760 million (share price: €19.63 in July 2025), the equity essentially reflects:
The mature PCB business,
Minus net debt,
And assigns about -€290 million to the Substrate segment, despite over €2 billion of capacity and growing signs of successful ramp-up (e.g. AMD supplier awards, Intel test engagements, rapid progress in Leoben and Kulim).
This thesis concludes:
Conservative Base Case: ~€28.50/share (roughly 45–50% upside),
Bull Case (if margins & utilization match peers): €60–90/share (3–4.5x upside),
Bear Case: €15–18/share, assuming only the PCB segment retains value and no substrate upside materializes. (Unlikely)
The current price offers a margin of safety of ~40–50% to base case intrinsic value—and the possibility of multi-bagger upside if the substrate business reaches even modest efficiency and yield. Note: The part 3 and Monte Carlo simulation rise as high as ~€51.33/share to €59/share for the base case.
How to Read This Thesis
The full write-up is ~25,000 words.
Skim Reading (1.5–2 hours): Focus on the summaries, segment valuation logic, risk sections, and conclusion.
Focused Reading (3–5 hours): For full due diligence—dive into peer benchmarking, detailed segment models, scenario analyses, and the raw numbers.
This isn’t a polished sell-side report. It’s a practical deep dive for personal conviction-building—cleaned up so that it can serve as a reference for a future re-check or sharing with other investors. Use it as a thinking aid, not gospel.
1. Business & Industry Understanding
1.1 Segment Overview: What AT&S Does (Plain-English)
Advanced PCBs (Electronics Solutions Segment): AT&S’s original core business is manufacturing printed circuit boards – the thin, layered boards etched with conductive pathways that mechanically support and electrically connect electronic components. AT&S specializes in high-end, high-density PCBs (like HDI – high-density interconnect boards, rigid-flex boards, and other complex multilayers) used in mobile devices (smartphone mainboards, wearables), automotive electronics (ADAS systems, engine control units), industrial controls, medical devices, and more. These are not commodity PCBs, but rather boards with miniaturized features and multiple layers to accommodate advanced chips and dense circuitry. For example, AT&S is a leading supplier of Apple’s SLP (substrate-like PCB) mainboards for iPhones and Apple Watches, which require ultra-fine circuit patterns. In plain terms, AT&S makes the “nervous system” of electronic gadgets – the green boards studded with copper traces that you’d find inside a smartphone or car module – and it does so with cutting-edge precision to fit more capability into smaller, faster devices. The Electronics Solutions segment also offers related services like design co-development and module assembly (e.g. integrating components onto boards) to provide turnkey “interconnect solutions” for customers. This segment is the mature, cash-generative side of AT&S, leveraging decades of PCB know-how. It accounted for roughly 60–70% of AT&S’s revenue until recently and remains critical to funding the company’s new ventures.
IC Substrates (Microelectronics Segment): The Microelectronics segment produces integrated circuit (IC) substrates, which can be thought of as an intermediate mini-circuit board that sits between a microchip and the main PCB in advanced electronic systems. In modern high-performance processors (like CPUs, GPUs, AI accelerators), the silicon chip cannot be soldered directly onto the rougher, larger PCB; instead, it’s mounted on a substrate – a smaller board made of specialty resin (often using Ajinomoto Build-up Film, or ABF, as insulation) with incredibly fine wiring that fan out the chip’s microscopic I/O contacts to the PCB’s larger contacts. AT&S’s substrates are used in data center processors, high-end AI chips, and other computing applications where high I/O density and signal integrity are paramount. Essentially, substrates are the “silicon chip package” platform that ensures the chip can communicate with the rest of the system. This business is relatively new for AT&S (entered mid-2010s) but is its biggest growth focus – by 2025 AT&S aims to be among the world’s top 3 substrate suppliers. The company has built state-of-the-art substrate fabrication cleanrooms in Chongqing (China) and recently in Kulim, Malaysia, as well as an R&D and pilot line in Leoben (Austria). In simple terms, AT&S is leveraging its PCB expertise to tackle the even more complex task of making “PCBs for chips”, which involve hundreds of microscopic layering steps and near-pristine manufacturing to achieve yields. This segment currently contributes ~30–40% of revenue, but is expected to drive virtually all the company’s growth going forward.
1.2 Industry & End-Market Exposure
End-Use Industries by Revenue: AT&S’s revenue mix is fairly diversified across major electronics end-markets, though consumer devices dominate. As of FY2023/24, approximately 59.5% of revenue came from “mobile devices and media”, which includes smartphones, tablets, laptops, wearables and other consumer electronics. The remaining 40.5% is from “industrial, medical and automotive” applications. In practice, smartphones are the single biggest driver (with Apple as a likely top customer), alongside other personal electronics – this makes AT&S partly beholden to the annual smartphone launch cycle and consumer upgrade rates. The automotive share (included in that 40.5%) has been growing as cars incorporate more PCB content (for EVs, ADAS, infotainment), providing a secular tailwind. Industrial and medical uses (e.g. factory automation, imaging equipment) are smaller contributors but offer stable demand. Meanwhile, the IC substrates business specifically is tied to the semiconductor cycle – notably PC/server CPUs and AI accelerators for data centers, as well as some high-end consumer processors. Recent disclosures suggest rising substrate demand from data center, AI, VR/AR, and high-performance computing customers, evidenced by AT&S’s contract to supply AMD’s epic data-center chipse. Geographically, AT&S’s end demand is global: ~75%+ of revenue ultimately comes from Americas (largely via tech customers who sell worldwide). In summary, AT&S is exposed to fast-moving tech hardware cycles (smartphones, PCs, cloud computing) for the majority of its business, with a significant minority linked to more traditional cyclical sectors like automotive and industrial.
Cyclicality of Demand (By Segment): Demand in both segments is highly cyclical, though for different reasons:
PCB (Electronics Solutions) Cyclicality: This segment’s demand tends to follow general electronics cycles and consumer spending. For instance, smartphone demand can swing with handset upgrade cycles – AT&S saw a smartphone slowdown in 2019 which caused its mobile revenues to dip ~5% in the first half of FY2019/20. In 2020, COVID initially disrupted electronics supply chains but stimulus and remote-work trends then boosted device demand; AT&S’s revenues actually grew from €1.0B in FY2020 to €1.2B in FY2021 despite the pandemic. Nonetheless, consumer electronics are inherently cyclical, and inventory corrections (like the one in late 2022–2023 when smartphone and PC sales cooled off sharply) hit PCB orders hard – e.g. AT&S’s PCB segment revenue fell in FY2023/24 due to a “significantly softened” mobile and industrial market in H2. Automotive PCB demand is likewise cyclical, tied to car production which fluctuates with economic conditions (though increasing electronic content per vehicle provides a secular growth offset). Importantly, PCBs are somewhat commoditized over the long run – during downturns, excess PCB capacity and pricing pressure can emerge, squeezing margins. AT&S mitigates this by focusing on the high-end niche where fewer competitors can match its technology, but even here, pricing fell in 2023 and is under “persistent high price pressure” now. Thus, the PCB segment is moderately cyclical, with troughs during global recessions or electronics lulls (2008 saw a demand dip; 2019 smartphone pause; 2023 broad electronics slump) and peaks when new device cycles or tech upgrades (4G/5G, etc.) drive booms.
IC Substrate (Microelectronics) Cyclicality: The substrate business is closely linked to the semiconductor capital cycle and HPC demand, which can be even more volatile. In recent years, ABF substrates experienced a boom-bust cycle: a severe supply shortage in 2021–2022 (driven by surging demand for PC and server processors, GPUs for gaming and crypto, and later AI chips) led to record revenues for substrate makers. AT&S’s substrate sales climbed strongly through FY2022/23 (contributing to record group revenue €1.8B). However, in FY2023/24, weakness in the server market caused substrate segment revenue to “decline slightly” and orders for PC-related substrates dipped (post-pandemic PC slump). The substrates market is subject to the silicon cycle: when semiconductor customers see demand drop, they burn off inventory and substrate orders can fall sharply. Additionally, substrate capacity expansions (by AT&S and rivals) are coming online just as the market is digesting a slowdown in classic servers – raising risk of short-term oversupply. AT&S explicitly notes that while AI chips and client PC chips recovered in recent quarters, “the classic server segment continues to be subdued”, and a general upturn in substrate demand likely awaits broader economic recovery. Therefore, this segment could face periods of low utilization if end-demand lags, and its fortunes are tied to big cycles in tech (e.g. cloud capex cycles, new CPU/GPU launches, etc.). Overall, both segments are cyclical, but the drivers differ: consumer/product cycle swings for PCBs, and semiconductor/HPC investment cycles for substrates.
Competitive Dynamics (Porter’s Five Forces by Segment):
Electronics Solutions (High-End PCB) Competition:
Rivalry: Competition among PCB makers is intense globally, but technical capabilities act as a moat at the high end. AT&S is the world’s #2 producer of high-end PCBs by volume, competing with Asian manufacturers like Zhen Ding Technology (Foxconn’s PCB arm, major in iPhone boards), Unimicron (Taiwan, also does HDI boards), Tripod Technology (Taiwan), Nippon Mektron, and others. U.S.-based TTM Technologies is another competitor especially in aerospace/defense boards. Rivalry is moderated by the fact that the most advanced PCB orders (e.g. Apple’s substrate-like boards) are split among only a few capable suppliers. However, in less advanced PCBs, Chinese firms have flooded the market, keeping prices low. AT&S tries to stay on technology’s cutting edge (e.g. ultra-fine lines, embedded component packaging) to avoid commodity fights. Still, price competition is a factor – AT&S reported significant pricing pressure in PCBs in 2023/24, implying competitors willing to cut margins to win business in a weak market.
Threat of New Entrants: The PCB industry has relatively low entry barriers at the low-end (many small fabricators exist), but very high barriers at the high-end. New entrants capable of matching AT&S’s precision and scale would require massive capital investment, engineering talent, and customer qualifications. China’s government has subsidized some PCB players, but so far the likes of Shennan Circuits or Wus Printed Circuit (leading Chinese PCB firms) focus on slightly lower-end or mid-tier HDI boards. Entrants also face the hurdle of customer trust – top OEMs (like Apple) are risk-averse in sourcing PCBs. Thus, threat of effective new entrants in AT&S’s niche is low.
Supplier Power: Key inputs for PCBs include laminate materials, copper foil, specialty chemicals, etc. Suppliers like DuPont (for laminates) or copper-clad laminate vendors have some power when shortages occur (e.g. past spikes in copper prices or ABF film supply constraints). However, these are fairly commoditized upstream markets with multiple suppliers. AT&S’s scale likely earns it some preferential supply, but shortages (like the ABF film shortage that also hit substrate production) can be a bottleneck – ABF (made by Ajinomoto) is essential for certain high-density boards, so limited suppliers there do have power. On balance, supplier power is moderate.
Customer Power: AT&S’s customers, especially in PCBs, include major OEMs and electronic manufacturing service (EMS) giants. They tend to have substantial bargaining power. For example, Apple (a known end-customer via its subcontractors) and big auto Tier-1s or EMS companies (Foxconn, Bosch, etc.) are huge and demand cost reductions and flawless quality. AT&S’s high-end tech gives it some differentiation, but customers often dual-source to keep pressure on suppliers. The fact that AT&S is one of only a couple of suppliers for certain iPhone boards gives it some leverage (Apple can’t easily drop it without risking supply), yet Apple is notorious for squeezing suppliers’ margins. In automotive, long-term relationships and qualification processes give the supplier some stickiness, but pricing is tight. So customer power is high in this industry.
Substitutes: There is no direct substitute for PCBs in electronics – you need an interconnect substrate of some kind. The main “substitute” is technological advancement: for instance, module integration or semiconductor packaging that bypasses some PCB content. But fundamentally, as long as electronics need wiring, PCBs (or their evolutions) are required. Thus, substitute threat is low, albeit one can argue that higher-level integration (like System-in-Package designs) can reduce the number of separate PCBs needed in a device (a trend to watch).
Overall, the PCB segment’s competitive environment is moderately challenging: high rivalry and customer power exert margin pressure, but high-end tech acts as a protective moat for AT&S.
Microelectronics (IC Substrates) Competition:
Rivalry: IC substrates is a much more oligopolistic market. As of 2022, the top five substrate manufacturers controlled the majority of the global market, with AT&S holding roughly ~9% share. Key rivals include Unimicron (#1, Taiwan), Semiconductor Manufacturing Electronics (SEMCO) of Samsung (Korea), Ibiden and Shinko (Japan), and Nan Ya PCB (Taiwan). These incumbents have decades of substrate experience and deep ties to chipmakers. Rivalry is tempered by surging demand (in recent years, everyone’s capacity was fully booked) but is now increasing as new capacity comes online. AT&S is a newcomer challenging to become top-3 by volume. There is rivalry in technology (e.g. race to build substrates with more layers and finer pitches required by next-gen chips) and in capturing marquee customers. However, given the overall market growth (forecasted to double from ~$15B in 2022 to ~$29B by 2028), rivalry has been more about ramping to meet demand than price wars. In a downturn, this could change: if substrate capacity outstrips demand, price competition could emerge, but customers also prioritize reliability and precision highly. AT&S’s move into substrates has certainly raised competitive eyebrows – incumbents will defend their turf, potentially by leveraging their quality track records. That said, with few capable players, the rivalry is limited to a small club where each has historically had enough business to grow.
Threat of New Entrants: Very low – the barriers to enter IC substrate manufacturing in 2025 are enormous. It requires multi-billion euro investment (AT&S is spending ~€1.7B on just Phase 1 in Kulim) and a multi-year learning curve on yields and qualifications. Only a handful of companies globally have succeeded. The only notable new entrants on the horizon are potential Chinese-backed firms (China has been trying to build domestic substrate capacity, but still lags in technology) and possibly Intel (which has in-house packaging, though it focuses on its own needs rather than external sales). There’s also research into glass substrates as a next-gen technology, but that’s a longer-term potential substitute (late this decade) and would likely be adopted by existing players. In short, new entrants in the traditional organic substrate space are highly unlikely in the near term.
Supplier Power: A critical material is the ABF film (made by Ajinomoto) – for a time, this was a chokepoint with limited suppliers, giving them power. AT&S and others depend on reliable shipments of these films and advanced fabrication equipment (from suppliers like Orbotech/KLA for laser drilling, AsmL for some lithography steps, etc.). The equipment suppliers for substrates (similar to semiconductor fab tools) are quite specialized and have backlogs; this can constrain how fast capacity ramps. Overall, suppliers have moderate power – e.g., if Ajinomoto doesn’t expand ABF film production, it limits substrate output for all players. However, large substrate makers often have strategic relationships to ensure supply (perhaps even prepaying for materials). On balance, input supplier power is a medium risk: not a daily concern, but critical in planning.
Customer Power: AT&S’s substrate customers are mainly semiconductor companies or their OSAT partners. These include giants like AMD (already secured as a customer), potentially Nvidia, Intel, Qualcomm, Apple (via its chip packaging subcontractors), etc. Such customers are highly demanding (“AMD can be a very challenging and demanding customer,” AMD’s VP noted) and typically multi-source substrates to ensure supply continuity. Their power is significant: for instance, AMD awarding business to AT&S (a new supplier) is a big win for AT&S, suggesting AMD had leverage to negotiate favorable terms and performance guarantees. If a substrate supplier falters (quality or delay), the chip company can shift volume to competitors (though not instantly, since qualifications are needed). Furthermore, pricing for substrates often involves annual or multi-year agreements, and customers push for cost reductions over time. Thus, customer power in substrates is high, mitigated only by the limited pool of qualified suppliers – if a customer needs more volume than incumbents can supply, it may accept a higher price or support a new supplier (as AMD did by engaging AT&S, diversifying away from solely Taiwanese/Japanese sources). In summary, large chipmakers have clout, but they also need enough substrate supply – a dynamic of mutual dependence.
Substitutes: The main threat here is technological substitution in chip packaging. There is continuous R&D in advanced packaging that could reduce reliance on conventional ABF substrates – for example, wafer-level packaging, 2.5D/3D integration, or future glass substrates. Currently, for high-performance chips, ABF substrates are the industry standard and there’s no near-term drop-in replacement that’s broadly viable. However, if in 5–10 years technologies like silicon interposers or direct chip-to-PCB bonding matured, they could cut into substrate demand. Given the time horizon and technical challenges, we consider substitutes a low near-term threat, but a longer-term strategic risk.
Overall, the IC substrate segment has favorable industry structure with few players, high entry barriers, and robust demand growth – which is why AT&S is betting heavily on it. But it is also a high-stakes, winner-takes-most game where execution matters immensely and customers hold stringent expectations.
1.3 Customer Structure and Relationships
Top Customers & Concentration: AT&S does not publicly name its customers in detail (typical due to NDAs), but industry reports and disclosures give strong clues. Apple Inc. is a very likely top customer for the PCB segment – AT&S has been identified as a key supplier of substrate-like PCB mainboards for iPhones and Apple Watches, sharing ~30% of Apple’s volume for those with one other supplier. This implies Apple-related business could be on the order of a few hundred million euros annually (Apple’s name even appears on AT&S’s sustainability reports as a supply chain partner). In substrates, AMD is confirmed as a major customer: AT&S’s new Kulim plant’s first product is substrates for AMD’s Epyc data-center processors, and AMD’s CEO Lisa Su publicly praised the Kulim facility as “incredible” and key to AMD’s supply chain.
AMD’s CEO Lisa Su on a AT&S Youtube video for the Plant Opening Ceremony in Kulim. - - AT&S Malaysia will produce cutting-edge IC Substrates, which are an integral part of high-performance data processors for computers, datacentres and AI-infrastructure. “AT&S has proven to be a very capable additional source of high-quality IC Substrates for high-performance AMD data centre processors, strengthening our global supply chain. We look forward to leveraging the leading-edge technologies manufactured in this new plant”, remarked Scott Aylor, Corporate Vice President for Global Operations Manufacturing Strategy at AMD.” -
It’s likely that Intel and Nvidia are target customers (if not already trialing AT&S’s output) given they consume huge substrate volumes, but incumbent suppliers currently serve them. Other significant customers in automotive could include Bosch, Continental, or auto ECU makers for PCBs, and possibly Medtronic or Siemens for medical PCBs. Historically, AT&S did a lot of business in mobile and communications (e.g. Huawei was previously a customer) and in industrial electronics (e.g. aerospace). The company has been trying to diversify its customer base in recent years, explicitly stating success in winning new accounts outside its traditional ones. Still, we infer that the top 2–3 customers likely make up a substantial portion of revenue – potentially >30%. For instance, losing Apple or AMD as a client would be a major blow.
In AT&S latest Youtube video, we can see executives from Marvell (develops and produces semiconductors and related technology - $62B valuation) and Ampere Computing Holdings LLC (“a U.S.-based semiconductor design company focused on high-performance, energy efficient, sustainable AI compute based on the ARM compute platform” which just got acquired by Softbank).
Switching Costs & Dependency: For high-end PCBs and substrates, switching suppliers is non-trivial due to qualification times and reliability requirements. Once a PCB is designed into a device, OEMs are reluctant to switch unless there’s a cost or supply issue. In automotive and medical, qualifications can take months or years, giving suppliers some protection (incumbency is valuable). Similarly, substrate suppliers go through extensive qualification by chipmakers; AMD took about 2–3 years working with AT&S Kulim (the plant was built in 2 years and then took another year for product qualifications and ramp). This creates a moderate lock-in – chip companies won’t lightly drop a qualified substrate supplier, especially if capacity is tight. That said, dependency risk cuts both ways: AT&S is highly dependent on a few customers’ programs (e.g. Apple’s iPhone cycle each year, AMD’s chip roadmap). If a customer’s product underperforms in the market (say, lower iPhone sales one year or AMD losing share to Intel/Nvidia), AT&S feels it. Also, a powerful customer like Apple or AMD can dual-source to keep AT&S from having too much leverage. So while technical integration creates some stickiness, the bargaining asymmetry means AT&S remains quite dependent on its large clients’ fortunes and procurement strategies.
Contract Terms & Visibility: Specific contract terms are not disclosed, but some general patterns can be assumed. For mobile/consumer PCB business, AT&S likely operates on purchase orders and short-term commitments, aligned with product launch cycles (Apple, for example, typically doesn’t sign very long-term fixed-price contracts for boards, instead it might allocate volumes among suppliers each year). This means visibility is limited to maybe a few quarters; indeed, AT&S has shifted to giving only quarterly guidance because of volatile order behavior of a key customer. For substrates, there might be more forward visibility due to capacity reservation agreements – e.g., AMD likely committed some baseline volume to enable AT&S’s investment. It’s possible AT&S has multi-year framework agreements for substrates (to justify the capex, customers often give letters of intent or long-term supply MOUs). However, even those likely have flexibility and are contingent on meeting yield/quality targets. The fact that AT&S cited “volatile order behavior of a key customer” impacting Q1 2025/26 suggests that even in substrates, orders can swing quarter to quarter (this likely refers to either Apple reducing orders or AMD adjusting forecasts). We also know AT&S sold its peripheral Korea plant to focus on “core business” with presumably tighter customer alignment. The sale of the Korea plant (which made older tech boards) might indicate an attempt to shed business with less visibility or lower margins.
In summary, AT&S’s revenue stream is concentrated and dependent on a few big tech players. Switching costs and engineering integration offer some protection once AT&S is an approved supplier (it took a huge effort to become AMD’s certified high-volume supplier). Nonetheless, customer power remains high – AT&S is essentially part of the supply chain of much larger companies. Contractually, the business likely lacks long-term fixed take-or-pay contracts; rather, it relies on forecasted orders that can be revised. This inherently limits forward visibility and is a reason we apply a strong margin of safety.
Dependency Risk Example: It’s worth noting a real example: when smartphone markets softened and “a volatile order behavior of a key customer” continued into early FY2025/26, AT&S had to refrain from giving full-year guidance. This likely alludes to Apple adjusting orders. That underlines how AT&S’s fate can swing with one customer’s inventory and sales dynamics. On the other hand, the AMD partnership gives some counterweight – AMD explicitly wanted a second source (beyond its traditional Taiwanese suppliers) and is invested in AT&S’s success (AMD even awarded AT&S “Certified HVM [High-Volume Manufacturing] Site” status after the Kulim ramp-up). Such partnership can reduce risk of sudden demand evaporation in substrates, at least for the committed programs.
Overall, while AT&S’s technology commands respect, the company must continuously navigate a tightrope of customer concentration and maintain world-class execution to keep these tech giants as clients. The substantial dependency on a few relationships is one of the reasons we insist on a large margin of safety – a stumble in any one key account could materially affect intrinsic value.
2. Segment-by-Segment Financial Deep Dive
We will now examine each segment’s financial performance and prospects in detail, applying a 10-year conservative lens favored by value investors. We focus on revenue trends, profitability (using EBITDA and Owner Earnings), capital expenditure needs (maintenance vs growth), and returns on invested capital, culminating in a cautious valuation for each segment.
2.1 Advanced PCBs (Electronics Solutions Segment)
Revenue Growth (10-Year Retrospective): A decade ago, AT&S’s PCB business was much smaller – for context, in FY2013/14 AT&S’s total sales were ~€590 million, almost entirely PCBs (since substrate contribution was negligible then). Fast-forward to FY2021/22, Electronics Solutions segment revenue had grown significantly as AT&S rode the smartphone boom and increased content per device. Group revenues hit €1.6 billion in FY21/22, and PCB likely constituted >€1.0–1.2 billion of that. The segment benefited from key wins like Apple’s iPhone board business around 2016-2020, and strong demand in automotive and industrial through 2018. However, growth has not been linear: cyclical dips occurred (e.g., roughly flat revenues around 2015–2016 amid smartphone maturing, a slight dip in 2019 as noted). Over 10 years, a reasonable estimate is that the PCB segment’s revenue roughly doubled, equating to a mid-single-digit CAGR. The peak was FY2021/22 when Electronics Solutions hit record sales alongside broader industry highs. In FY2022/23, PCB revenue likely exceeded €1.2 billion (since total was €1.79B and substrates were ~€0.5B) – a record level – before dropping ~13% in FY2023/24 as demand softened. Indeed, management reported the PCB segment’s revenue “fell short of the strong figures of the previous year” due to a weaker product mix and pricing pressure. We infer FY23/24 PCB sales around €1.0–1.1B (down from perhaps ~€1.3B in FY22/23). This retrenchment, while significant, still leaves the segment at a much higher base than 5–10 years ago. Going forward, we assume little to no growth for this segment in our conservative base case – perhaps a normalized growth rate of ~3% or less per annum, roughly in line with global PCB market GDP-like growth. In fact, management itself guides for only low growth in mature segments; the doubling of revenue by 2026/27 is to come from substrates, with PCB being stable or modestly growing. We will later apply ~0–5% growth in our valuation, to be safe.
Profitability & Margins: The PCB segment historically has had solid margins in upcycles and thin margins in downcycles. Over the last decade, EBITDA margins ranged roughly 15%–25% for AT&S’s PCB business. For example, in FY2013/14, AT&S overall had ~€127m EBITDA on €590m revenue (~21.5% margin), mostly from PCBs. In the recent peak FY2021/22, segment EBITDA was exceptionally strong at €310m. Assuming PCB revenue around €1.3B that year, the segment EBITDA margin was roughly 24% – buoyed by high utilization and favorable mix (tech customers ramping orders). However, in the downturn FY2023/24, PCB EBITDA plunged 32% to €210m on lower revenue ~€1.0–1.1B, implying margin compressed to ~19–20%. This was due to a “less favorable product mix” (likely fewer high-margin smartphone boards and more lower-end or idle capacity) and price erosion. In other words, margins swing widely with volume and pricing. During the Great Financial Crisis (2008–2009), AT&S’s margins also shrank drastically (the company made losses in some earlier downturn years), illustrating limited fixed-cost absorption when volumes fall. To gauge through-cycle profitability, we consider a normalized EBITDA margin around 20% for this segment – conservative given their technology edge, but prudent given competition.
To approximate Warren Buffett’s Owner Earnings (OE), we start with cash flow from operations (CFO) and subtract maintenance capital expenditures—the cash outlay needed merely to keep current capacity and competitive position intact. AT&S has historically guided to €40-50 million a year of “maintenance capex and minor technology upgrades” for the whole group. Scaling that guidance to FY 2023/24 revenue (group €1.55 bn, Electronics Solutions c. €0.93 bn) implies that a realistic steady-state maintenance charge for the PCB division is about €45 million, or ≈ 3 % of segment sales.
FY 2023/24 segment CFO (proxied by EBITDA less Δworking-capital cash outflow) was €235 m.*
Less maintenance CAPEX €45 m
Owner Earnings ≈ €190 m, equal to a low-20 % yield on segment revenue.
Context: Group depreciation was €276 m in FY 2023/24, so the €45 m assumption implies maintenance CAPEX at ~16 % of D&A, a ratio typical for PCB plants once the big growth projects are carved out.
*AT&S does not publish segment-level cash-flow statements; CFO is estimated from segment EBITDA minus the working-capital change disclosed in the notes. While not perfect, it is a closer economic measure than headline EBITDA.
Using FY2023/24 as a depressed case: PCB segment EBITDA €210m, depreciation perhaps ~€100m (a rough estimate for that segment), yields EBIT ~€110m. If maintenance capex ~€50m, and ignoring growth capex, the owner earnings might be on the order of €110m (EBIT) + €100m (depreciation) – €50m (maintenance) – €(taxes). After ~25% tax, OE ≈ €120–130m in that down year. In a better year like FY2021/22, PCB OE would have been much higher (EBITDA 310m, maintenance 50m, taxes ~20% of EBIT, giving maybe ~€200m+ OE). For a normalized mid-cycle, we’ll take PCB owner earnings ≈ €150 million as a conservative representative figure. This equates to, say, €1.1B revenue * 20% EBITDA – maintenance – tax.
Capital Expenditures – Maintenance vs Growth: The PCB segment is capital intensive, but much less so than substrates. AT&S has established PCB fabs in Shanghai (China), Nanjangud (India), and a smaller one in Austria, etc. Most of these have been built and expanded in years past. Growth capex in recent years overwhelmingly went into substrates, not PCB. The segment’s needs now are largely maintenance plus incremental upgrades (like adding a capability for finer line width, or replacing an older drill machine with a laser drill). Growth capex for PCB has essentially paused – in fact, AT&S opted to sell its Korea PCB plant (Ansan) in 2025 for €405m.
On September 23, 2024, AT&S and SO.MA.CI.S. S.p.A. entered into an agreement to sell AT&S Korea Co., Ltd. (i.e., the Ansan PCB plant) for an equity value of approximately €405 million. The transaction was finalized on January 31, 2025, when AT&S completed the sale of its Ansan, Korea, plant to SO.MA.CI.S.
That plant made less advanced boards and was sold to refocus resources. This divestiture suggests AT&S doesn’t plan new PCB factories soon; rather, it will optimize existing lines and only invest in efficiency or minor capacity tweaks. Thus, future capex for PCB should be low relative to history. In the past, when AT&S built new HDI lines in China around 2010s, growth capex was significant, but now that footprint suffices for current demand. We can reasonably treat nearly all future capex for AT&S as pertaining to substrates, with PCB getting its ~€50–60m maintenance slice. The ROIC in PCB business historically has been decent in good years (teens %) but poor in bad years. For instance, in FY2022/23 the segment likely generated ROIC in the ~15% range given high earnings, whereas in FY2023/24 ROIC collapsed to a few percent. Over 10 years, AT&S’s PCB ROIC would average out perhaps high single digits – not fantastic, reflecting the competitive nature, but improved by focusing on high-margin niches.
Competitor Comparison: Compared to peers, AT&S’s PCB segment holds up well on tech and margins but is smaller in sheer scale than some Asian rivals. Unimicron and Zhen Ding each have PCB revenues well above $1.5B and also do substrates. Tripod Technology (Taiwan) had revenue around $1B+ and typically low-teens EBITDA margins (Tripod’s forward P/E ~10, EV/EBITDA ~7–8 historically, indicating modest profitability). TTM Technologies (U.S.) has EBITDA margin ~15% in recent years and trades around ~12× EV/EBITDA in mid-2025. AT&S’s ~20% segment EBITDA margins in good times are slightly above many PCB peers (thanks to its specialization), but not unheard of – high-end PCB work can command ~20–25% margins at peak. In downturns, AT&S cut costs aggressively (e.g. they initiated an efficiency program and headcount reduction of 1,000 in FY24*) to defend margins. Chinese HDI players often accept lower margins for market share, but AT&S avoids competing on the lowest cost tier.
Regarding the efficiency program:
Programme launch and timelines
Initial rollout: AT&S first announced “comprehensive cost optimisation programmes” in February 2023, with a Corporate News release on March 16, 2023 detailing that these measures would focus on productivity, materials and purchasing.
Expected savings: These initial programmes were projected to deliver € 440 million of cost savings over the two years following FY 2022/23.
Intensification: On May 14, 2024, alongside its FY 2023/24 results, AT&S confirmed that it had “intensified our ongoing efficiency programs” and would implement a headcount reduction of up to 1,000 employees at existing locations to defend margins in FY 2024/25.
Implementation costs
In its October 28, 2024 ad-hoc announcement, AT&S disclosed that it expects to incur one-off costs of up to € 110 million in FY 2024/25 for implementing the cost-optimisation and efficiency programme.
Headcount impact and scale
As of March 31, 2024, AT&S employed 13,507 people (average FY 2023/24 headcount: 13,828). The 1,000-job reduction therefore equates to ≈7 % of the overall workforce.
Earlier (March 2023) disclosures put total headcount at “roughly 15,000”, which would make a 1,000-person cut about 6.7 % of staff.
AT&S does not publicly break out headcount by segment (e.g. PCBs vs. IC substrates), but since PCB manufacturing remains its core business, this reduction is proportionally significant to its printed-circuit-board operations.
Bottom line: AT&S’s efficiency drive kicked off in early 2023 (savings target € 440 million), was ramped up in mid-May 2024 (1,000 jobs cut), and carries one-off costs of up to € 110 million in FY 2024/25. The headcount reduction represents around 7 % of its total workforce—a substantial adjustment for its PCB-focused operations.
Downturn Resilience: Historically, AT&S’s PCB segment has not been immune to recessions – during the 2008–2009 crisis, AT&S’s revenue fell and it reported losses (though it was smaller then). In the smartphone plateau of 2019, AT&S had to digest a 5% revenue drop in H1 and lower utilization. However, management has shown discipline in such times: They focus on cost cuts (the current program aims €250m savings over FY24-26), and they diversify end-markets to offset. For instance, weakness in mobile in 2H 2023/24 was partly offset by strength in “armament industry and data centers” PCBs – AT&S noted defense and server investments picked up in early 2024, softening the blow. The segment’s resilience thus depends on having a mix: consumer vs industrial vs auto. It appears that while auto PCBs slowed in late 2023 (due to high inventory), other areas helped. The PCB segment still produced +€210m EBITDA in a very tough FY23/24, remaining profitable. This suggests a baseline resilience: even in bad years it can likely generate positive cash (helpful for covering corporate costs and interest). We stress-test a severe downside where smartphone demand drops 20% and auto/industrial stalls – PCB revenue could fall perhaps 15–20%, and margins compress to low teens or single digits. In such a case, segment EBITDA might only be, say, €100m (near break-even net profit after depreciation). That’s about as grim as 2009, which we consider an extreme. The base-case is far brighter, but we will account for this scenario in risk analysis.
Growth Ceiling Estimate & Normalized Growth: The PCB business for AT&S is relatively mature. Management’s commentary implies that beyond cyclical recovery, growth will be low to mid-single digits. Secular trends like 5G phones, EVs, and IoT could expand PCB content, but conversely, some device consolidation (e.g. fewer discrete PCBs if modules combine functions) could cap growth. We conservatively estimate a ~3% CAGR long-term for this segment’s revenue – essentially tracking global electronics growth. The “ceiling” is not high – AT&S likely cannot double PCB revenues without major new capacity or acquisitions (and they seem disinclined to invest heavily here now). So, we treat the segment as a cash cow with modest organic growth potential.
Segment Valuation (Conservative): Using an Owner Earnings multiple approach, we value Electronics Solutions on a stand-alone basis:
Normalized Owner Earnings: We use ~€150 million as discussed. This is a mid-cycle figure reflecting, say, ~€1.1B revenue at 20% EBITDA margin, 5% maintenance capex, 25% tax. It’s arguably cautious (below the FY22 boom, above the FY24 bust).
Growth Assumption: We assume 0–3% growth for valuation purposes. To be conservative, we might capitalize it as if no growth (or minimal) – effectively using an earnings yield approach.
Appropriate Multiple: In a no-growth, cyclical manufacturing business, a conservative multiple might be 8× Owner Earnings (which equates to a 12.5% earnings yield). For context, peers trade at maybe 10–12× EBITDA in current market, but we choose a lower multiple to embed caution (and to account for higher risk of PCBs commoditizing further). Benjamin Graham might even argue for 6–7× in a very uncertain case, but given AT&S’s high-tech niche, we’ll use 8× as base, and at most 10× in a more optimistic scenario.
Applying 8× to €150m gives €1.2 billion enterprise value for the PCB segment. At 10×, it would be €1.5B. We lean toward the lower end for TGTP calculations.
One can cross-check this: €1.2B EV for PCB corresponds to ~0.8× sales (if sales ~€1.1–1.2B) and ~4× EBITDA (if EBITDA ~€300m peak, but on normalized ~€220m EBITDA it’s ~5.5×). This looks low relative to market comps (TTM’s EV/S ~1×, EV/EBITDA ~12×). The disparity is by design – we are building in a margin of safety at the segment level.
This segment is the crux of AT&S’s future – and also its biggest source of uncertainty. We will analyze its development timeline, investment and ramp progress, yield dynamics (with an engineer’s perspective on learning curves), scenario risks, and then attempt a DCF valuation with conservative assumptions.
Development Timeline: AT&S’s journey in substrates began mid-2010s:
Around 2016–2017, AT&S started volume production of IC substrates at its plant in Chongqing, China (Plant 1). This was a strategic leap from PCBs to substrates, aided by technology transfer and a big capex. The Chongqing facility ramped through late 2010s; AT&S became a qualified supplier, likely for Intel’s smaller chips or other firms, but remained a minor player (global rank ~#5 by 2020).
In the early 2020s, seeing surging demand (ABF substrate shortages), AT&S committed to massive expansions:
It built Chongqing Plant 2 for additional substrate capacity (some of the €600m capex in 2022 went here).
It constructed a new substrate R&D and pilot line in Leoben, Austria, as a “European competence center” (officially opening June 2025). This smaller line focuses on new technology development and will also produce small series for European chip clients.
Most notably, in mid-2021 AT&S announced a greenfield investment in Kulim, Malaysia – its first Southeast Asia site – committing €1.7 billion for Phase 1 to build a giant substrate campus. Construction began Nov 2021. Impressively, by late 2023, Plant 1 in Kulim was erected and by May 2025 AT&S declared it ready for high-volume manufacturing (HVM). This is extremely fast by industry standards (a potential “world record” 2-year build + 1-year ramp).
A Plant 2 shell in Kulim was also built (wind-and-water-tight by late 2024), but AT&S is holding off equipping it until market conditions justify (i.e. until a major customer’s demand improves enough to fill it).
By mid-2025, AT&S effectively has a “substrate triangle”: Chongqing (mature, supplying various customers), Kulim (new, focused initially on AMD), and Leoben (new R&D line). The heavy lifting of capex for Phase 1 is largely done; now it’s about ramping yields and utilization.
Capex History & Depreciation Base: The substrate expansion was extremely capex-intensive:
In FY2021/22 and FY2022/23 combined, AT&S spent ~€1.8–1.9 billion in capex, predominantly on substrate projects (Kulim, Leoben, Chongqing expansion). For instance, in FY22/23 alone net capex was €996m, and another €855m in FY23/24. By FY24/25 capex slowed to €415m as projects neared completion.
Total Phase 1 Kulim investment is ~€1.7B planned, with ~€1.0B invested by mid-2024. The remaining spend likely occurs in FY25/26 for final equipment installations on additional lines.
Regarding Leoben’s IC substrate center:
AT&S’s new Leoben IC-substrate Competence Center (“Hinterberg 3” or HTB3) and the government support behind it:
Project scope, cost and timeline
Official opening: June 3, 2025, at the company’s headquarters in Leoben-Hinterberg.
Construction period: Three years.
Facility size: 11 000 m² of R&D and production space.
Total investment: More than €500 million.
Staffing: Approximately 420 new employees already on the payroll at launch.
Austrian government funding (via IPCEI)
The project falls under the EU’s Important Projects of Common European Interest (IPCEI) for microelectronics.
Austria’s federal government is investing nearly €400 million into two IPCEI microelectronics projects—part of a broader national push to develop high-tech industries.
AT&S CEO Michael Mertin has acknowledged the “close coordination between AT&S, the government of Austria, and the European Commission” that made HTB3 possible.
Additional public-sector financing
On May 9, 2023, the European Investment Bank granted AT&S a €250 million loan to support R&D activities and help finance the Leoben research centre’s construction.
Taken together, the Leoben HTB3 center represents a € 500 million-plus private investment, leverages nearly € 400 million of Austrian federal grants under IPCEI, and benefits from a € 250 million EIB loan—underscoring strong public-private collaboration to establish Europe’s first local IC-substrate R&D and production hub.
As a result, AT&S’s gross fixed assets (and depreciation base) have skyrocketed. Depreciation & amortization rose to €328m in FY24/25 (21% of revenue), up from €276m in FY23/24, and this will rise further as Kulim’s equipment depreciates over useful life (likely 5–10 years). We estimate when Kulim Plant1 is fully operational, total annual D&A could approach €400m. While this depresses accounting earnings in early years, one must remember much of it relates to assets with 10+ year lifespans – relevant in judging economic profits.
The substrate capex has been funded by a mix of debt and internal cash; notably AT&S issued €350m of hybrid bonds in 2022 to bolster capital for this expansion.
Issue date and volume AT&S successfully issued a deeply subordinated “hybrid” bond of € 350 million on 18 January 2022, exceeding its € 300 million target.
Instrument structure
Indefinite term, with a first call date after five years (i.e. callable as of 18 January 2027)
Deeply subordinated (ranked below all senior and most other subordinated debt)
Treated as equity for regulatory and credit‐rating purposes.
Coupon and rate
Fixed annual coupon of 5.00 % until the first call date.
(After 20 January 2027, the coupon resets annually based on the 5-year EUR mid-swap rate plus a fixed spread of 494.2 basis points. The bond follows typical reset conventions (e.g., ACT/ACT ICMA day count, modified following business-day convention), though the full fallback and rounding provisions are only detailed in the listing prospectus.)
Listing and placement
Placed exclusively with institutional investors at a minimum denomination of € 100 000
Admitted to trading on the Regulated Market of the Vienna Stock Exchange.
Basically, AT&S borrowed at a fixed 5.00 % yield on a perpetual, after-five-year call hybrid instrument—an expensive but equity-like form of financing—to underpin its substrate investment alongside internal funding.
Ramp Progress: AT&S has made faster progress than many expected:
Kulim Plant 1: By May 2025, the site achieved “Certified High-Volume Manufacturing” status from AMD and began shipping substrates in volume. According to AT&S, the campus was built in record time (two years construction, one year to reach HVM). It’s producing advanced substrates for AMD’s Genoa/Epyc processors and “other customers” (likely smaller volume for now, but expected to “grow significantly with additional high-profile clients joining this financial year”). The initial output is probably a fraction of full capacity – ramp-up is gradual as more lines are installed and qualified through FY25/26. But hitting HVM means yields have reached a threshold where consistent production is possible.
Leoben substrate line: Also started production in FY24/25 (by early 2025). This will contribute small volume but is key for technology development (perhaps for future 3D or glass substrate tech).
Chongqing: Running more or less at steady state. It invested €600m in Chongqing in FY22/23 to expand capacity, likely to add more ABF lines. This indicates Chongqing’s substrate output was growing through 2023.
Chongqing Expansion – FY 2022/23
Capex amount: ~€600 million dedicated to the Chongqing facility during FY 2022/23.
Primary goal: Scale up ABF substrate capacity—critical for high-performance computing, servers, PCs, and 5G base stations—addressing strong customer demand.
Strategic context:
The company’s investor presentation (Q3 2021/22) showed a multi‑year investment roadmap of ~€1.2 billion for ABF capacity (Chongqing + Kulim), aiming to dramatically increase substrate throughput.
AT&S later confirmed that “Plant III” in Chongqing was progressing and the “additional capacity for ABF substrates” was a key driver behind its record €1.07 billion H1 2022/23 revenue.
Overall, by FY2024/25, AT&S’s substrate segment had near-flat revenue (~slight decline) as existing lines slowed but new ones hadn’t contributed fully. FY25/26 should see a pickup as Kulim’s shipments ramp. AT&S had initially guided FY24/25 group revenue €1.7–1.8B with new plants contributing late, but ended at €1.59B – implying ramp was a bit delayed or slower than hoped in H2. However, the Outlook for FY2025/26 is optimistic: they expected ~€400m revenue in Q1 (vs €349m prior year), a ~14% jump mainly due to new capacity coming online. And crucially, they project doubling revenue to ~€2.2B by FY2026/27 largely from Kulim and Leoben contributions. We must temper this with caution but note it as their target.
Yield Curve Analysis: Yields are —THE— critical factor in substrate economics. When a new substrate plant starts, yields (the percentage of produced units that meet all specs) are low – meaning a large portion is scrap/rework, driving up cost per good unit. Over time, as the team fine-tunes processes, yields improve, reducing scrap and improving margins. We attempt an “engineer’s forecast” of yields:
Expected Initial Yield (start-of-ramp): For complex ABF substrates, initial yields can be on the order of 50–70% in early mass production. Given AT&S’s fast ramp, let’s assume Kulim’s initial HVM yield was in the upper part of this range – perhaps ~70%. Indeed, achieving HVM certification from AMD suggests AT&S hit a threshold yield (maybe around 80% for that specific qualified product) – AMD would not rely on them if yields were disastrously low. However, since new production lines are still being added, each new line may start at lower yield and climb up the learning curve. So across the whole plant, initial average yield could be ~60–70% in FY24/25.
Conservative overlay: Industry data isn’t public, but conversations with semiconductor packaging engineers suggest initial yields for new substrate lines often start around 50-70%. We lean to 70% for AT&S Kulim because of their prior experience and focusing on one lead customer (AMD), which helps debug issues faster.
Historical Yield Improvement (Learning Rate): Semiconductor manufacturing often follows a learning curve where yield improves as cumulative output increases. In substrates, one might expect significant improvement within the first 1–2 years of operation. For example, if starting yield is 60%, it might reach ~85% after one year of ramp and ~90+% after two years, assuming competent execution. AT&S’s experience in Chongqing would help shorten learning at Kulim – they can transfer know-how (they explicitly share know-how across sites). Peers like Unimicron or Ibiden have indicated yields for new advanced substrate lines ramp to ~90%+ over a couple of years under normal circumstances. Let’s assume AT&S can get to ~90% yield on mature lines by FY2026 (2 years post-HVM), and possibly ~95% in steady state.
Learning curve models (like Wright’s Law) in semiconductors show significant yield improvements as cumulative output doubles. If AT&S doubles output every few months in early ramp, yield could increase 5-10 percentage points within a year. We assume a roughly linear climb from ~70% to ~90% over two years for simplicity.
Impact on EBITDA Margin: Yield directly hits cost of goods: e.g. at 70% yield, for every 1 good unit, ~0.43 units worth of materials/labor are wasted (1/0.70 = 1.43x cost factor). At 90% yield, waste is only 0.11 units per unit (1/0.90 = 1.11x cost factor). We can model substrate gross margin as a function of yield: If at perfect yield (100%) EBITDA margin would be perhaps ~30% (comparable to top peer margins in tight supply), then at 70% yield the margin might drop near 0%. Indeed, a quick estimate: assume at 100% yield, cost of goods is 70% of revenue (30% margin). If yields are 70%, effective cost becomes 70% * (1/0.70) = 100% of revenue – essentially erasing margin. At 85% yield, effective cost ~82% of revenue (yield factor 1/0.85=1.176, *70% = 82.3%), giving ~17.7% margin. At 90% yield, cost ~77.7%, margin ~22.3%. At 95% yield, cost ~73.7%, margin ~26.3%. This simplified model shows how margin could improve from ~0% at start to ~25%+ once yields stabilize in the 90s%. AT&S’s own adjusted EBITDA margin outlook for FY26/27 is 24–28% which aligns with achieving high yields by then.
NOTE: If effective cost equals revenue, EBITDA margin ~0%. Our simplified model treats 30% as the inherent gross profit margin at perfect yield; actual fixed costs mean margin wouldn’t fully drop to zero at 70% yield (there might still be a small positive margin if selling price covers variable costs). But it would be very slim, consistent with management saying start-up costs burden margin heavily. The sensitivity is for directional understanding.
We present a sensitivity table (approximate) linking yield to margins and annual Owner Earnings for the substrate segment (assuming future revenue ~€1.2B and maintenance capex 5% of rev)
Table: IC Substrate Yield vs. Profitability Sensitivity (illustrative). Higher yields dramatically improve margins. (Assumptions: ideal margin ~30% at 100% yield; tax ~25%; maintenance capex 5%.).
In probability terms, early ramp years (FY25–26) might see yields in the 75–85% range (translating to low or mid-teens margins). By FY26/27, we expect yields approaching 90%, unlocking margins ~22–25%. Our base case will assume they ultimately reach ~90% yield (mid-20s% EBITDA margin) by year 3 of ramp, but we’ll also consider a downside where yields plateau at e.g. 85% (limiting margin <20%).
Risk Analysis – Substrate Segment: This expansion comes with significant risks, which we enumerate with conservative vigilance:
Start-up Drag: As shown, the new capacity initially operates at suboptimal yields and under-utilization, creating a drag on earnings. In FY2023/24, AT&S incurred heavy start-up costs in Kulim and Leoben that cut segment EBITDA to €96m (down 17%) despite decent revenue. This drag continues into FY2025/26 – management guided only ~16% EBITDA margin in Q1 25/26 because further line start-ups will “continue to burden the earnings/revenue ratio”. We must expect a couple more quarters (or even years) where substrate operations contribute little or even negative EBIT until fully ramped. We account for this in our DCF by low margins in initial years. The risk is that the drag lasts longer than expected, eating into corporate cash and testing debt covenants.
ABF Supply Constraints: The substrate build-up relies on third-party materials like ABF films. In the last few years, ABF film supply was so tight it constrained substrate production globally. If Ajinomoto or others cannot supply enough film (or if prices spike), AT&S’s costs could rise or output be limited. So far, the industry is investing in more film capacity, and with a slight market lull, this may not bite immediately. But if AI-related substrate demand soars unexpectedly (e.g., Nvidia orders skyrocket), material bottlenecks could resurface. We view this risk as moderate, as supply chain is catching up, but we flag it as a potential margin suppressor (higher input cost or forced idle time waiting for materials).
Customer Qualification Failure or Delay: A worst-case scenario would be if AT&S’s new substrates fail to meet a key customer’s specs, causing rejection of parts or loss of future orders. This is somewhat all-or-nothing: For AMD, AT&S did get certified and is shipping, mitigating this risk for that program. But as AT&S seeks “additional high-profile clients”, each new engagement carries risk—customers will sample products and could decide yields/quality aren’t sufficient and stick with incumbents. Any such failure would strand some capacity (at least temporarily) and tarnish AT&S’s budding reputation in substrates. So far, we have positive signals (AMD’s endorsement, and presumably other unnamed clients have started small orders), but we remain cautious: execution must be flawless to expand the roster.
Research memo – Customer qualification failure / delay risk
Thesis. In advanced IC-substrate contracts the commercial upside is binary: if a product does not clear a customer’s electrical, mechanical and reliability tests, volumes disappear and installed capacity sits idle. AT&S has demonstrated the ability to pass at least two tier-1 qualifications (AMD and Intel awards), but every new design-win cycle effectively “resets the clock.” The risk therefore moderates – but never disappears – as the customer roster broadens.
Evidence to date
AMD public certification & shipments - AT&S substrates are already in high-volume AMD data-centre processors and accelerators. AMD publicly endorsed AT&S’s “very high quality requirements.” Source: ats.net
Kulim plant goes HVM for “AMD and other customers” - The newest Malaysian fab has moved from pilot to high-volume mode and is shipping not only to AMD but to “other customers,” implying at least one additional programme certified. Source: evertiq.com
Intel EPIC Supplier Award 2025 - Only 26 suppliers globally received Intel’s highest vendor accolade in 2025; criteria include “quality, cost, availability and technology.” This is a strong, though not definitive, signal that AT&S has cleared Intel’s substrate-quality bar. Source: LinkedIn
Management disclosure: “customer diversification successfully initiated” - AT&S states it has secured additional IC-substrate customers that are co-financing capacity in Leoben, signalling concrete design-wins beyond AMD. Source: ats.net
Why the risk still matters
Each node = new qual cycle. Every time a customer shrinks line-widths, adds layers or introduces new materials (glass test vehicles, ultra-fine ABF, etc.), AT&S must re-qualify – success in an N-1 generation is no guarantee in N.
Process-yield tightrope. Early ramp yields are volatile; a minor defect spike can push yield below the contractual floor and trigger re-inspection or RMA, creating weeks-long gaps in shipments.
High customer concentration. Even with diversification, two or three hyperscale CPU/GPU vendors will still make up the majority of substrate revenue; losing one design window can strand an entire line.
Reputation lock-in. Incumbents like Ibiden, Unimicron and Samsung Electro-Mechanics already enjoy multi-decade track records. Any headline failure by AT&S would hand ammunition to procurement teams and slow follow-on engagements.
Probability & impact assessment
Market Risk – Demand Shortfall: The expansion was predicated on robust demand forecasts (like the €3.5B FY2027 revenue goal mentioned by prior CEO). However, the market can shift. PC and standard server demand has been weak; AI chips are hot but could also face cycles or double-ordering risk. If global recession hits or tech capex slows, AT&S might find itself with excess substrate capacity. This is particularly concerning given Plant 2 in Kulim is ready to equip – if demand doesn’t justify it, that sunk building is an idle asset generating no return (but still maybe incurring some maintenance and interest). We consider a scenario where substrate revenue tops out at e.g. €700–800m (instead of >€1B) if AI/CPU demand disappoints. In that case, the segment would underutilize its new lines, and ROIC would be poor. We mitigate this by using conservative revenue ramps (slower than company’s optimistic case) in our base valuation.
Debt Funding & Covenant Breach: The substrate build was largely debt-funded. AT&S’s net debt/EBITDA spiked to 6.1× by Dec 2024 (prior to the Korea plant sale) – an extremely high level. The sale brought leverage down to 2.5× by Mar 2025, but that includes the one-time gain in EBITDA; on an adjusted basis it’s closer to ~3–4×. The company has a hybrid bond €350m (4.75% coupon) treated as equity in accounts, and significant bank debt. Should the substrate ramp falter (yield or demand issues), EBITDA could undershoot and leverage spike again, potentially breaching covenants. We don’t have exact covenant terms, but typically banks might require e.g. net debt/EBITDA < 4.0×. A scenario of prolonged low EBITDA (e.g. <€200m) while debt stays >€1.0B could indeed trigger breaches. AT&S is mitigating by not paying dividends and cutting costs, and it has some headroom with €256m undrawn credit. But in a downside scenario, the risk of dilution via equity raise or an expensive refinancing is real. For instance, if FY2025/26 saw another operating cash outflow and EBITDA only ~€300m (with interest >€80m), AT&S might need to raise capital by FY2027 to relieve debt. We account for this risk by increasing the margin of safety required.
Geopolitical risk (China etc.): Although not specific to substrates alone, any geopolitical event (e.g. China-Taiwan conflict) could roil the substrate market. Paradoxically, it could benefit AT&S (if Taiwanese suppliers are disrupted, AT&S’s Kulim and Austrian output become more critical), but it could also hurt if global tech trade is disrupted. It adds uncertainty to long-term substrate demand patterns (one reason AT&S diversified geography to Malaysia).
Technology Risk: The substrate business faces the long-term challenge of evolving technology. While likely not before 2027, future packaging innovations (like multi-chip modules, 3D stacking with interposers, or even new substrate materials) could reduce the need for large conventional substrates. AT&S is working on next-gen (hence the R&D center in Leoben), but if it misses a tech shift, some capacity could become obsolete. We assume no drastic disruption in our near-term analysis, but acknowledge this as a “terminal value” risk (hence we use conservative terminal assumptions).
Given these risks, we may allocate a higher internal discount rate for substrate cash flows. In our DCF, we will stress test downside cases (e.g. slower yield ramp or half-filled plant) to ensure the valuation is robust only to the more likely moderate success scenario, not requiring perfection.
Segment Financials & Current Performance: As a snapshot, in FY2023/24 the Microelectronics segment had external revenue likely around €500–550m (slightly down from ~€550+ in FY22/23) and reported EBITDA €96m (margin ~17%). This included startup costs burden; excluding those, adjusted segment EBITDA might have been higher (the company said overall adjusted EBITDA excluding startup was €384m vs reported €307m, so roughly €77m of startup costs group-wide, largely substrate-related). Thus, substrate core profitability was probably still healthy if we exclude expansion costs – a sign that Chongqing’s established lines were profitable, but Kulim/Leoben pre-revenue costs dragged the number down. This nuance is crucial: the underlying substrate business (existing operations) is sound, but the new capacity is pulling it down temporarily. We expect FY2024/25 saw a similar dynamic: some startup cost, offset partly by initial Kulim revenue plus Chongqing’s base. For valuation, we focus on the future steady-state potential once ramps are done.
Segment Valuation via Conservative DCF: Valuing the substrate segment is inherently speculative given the forward-looking ramp. We approach with a scenario-based DCF using realistic yield and utilization trajectories:
Revenue Projections: In a base-case (conservative) scenario, we assume the substrate segment’s revenue grows from ~€500m in FY2023/24 to approximately €700m in FY2025/26 (as new capacity contributes for a full year) and to €1.1–1.2 billion by FY2026/27 (when Kulim Plant 1 is fully utilized). This aligns with management’s lower-end guidance of ~€2.1B total revenue in 26/27, of which roughly half could be substrate. We then assume growth tapers – maybe reaching ~€1.3B by FY2028/29 if Plant 2 isn’t equipped, or higher if partial Plant 2 added. For prudence, we won’t assume Plant 2 expansion in our base case (that would be upside we can discuss separately).
Margin (EBITDA) Trajectory: We incorporate yield improvement in margins:
FY2024/25: margin low (actual was ~19% EBITDA for segment, aided by one-time sale adjustment in other segment though). We assume ~15% true underlying margin.
FY2025/26: margin improving to ~15–18% as yields maybe ~80–85% average (some lines still ramping).
FY2026/27: margin ~24% (mid-point of co. outlook) as yields ~90% and volume high.
FY2027/28 and beyond: margin perhaps 25–26% steady (we cap it below peers like Ibiden which might do ~30% at peak, to reflect AT&S likely still has slightly higher costs or customer-mix concessions).
Capital Expenditures: The heavy growth capex phase is essentially ending in FY25/26. We model capex in FY25/26 still high (maybe €300–400m to finish Kulim lines), then dropping sharply. From FY26/27 onward, we assume only maintenance capex ~5% of revenue for substrate (since any further big expansion would be discretionary and we’d treat it separately). Maintenance 5% of, say, €1.2B is €60m. FALSE, see research note below , I would argue 10% is better which would run €120m.
Research Memo about the 5%:
Does a “maintenance capex = 5 % of substrate revenue” rule-of-thumb hold up for AT&S?
What that tells us
5 % looks too skinny. Every hard datapoint—AT&S depreciation, Ibiden depreciation, Unimicron “post-boom” capex—clusters in the high single to low-double-digit range.
Why depreciation is the right anchor.
IFRS depreciation for substrate tools is typically 5-7 years; if new investment slows, cash maintenance capex should gravitate toward that depreciation line.
AT&S’s €276 m D&A implies a capital base of ~€1.6 bn; replacing just 1/7th of that each year would already require €230 m+, or ~13 % of €1.8 bn sales.
Reasonable maintenance band for substrates:
Base case: 8 – 10 % of sales (aligns with Unimicron’s forward guide once residual growth spend is gone).
Bare-bones “run to fail”: maybe 6 – 7 %, but that risks higher unplanned downtime and hit to yields.
Rule-of-thumb for modelling AT&S from FY 26/27 onward
Use €90 – 120 m (7.5 – 10 % of €1.2 bn substrate revenue) as maintenance capex.
Treat anything above that as capacity-adding “growth” spend; anything below it effectively assumes life-of-tool extensions the auditors will almost certainly push back on.
Bottom line
A 5 % ratio would imply AT&S can run an asset base that depreciates in ~20 years—twice the economic life the rest of the industry experiences. Unless management explicitly guides to radically lower capital intensity (and auditors bless much longer depreciation schedules), a more defensible maintenance figure is in the 8 – 10 % of sales range. Modelling €60 m on €1.2 bn risks understating ongoing cash needs by ~€40-60 m a year.
Working Capital: We won’t delve deeply; assume it scales modestly with growth. AT&S did reorganize its factoring program, which might temporarily impact cash flow, but that’s a one-time effect not fundamental (factoring removed in FY24/25 caused CFO to drop). For DCF simplicity, we assume changes in working capital roughly net out over the period.
Discount Rate: Given the risk, a conservative investor might use a high hurdle (say 12–15 %). We’ll use ~12 % (realistic for a high-risk segment).
Putting numbers (approximate, in € millions):
Thereafter we assume terminal growth of ~2 % (inflation-like) with a ~27 % EBITDA margin and capex continuing at ~10 % of sales.
Summing all = €930 million enterprise value for the substrate segment.
This is a rough figure, but it gives the order of magnitude: ~€930 million EV. We should view this as the value in a conservative success case (Kulim fully ramped, yields in the low-90 % range by 2028). The DCF already embeds two cash-burn years and a heavy discount on distant profits, so the stance is appropriately cautious.
We could scenario-weight a downside: suppose yields stall at 85 % (EBITDA margin capped around 18-20 %, revenue peaks near €800 m). In that case, steady-state FCF collapses and the segment might be worth only €500-700 m. Conversely, if a second Kulim line (Plant 2) adds another €1 bn of revenue by FY-2028, the substrate EV could rise toward €1.6-1.8 bn. As conservative investors, we prefer to value only what is visible and already financed—namely Kulim Plant 1 plus the Leoben competence centre.
To bake in extra prudence we could trim this further—or, equivalently, apply a probability-of-success haircut—but remember we will impose a firm margin-of-safety at the total-company level. Using the €0.93 bn base EV is therefore acceptable for now; we will revisit in the overall valuation section when we allocate debt.
Check vs. peer comps: if AT&S eventually delivers ~€1.2 bn revenue and ~25 % EBITDA margins, that is roughly €300 m EBITDA. Japanese peers like Ibiden currently trade around 8-10× EBITDA (i.e., €2.4-3.0 bn). Our €0.93 bn figure equates to just ~3× that future EBITDA—clearly conservative and consistent with not baking in rosy execution.
Finally, when apportioning net debt across segments we will implicitly attribute part of the borrowings to substrates, perhaps ~€600 m, with the remainder charged to the PCB division. In the end, the group valuation subtracts total debt from the sum-of-parts, so this internal split only affects segment optics, not consolidated equity value.
Having valued each segment conservatively (PCBs ~€1.2B, Substrates ~€0.93B EV), we will proceed to look at the company’s overall structure (ownership & governance) and then combine these values in Section 5 (Valuation) after assessing net debt and other adjustments.
3. Ownership & Governance
Understanding who owns and runs AT&S is important for a value investor, as it sheds light on alignment of interests, capital allocation discipline, and potential governance risks. We examine the shareholder structure, insider incentives, and any governance red flags.
3.1 Shareholder Breakdown
AT&S’s ownership is characterized by two significant long-term shareholders and a broad base of institutional investors:
Dörflinger Private Foundation (~18% ownership): This is the largest shareholder, with approximately 18.1% of outstanding shares. The Dörflinger family has been involved with AT&S since the 1990s. The foundation likely represents the interests of Johannes Dörflinger (or his family), who was a co-founder or early investor in AT&S. Indeed, historical records show Dr. Hannes Androsch (see below) and a partner acquired AT&S in 1994 from Philips, and that partner was Willibald Dörflinger – presumably related to Johannes Dörflinger. This stake suggests a stable, insider-like investor with a multi-decade perspective.
Androsch Private Foundation (~17.6% ownership):Dr. Hannes Androsch – a prominent Austrian industrialist and former finance minister – has been AT&S’s Supervisory Board Chairman for many years and, through his private foundation, owns about 17.5% of the shares. As of a recent disclosure, Androsch and Dörflinger together control ~35.6% of AT&S, which effectively means the company has a strong insider bloc. Androsch is essentially the company’s elder statesman; his involvement since 1994 (when he led the buyout that established AT&S as an independent company) indicates deep commitment. This ~35% combined holding by the two foundations is crucial: it likely prevents any hostile takeover and provides strategic guidance continuity. (NOTE: He passed in december 2024)
Erste Asset Management (~4%): The next known holder is Erste Group’s asset management arm with about 4.0%. This represents institutional ownership (likely Austrian/European funds). At 4%, Erste is a notable but not controlling investor.
Others / Free Float (~43%): The remainder (around 43–45%) is free float held by various institutional investors (some names appear in fund databases) and retail shareholders. According to one source, as of mid-2025 institutions overall held ~34.7%, which is in line with the known big ones plus numerous small positions. There is no single strategic corporate owner or government stake visible; the ownership is primarily these two private foundations and market investors.
Impact on Valuation: The concentrated insider ownership can be a double-edged sword:
On the positive side, having ~35% held by insiders (Androsch & Dörflinger) means management decisions are likely aligned with shareholder value creation in the long run – these owners have a lot of skin in the game. It also suggests any major capital decisions (e.g. equity raise, M&A) would need their buy-in, which likely ensures caution and a focus on not diluting value. For value investors, such strong insider alignment is usually a good sign – it can lead to prudent capital allocation and resistance to short-term market pressures.
On the other hand, with >1/3 control, these insiders can sway corporate votes; minority shareholders must trust their judgment. The upside is they’ve been steady stewards (no indication of self-dealing or exploitation in the record). The downside could be if they have personal agendas (e.g. empire building) that diverge from minority interests. However, given that both insiders are in their 70s/80s (Androsch is in his 80s, I believe), they likely prioritize the legacy and success of AT&S rather than reckless expansion – though one could argue the huge substrate gamble itself was a bold, perhaps risky call presumably backed by these owners.
Another effect on valuation: the relatively low free float (~45–50%) and the mid-cap size (market cap ~€760m at current price) means liquidity is somewhat limited, possibly leading to a “conglomerate” or liquidity discount in the market. But for our intrinsic value analysis, that is less relevant except to note that acquiring or exiting a large position might move the stock.
In sum, ownership is tightly held by aligned insiders, which we view favorably from a governance perspective (with some caveats noted below).
3.2 Insider Alignment & Stock-Based Compensation
Insider Ownership vs Industry Norms: At ~35% insider/founder ownership, AT&S is far more “owner-led” than a typical European tech manufacturer. This high insider stake is actually reminiscent of many family-owned European industrials. It’s a positive sign for alignment: management is effectively working for themselves as shareholders. It also means less risk of hostile takeover without insiders’ approval, allowing management to focus on long-term strategy (like the substrate expansion) even if short-term earnings suffer. By comparison, most peers (e.g. large Asian PCB firms) are more widely held or part of conglomerates. The insiders’ large stake likely gave confidence to creditors when raising debt for expansion – they know the owners have much to lose if things go awry.
Recent Insider Buying/Selling: We should check if insiders have been buying more shares or selling:
There was a news of Androsch transferring 202,095 shares in Dec 2021. It appears to be a transfer (maybe to a family trust or between entities), not an open-market sale. Actually, it is suggested that by December 2021 Androsch and the Dörflinger foundation increased their combined stake from 16.3% to 17.55% (which likely was Androsch buying some shares to strengthen position). Indeed, an AT&S ad-hoc release (Sept 2021) noted that these insiders raised their holdings to 17.55%. This suggests they doubled down around the time the Kulim project was launched – a reassuring sign that they believed in the strategy enough to up their stake.
We haven’t seen reports of large insider sales in the market. In fact, given the share price declined in 2022–2023, insiders did not panic-sell as far as disclosed. This stability is encouraging.
The only caution is Androsch is aging; at some point, succession or distribution from his foundation could lead to shares being sold (e.g. his sale of a small portion in 2021 might have been estate planning). But with two large holders, even if one trimmed, the other might maintain influence.
Management Compensation Structure & KPIs: AT&S’s executive board (the CEO, CFO, etc.) likely have a compensation mix of base salary, annual bonus tied to financial KPIs, and possibly a long-term incentive plan (LTIP). While we don’t have exact figures from an annual report here, typical KPIs in this industry are EBITDA margin, revenue growth, ROCE, and project milestones (like successful ramp of Kulim). The fact that AT&S in FY2023/24 had a loss and still cut workforce suggests management felt pressure – indeed they took steps like zero dividend and cost cuts, implying their incentives are aligned with turning a profit again. Austrian corporate governance code would require disclosure of pay and KPIs; reading between lines, they likely emphasize ROCE and EBIT margin because of the heavy capex (there was an IR remark that sustainable cost optimizations would yield €180m savings from 23/24 onward, hinting management’s targets include cost and margin improvement). For value investors, it’s good if management is return-focused (ROCE) versus just growth. We suspect after an investment binge, there’s pressure now to show returns.
Research Memo: Management Compensation & KPI Alignment at AT&S
Compensation mix
Base salary paid in 14 equal instalments, covering overtime and travel.
Annual bonus tied to financial targets (e.g. EBITDA margin, revenue growth, ROCE) and project milestones (e.g. Kulim ramp)
Long-Term Incentive Plan (LTIP) (PSUs & RSUs), with performance metrics including adjusted EBIT/EBITDA and relative TSR vs. ATX/MDAX peers; clawback on materially misstated results.
Regulatory framework
Under the Austrian Corporate Governance Code, the Supervisory Board must publish a clear remuneration policy detailing fixed vs. variable pay, all performance criteria (financial & non-financial), waiting/retention periods, and clawback provisions.
Evidence of KPI focus & alignment
ROCE target >12 % baked into forward guidance for FY 26/27
Cost-savings KPI: “sustainable cost optimisation” programmes aimed at ~€ 180 million p.a. savings from FY 23/24 onward
Efficiency execution: FY 23/24 saw an intensified cost-cutting drive and headcount reduction of up to 1,000 employees to defend margins
Incentive alignment signals
Zero dividend proposed for FY 23/24—management foregoing payout to prioritise balance‐sheet and reinvestment objectives
CEO/CFO commentary underscores using profits to “return to profitable growth” and enhance company value
Conclusion: AT&S’s executive pay structure (salary + bonus + LTIP) and published targets (EBITDA margin, ROCE, cost savings, project milestones) align management incentives squarely with restoring and growing profitability post-capex. Continuous public disclosure under Austrian governance rules provides transparency on these KPIs.
Stock-Based Compensation (SBC): AT&S is not a Silicon Valley tech firm, so SBC is relatively minor. Let’s see the share count: it has remained stable at 38.85 million shares for several years. It stayed the same from Mar 2024 (39M) to Mar 2025 (39M). This indicates no significant equity dilution recently, aside from maybe small issuances for stock options. If any management stock option plan exists, it’s not large enough to move the outstanding share count meaningfully (likely <1% per year). Indeed, if maintenance capex was capitalized from a hybrid or debt, they avoided issuing new equity during the expansion – which, from an alignment perspective, shows they preferred to sacrifice credit rating rather than dilute shareholders. This is a double-edged sword (more debt risk), but it signals management/owners strongly care about existing shareholders’ value (since they themselves are the largest shareholders).
We should still check if SBC is expensed properly and how much:
AT&S probably has a Long Term Incentive Plan (LTIP) granting performance shares to executives if targets are met. The cumulative dilution from these could be a few hundred thousand shares over years (<<1% of float).
We didn’t find an explicit number in sources, but given share count fixed at 38.85M, we infer any issuance was offset (maybe by buying back small number to satisfy plan, or negligible).
From a value perspective, dilution has been minimal (~0% last 5 years), which is excellent. The company even paid small dividends historically (e.g. €0.40/share for FY2021/22) until they halted it to conserve cash in FY22/23 and 23/24.
Insider Alignment Summary: With ~35% held by insiders and negligible SBC dilution, AT&S’s leadership is highly aligned with shareholders. We have essentially an owner-operator model. This gives some comfort that decisions – like the bold substrate expansion – were taken with an owner’s risk-reward calculus (they truly believed in long-term value creation, not just empire-building on OPM (other people’s money), since it’s largely their own money at stake via equity). For instance, when they suspended dividends for two years to strengthen the balance sheet, insiders suffered equally from not receiving those payouts, signaling commitment to long-term health.
One minor concern: hybrid bond treatment of interest – hybrids typically allow deferring coupons. If things got tight, management might defer hybrid interest (which wouldn’t be great for credit but avoids default) to avoid equity raise, again preserving shareholder value at possibly debt holders’ expense. This is more of a financial engineering point, but it shows they have tools to avoid dilution.
In conclusion, insiders are strongly aligned, and SBC is not a material drag. The key person risk might be if these anchored owners step back (succession planning – e.g. Androsch eventually retiring). However, AT&S did recently get a new CEO (Mr. Michael Mertin, effective 2023) after long-time CEO Andreas Gerstenmayer stepped down, indicating some renewal. We should watch how the new management performs. But presumably, the foundations maintain oversight via the Board.
3.3 Governance Red Flags
Let’s scrutinize governance aspects for any red flags:
Dual-class Share Structure: AT&S has a traditional single-class share structure (one share, one vote). There are no multiple voting shares or special preferential shares as far as disclosed – both major owners hold ordinary shares. So minority shareholders are not structurally disenfranchised by share class. This is good.
Board Independence: The Supervisory Board is chaired by Dr. Hannes Androsch (the largest individual shareholder). While he is clearly not “independent” (he’s an insider/owner), his presence is arguably a positive given his stake, as he represents shareholder interests. Other board members should be checked for independence and expertise. Historically, AT&S’s board had a mix of industry and finance professionals, including presumably independent directors (Austrian law often requires some representation of different constituencies). We do not have the full list here, but no obvious controversy is known. The heavy insider presence could be a concern if they override minority voices; however, Androsch’s long tenure and the company’s growth suggest a generally effective board. We don’t see egregious related-party deals or anything that would signal board capture.
Research Memo: Supervisory Board Independence & Composition at AT&S
Chair & major shareholder
Until December 2024, Dr. Hannes Androsch (Estate of Androsch Hannes) served as Chair and held 17.55 % of voting rights, aligning leadership with major‐shareholder interests but compromising formal independence.
On July 3, 2025, Andy Mattes (McKinsey veteran) was appointed Chair—classified as independent under ÖCGK C-Rules 53 & 54.
Board composition & independence
Total members: 8 active shareholder‐elected seats.
Independent directors (5):
Andy Mattes (Chair) – independent per ÖCGK C-Rules 53 & 54
Dr. Georg Riedl (Deputy Chair) – independent per C-Rule 53
Dr. Gertrude Tumpel-Gugerell – independent per C-Rules 53 & 54
Georg Hansis – independent per C-Rule 53
Dr. Karin Schaupp – independent per C-Rules 53 & 54
Insider representative (1): Dr. Hannes Androsch (former Chair)
Employee representatives (2): Works-Council delegates (Ronald Arh, Christa Köberl, Günter Pint; three delegates rotate within eight seats).
Governance code compliance
Austrian Corporate Governance Code (ÖCGK) C-Rule 53 mandates a minimum of three independent shareholder representatives; C-Rule 54 requires independent majorities on audit and remuneration committees. AT&S exceeds both thresholds.
Expertise & oversight
Directors bring diverse backgrounds: legal (Georg Riedl), finance and banking (Tumpel-Gugerell), engineering and operations (Hansis, Schaupp), and strategic advisory (Mattes).
Three dedicated committees (Audit; Nomination & Remuneration; Financing & Strategy) are chaired by independents, strengthening oversight.
Related-party & controversy check
No material related-party transactions or shareholder‐conflict cases have been flagged in press releases or filings.
Board self-evaluations and commitment to ÖCGK show no governance concerns to date.
Conclusion: While the former Chair’s significant shareholding blurred formal independence, AT&S now features a clear independent majority—with an independent Chair and five of eight directors classified as independent under ÖCGK. This balance, combined with no apparent related-party controversies, supports robust oversight and alignment with minority‐shareholder interests.
Related-Party Transactions: No major related-party transactions are noted in filings or news (like the big capex was all arms-length purchases of equipment, etc.). One related party event was the sale of the Ansan, Korea plant – that was sold to WUS Printed Circuit (China) via Somacis (Italy). This appears to be a third-party sale (Somacis is unrelated to AT&S’s owners, as far as we know). The price was fair (€405m) and gave a gain, which is good for shareholders. We don’t see evidence of the foundations engaging in any dealings with AT&S (like renting property or anything). The only nuance: Androsch is an Austrian statesman, sometimes there could be government links (e.g. he might have helped secure subsidies or financing, but that’s beneficial, not a conflict necessarily). So, no red flags found.
Capital Allocation History: This is critical. AT&S historically paid small dividends when profitable (e.g. €0.40/share for FY21/22, and previously similar modest payouts). In FY22/23, they cut the dividend to €0 due to losses and heavy investment, and similarly propose no dividend for FY24/25. This is a prudent move to conserve cash during expansion – not a red flag, but something to note for income investors. They also issued a hybrid bond instead of equity, showing a preference to avoid dilution (which aligns with insider interests but increased leverage risk). They haven’t done share buybacks (none indicated), which makes sense given they needed cash for growth. There’s no significant M&A in the past decade aside from building new plants; the one divestiture (Korea) was actually a wise move to exit a non-core low-end business for a good price – that sharpened their focus and improved the balance sheet.
One could question: Was going into IC substrates reckless or smart? It was certainly bold – effectively a bet-the-company type move (investing over €2B in new tech capacity). From a conservative viewpoint, it’s risky to leverage a solid PCB cash cow to enter a volatile, capital-hungry semiconductor supply chain. However, management likely judged (correctly) that staying solely in PCBs would lead to stagnation or decline long-term (PCBs are maturing, possibly commodifying). The upside of substrates was huge if executed right, and AT&S had some successful initial experience in Chongqing. Moreover, they timed it when market demand was very high (2019–2021). In hindsight, the cycle turned down at the wrong time (2022–2023 slow patch), which has strained the company. But reckless would be if they ignored risks; they did, for example, strengthen equity via hybrid and get anchor customer AMD, which shows a measure of prudence. Still, a value purist might have preferred they grow slower with less debt. The counterpoint is now AT&S is among the few in a high-entry-barrier field – if they succeed, it secures a competitive position for a decade. We lean that it was a strategically smart but financially aggressive move. It definitely heightens near-term risk (hence our MoS).
Transparency & Reporting: Another governance aspect is how transparent management is. The company does provide segment reporting, and its IR releases have been candid about challenges (they issued profit warnings when needed, etc.). For example, they adjusted guidance as needed (cut FY22/23 outlook when market softened). This honesty is good. They also started giving only quarterly guidance due to uncertainty – some might see that as a lack of visibility, but it’s arguably realistic and protects them from over-promising. No evidence of accounting shenanigans or restatements.
In summary, no major governance red flags stand out. The structure is straightforward, insiders are heavily invested, and the board seems to act in shareholder interest. The main “flag” one could cite is the ambitious capital allocation to expansion, which, while not improper, increases risk. But as value investors, we account for that via required margin of safety rather than labeling it governance malpractice.
One more point: AT&S is listed in Vienna and follows Austrian Corporate Governance Code. Austria’s governance has had issues in some companies historically, but AT&S appears well-regarded. They even have ESG ratings (MSCI BBB, not bad). So likely they adhere to governance best practices like board committees, etc.
Capital Allocation – Dividend, Hybrids, Avoiding Dilution: Let’s highlight these:
Dividend policy: Historically, a small payout (~20% payout ratio in good years). They cut it to zero in FY22/23 and FY24/25 to conserve cash. We view that positively – management is prioritizing investment and balance sheet over appeasing yield-hunters.
Use of Hybrids: The hybrid bond (effectively perpetual subordinated debt) issuance of €350m in Jan 2022 was a tool to raise equity-like capital without diluting shareholders. They even did an exchange offer to replace an older €175m hybrid (which had 5% coupon) with the new one at 4.75% – likely extending maturity indefinitely. This shows financial savvy. Hybrids are counted as equity by rating agencies (usually 50%) and on balance sheet (if structured with discretionary coupons, IFRS counts them in equity). Indeed, the hybrid helped keep the equity ratio >20% even as debt soared. So, management was proactive to strengthen capital structure while shielding share count. The flip side is eventually that hybrid may need to be refinanced (perhaps in 2027 when first call date comes).
No share issuances: There’s no indication of rights issue or placements in the last decade, which is remarkable given the scale of expansion. Many companies would have issued equity for a project this size. AT&S’s owners instead took on debt – which implies great confidence (they’d rather lever up than dilute their stake). As investors, we must be cautious because this increases downside risk, but it also means if things go well, existing shareholders reap all the upside (rather than it being shared with new investors who came in cheap).
To sum up, governance quality is reasonably high for AT&S. The key players have significant skin in the game. We do not see evidence of self-dealing or poor transparency. The largest concern is the financial risk taken, but that seems a strategic choice rather than governance failure. For our valuation, that means we trust the financial statements and that reported numbers are not manipulated (e.g., no worry about hidden losses, etc., beyond normal adjustments).
4. Financial Quality & Risk Assessment
In this section, we evaluate AT&S’s overall financial condition, quality of earnings, and key risks from a conservative perspective. We’ll normalize the past decade’s financials to see through cyclicality, examine leverage and coverage ratios, working capital health, and exposures (currency, geopolitical). We’ll also model a realistic downside scenario to test resilience – particularly important given AT&S’s high leverage and expansion stage.
10-Year Cash Flow & Income Normalization: AT&S’s reported earnings have fluctuated dramatically in the last 2–3 years due to the expansion and cyclicality. To gauge normalized earning power, we consider approximately the past decade:
From FY2012 to FY2019, AT&S was a profitable PCB-centric company with revenue steadily rising from ~€0.5B to ~€1.0B and net profits in most years (though slim in downturns). For instance, net profit was only €19.8m in FY2020 (year ended Mar 2020) on €1.0B rev, indicating a down year margin ~2%. But it jumped to €136.6m by FY2023 (year ended Mar 2023), when revenue was €1.79B and cycle was up. So the 10-year average net margin might be around mid-single-digits.
The concept of Owner Earnings (Buffett’s definition: cash flow from ops minus maintenance capex) is apt here. We attempt to normalize OCF and maintenance capex:
AT&S’s operating cash flow (OCF) is volatile partly due to factoring and working capital changes. In FY2024/25, OCF was -€75m because they reorganized factoring (meaning they likely stopped selling some receivables, thus absorbing cash into working capital). In FY2023/24, OCF was a hefty +€653m, which was inflated by heavy factoring and perhaps advance payments. Over time, these swings even out.
We consider an average EBITDA margin ~22% (midpoint of up and down years), on average revenue say €1.2B (past few years average), gives ~€264m EBITDA average. If maintenance capex is ~€60–80m and cash taxes ~€25m (given historical low taxes due to incentives in China, etc.), then normalized Owner Earnings might be ~€150–180m per year. This is a rough ballpark: indeed, in FY2021/22 OCF was €349m (a record half-year suggests full-year strong), maintenance portion of capex was small so free cash ~€200m+; whereas in FY2019, OCF might have been ~€100m. So ~€150m average OE seems plausible.
Importantly, this normalized figure excludes growth capex. In reality, AT&S spent far above that in recent years for expansion, which turned free cash flow negative. But that’s intentional investment, not maintenance. For financial quality, we want to see if core operations generate solid cash absent expansion: they did, until they accelerated expansion from 2019 onward (when free cash flow turned deeply negative due to capex).
Earnings Quality: AT&S’s earnings have been impacted by certain one-offs:
Pre-FY2022: fairly straightforward, except in some years R&D costs, etc., but no major unusual items.
FY2022/23 and 2023/24: They had significant start-up expenses and restructuring costs. For analysis, AT&S provides adjusted EBITDA excluding those. In FY23/24, reported EBITDA €307m, adjusted €384m (excluding ~€77m of startup costs). So actual underlying performance was better than the raw number.
FY2024/25: contained a huge one-time gain from the sale of the Korea plant, which boosted EBITDA by ~€325m and net profit by ~€127m pre-tax (since profit went from -37m prior year to +90m, that delta includes the sale gain minus other differences). We must strip this out for any normalized view. They did by giving adjusted EBITDA €408m vs reported €606m.
Over 10 years, if we exclude one-offs, AT&S’s net income would roughly equal the cash flows we estimated above in good years, and near zero in tough ones (but rarely deeply negative until 2024’s loss of €37m).
We note that AT&S’s tax rate has often been low effective (sometimes <10%) due to tax incentives for their China investments (tax holidays etc.). Going forward, as those expire and as profits shift to Malaysia (which likely gave them tax breaks to invest), the cash tax might remain modest. This improves cash earnings quality.
Thus, normalized earnings are positive and fairly high quality (backed by cash flow historically), except that recently reported earnings are depressed by transient factors. We will definitely rely on our normalized segment analysis rather than recent GAAP net income for valuation.
Leverage & Coverage:
Net Debt/EBITDA: At Mar 31, 2025 net debt/EBITDA was 2.5× (using the boosted EBITDA). On an adjusted basis (using €408m EBITDA), net debt ~€1,020m (because 2.5×408 ≈ 1,020). Pre-sale, net debt was around €1.6B and net debt/EBITDA over 6× as noted. So the sale drastically improved the ratio. But we must include hybrid bonds to get a full picture:
The hybrid (€350m) is accounted as equity likely, but it’s debt-like (perpetual with coupon). If we add that, effective debt was ~€1.37B at Mar 2025. Using adjusted EBITDA €408m, net debt+hybrid/EBITDA ≈ 3.4×. That’s still high but more manageable than 6×. The trend is: leverage peaked in mid FY24, and has now declined. They also scaled back capex (net capex halved from €855m to €415m in FY24/25).
Interest Coverage: Finance costs net were €-83m in FY24/25, up from €-50m prior. With adjusted EBITDA €408m, EBITDA/Interest ≈ 4.9×. EBIT/Interest would be lower: EBIT adjusted was €97m (since D&A huge), which doesn’t cover €83m interest by a big margin (EBIT/Interest ~1.2×). That’s a concern – under IFRS, they had barely enough operating profit to cover interest in FY23/24 (EBIT only €31m vs €50m interest – coverage <1, hence net loss). This will improve if the expansion yields profits, but in the near term, interest burden is heavy.
The average interest rate on debt likely rose with interest rate environment. If €83m interest on perhaps ~€1.3B gross debt (excl hybrid interest which is ~4.75% on 350m = €16.6m included in that maybe?), the effective interest cost ~6–7%. Some debt might be fixed/hedged, but the hybrid and some loans incur ~5%+. If rates climb further or if covenants tighten, interest could increase.
Interest coverage is a key vulnerability. We prefer to see at least 3–4× EBIT/Interest. AT&S will likely only achieve that once Kulim contributes significantly (maybe FY26 onward). Meanwhile, they rely on EBITDA coverage (which is okay for now). If a downturn hit EBITDA to, say, €250m and interest stays ~€80m, EBITDA/Interest ~3× (safe), but EBIT could be negative, which is not sustainable forever.
Hybrid Classification: The €350m hybrid is classified as equity (as it’s perpetual, subordinated, coupon deferrable). This flatters the equity ratio and net debt metrics. We, being conservative, treat it as debt in substance for valuation. That said, a hybrid gives management flexibility – they can defer coupons if needed (coupons cumulate typically, but not paying doesn’t trigger default). This is an emergency lever: should cash flow be very tight, they might skip a hybrid interest payment (though that would hurt credibility). So hybrid acts as an equity buffer in distress, which is positive for solvency. But from a shareholder’s perspective, it’s still an obligation that eventually needs servicing or refinancing.
Working Capital Health: AT&S’s business involves significant working capital tied up in accounts receivable (from big OEMs/EMS) and inventories (especially substrate WIP which has long cycle times). To manage this, they use factoring:
They had an international factoring program that in FY23/24 helped boost OCF to a high level. In FY24/25, they “reorganized” it, presumably meaning they might have reduced the amount of receivables sold, leading to a cash drain of ~€200m (pure speculation as OCF swing was €728m). Actually, from FY23/24 to FY24/25, OCF fell by ~€728m, out of which about €300m is loss of EBITDA gain (strip out Korea sale effect), rest could be working capital changes ~€400m. Possibly ~€200–300m of that was factoring adjustment. They mention the program “started in Q1 25/26” after reorg, so maybe they paused it at FY25 year-end and resumed with new terms after.
Factoring usage is not inherently bad, but it does mask underlying working capital needs when active. It can also incur costs (factoring fees, interest).
AT&S’s inventory might be high now due to ramp (they had to stock materials for new lines). We don’t have figures, but days inventory likely rose.
Overall, working capital relative to sales historically wasn’t too bad (in mid-teens percent). If factoring is fully stopped, AR would reappear on balance sheet and increase debt effectively. But they likely will continue factoring since they mention the program in 25/26.
We should be mindful: if customers delay payments or if AT&S can’t factor as much, cash could tighten. Current ratio etc. aren’t given, but presumably:
They had €485m cash at Mar 2025 after the plant sale (healthy buffer).
Unused credit lines €256m available.
Short-term debt obligations might include some bonds or loans due, we’d want to ensure no near-term wall. They have a 5% bond (maybe €200m) due 2027 (just guess from Frankfurt listing info). They likely staggered maturities.
Currency & Geopolitical Exposure:
AT&S reports in EUR, but earns revenue primarily in USD or USD-linked currencies. Customers like Apple, AMD, etc., often transact in USD. The geographical breakdown shows 75.7% of sales “Americas”, which likely means end-customers or invoicing in USD to their hubs. Meanwhile, a chunk of costs are in RMB (for China ops), MYR (Malaysian ringgit), and some in EUR (Austria HQ, some capex).
So there is FX mismatch risk. Historically, a stronger USD helped AT&S – indeed in FY23/24, they said currency fluctuations had a +€6.8m effect on earnings. If EUR strengthens against USD, their revenue (in EUR) would fall, hurting margins. They might hedge part of this exposure via forward contracts (common practice to hedge some portion of USD sales). But we haven’t seen specifics.
Another aspect: large capital expenditures in Malaysia might be in various currencies (equipment from Japan, Germany, etc.), but that’s mostly done. Now, AT&S will generate revenue in USD and incur local costs: Chinese costs in RMB (which loosely tracks USD somewhat), Malaysian costs in MYR (pegged partially to USD historically, but with some flexibility). If USD weakens significantly, AT&S’s EUR-reported earnings would shrink. Conversely, in the last year USD strength probably cushioned them.
China risk: AT&S has two plants in China (Chongqing substrate, Shanghai PCB). Tensions or tariffs could impact either supply lines or ability to serve certain customers. For example, U.S. tech sanctions on China primarily hit chips; but if Sino-U.S. relations worsen, some Western customers might avoid sourcing from China-located suppliers. AT&S did partly mitigate by building Kulim to diversify from China (they explicitly said it’s not wise to put all eggs in one basket in China). Now about half their substrate capacity is ex-China. But their China exposure is still material (Chongqing is fully China domestic, and Chinese market sales only 1.4% because presumably stuff made in China is exported or counted under the “Americas” if for U.S. clients). If extreme events (trade war, blockade) occur, AT&S could see production disruptions. On the other hand, being in China also gives them local Chinese business (perhaps with companies like Huawei or emerging Chinese chipmakers).
Malaysia/Asia: Political risk in Malaysia is low; however, reliance on imported skilled labor or expats for such a tech plant could be a factor (they need to train many new staff – they hired 1,500 by 2025 and plan up to 6,000 in Kulim).
Europe factor: The Leoben plant did attract EU funding. In case of a sharp downturn, AT&S might seek government support (loans or subsidies) citing strategic importance (especially with talk of EU wanting packaging capabilities onshore). That could be a cushion in worst-case scenarios (not something to bank on, but worth noting as soft support).
Valuation Currency: We do our valuation in EUR, which is AT&S’s reporting currency. No need to convert, since stock trades in EUR.
Downside Scenario Modeling: Let’s articulate a realistic bear case and see what it implies:
Scenario: A global recession in 2025–2026 plus slower yield ramp. Smartphone and consumer electronics remain sluggish (another 10% drop in mobile revenue). AI chip euphoria cools; data-center spending is flat, substrates suffer low utilization.
Under this scenario, assume FY25/26 group revenue stays ~€1.6B (no growth vs FY24/25) because PCB maybe -5% and substrate +5% only. EBITDA margins could be pressured: perhaps PCB margins drop to mid-teens% due to underutilization (maybe PCB EBITDA ~€150m), and substrate margins remain ~10% (substrate EBITDA ~€70m on maybe €700m revenue). Add maybe -€20m from “Others/corporate”. Total EBITDA ~€200m. With D&A €330m, EBIT is deeply negative (-€130m). Interest ~€80m, so pre-tax ~-€210m. They’d be incurring losses >€150m.
Cash flow impact: EBITDA €200m minus interest €80m minus some working capital (but in downturn, maybe working capital is released, +ve) minus maintenance capex (~€100m, because they might halt all expansion, only do maintenance). Free cash flow might be ~€20m (since working capital release could offset the op loss – they might scale back inventories, etc.). Perhaps roughly cash break-even before any debt paydown. However, if they still had committed capex to finish Kulim, that’s say €200m more – then FCF would be -€180m. That would force drawing credit lines or new debt. Net debt would rise again.
Leverage in this scenario: EBITDA €200m means net debt/EBITDA skyrockets. If net debt was ~€1.0B, ratio becomes 5×. If it rose to €1.2B with more debt, then 6×. Covenants likely breached (assuming covenants around 4×). AT&S might then do an emergency equity raise or seek covenant waivers (perhaps banks allow a one-time breach if the company has a credible plan and maybe if insiders pledge to backstop, etc.). The hybrid might allow skipping coupons, saving €17m cash per year (small help).
Dilution risk: In such dire scenario, raising equity say €300m at a depressed share price (imagine €10/share) would dilute existing holders ~30%. That’s something a value investor must consider – the margin of safety should ideally factor that possibility. The insiders might try to avoid that at almost any cost (they’d rather defer expansion or sell another asset perhaps). Are there other assets? They sold Korea, not much left to sell except maybe a partial stake in a plant or sale-leaseback of real estate. Government might be asked to help (especially for Austrian jobs in Leoben).
Covenant details: We don’t have exact covenant thresholds, but typical might be Net Debt/EBITDA < 3.5 or 4.0× (looser during expansion?). They might have negotiated waivers given expansion plan. If breached, lenders could demand paydown or higher interest, or it can trigger cross-default in bonds/hybrids if not cured. Given the proactive sale of Ansan to cut debt, management is aware of this risk.
In sum, the downside scenario is survivable but painful: AT&S would likely suspend all non-critical spending (stop Plant2 entirely, cut workforce further beyond 1,000, possibly suspend even hybrid coupon), and possibly require external infusion if the trough lasts long. The worst-case outcome for equity could be a combination of dilution and value diminution (in a fire-sale scenario, maybe the stock goes to single digits). However, with big insiders, any equity raise might be rights issue which they’d participate in – at least minority could avoid being diluted if they also participate.
From capital preservation view, our TGTP tries to price the stock such that even if something close to this scenario happens, one’s downside is limited. For instance, at €19(current price), the market might already price in a mild version of that scenario (since it’s down ~60% from peaks). At our TGTP €25, we think the scenario has low probability, making it an acceptable risk. But we will explicitly stress test that later in Section 5.3 when picking MoS.
Other considerations:
Factoring Reliance: The fact that AT&S had to reorganize factoring implies banks/ factors might have reduced exposure or changed terms. If the macro gets worse, they might further curtail factoring or increase its cost, thus AT&S could see working capital needing more cash. This risk is often underappreciated; we mention it to be thorough. Currently, they resumed factoring, so Q1 25/26 might show improved OCF as they sell receivables again. It’s a lever they can pull to generate cash quickly (at cost).
Interest Rate Risk: Most of AT&S’s debt might be fixed via bonds/hybrid. But any floating-rate loans (bank loans, credit lines) have become costlier with rising EURIBOR. That partly explains interest jump from €22m to €50m to €83m over last 2 years. If interest rates rise further or if they need to draw more credit, interest expense could increase. We saw interest coverage is thin; a couple more percentage points in rates could push finance costs over €100m, which would be > all FY24 operating profit. They likely hedged some interest or have fixed rates, but new debt or refinanced portions will be at higher rates in future. Another reason to have MoS.
Research Memo: Interest Rate Risk at AT&S
Rising net finance costs Net finance costs surged from –€18 m in FY 2022/23 to –€50 m in FY 2023/24, and further to –€83 m in FY 2024/25, driven primarily by higher interest charges.
Floating-rate exposure amid EURIBOR hikes While €350 m of hybrid bonds (5 % fixed until 2027) and Schuldschein issuance lock in part of the debt at fixed rates, bank loans, credit lines and lease financing remain tied to EURIBOR. As EURIBOR climbed, gross interest expense jumped from €39.2 m (H1 FY 22/23) to €64.4 m (H1 FY 23/24)—including €8.1 m of lease “margin step-ups”.
Thin interest coverage FY 2024/25 EBIT of €277 m versus €83 m net finance costs yields an interest coverage ratio ≈ 3.3×. A further 2 pp rise in EURIBOR could tack on ~€20 m p.a. of extra interest, pushing coverage toward < 3×, a level that risks credit-rating pressure.
Limited hedging disclosure AT&S’s reports do not detail significant interest-rate hedges beyond fixed-rate instruments. New debt or refinancings will be priced at current elevated yields, embedding further cost risk.
Conclusion of Risk Assessment: AT&S’s financial quality is in a transitionary state: historically strong cash generation from PCB is being overshadowed by high debt and slim coverage due to the new venture. The next 1–2 years are critical – if they successfully ramp profits, financial quality will improve rapidly (EBITDA rising, debt falling, coverage restored). If they stumble, financial distress is possible. The company’s actions (asset sale, cost cuts, dividend halt) show they’re actively managing risk. But from a conservative lens, until we see consistent free cash flow and debt reduction, AT&S cannot be viewed as low-risk. It has above-average financial risk, which we counterbalance by requiring a higher margin of safety in valuation.
This analysis underscores why our TGTP demands at least 35% safety margin: to ensure we’re protected even if something like the downside scenario unfolds (such that the effective price paid maybe already factors in partial dilution or extended weak earnings).
Next, we incorporate these assessments into our sum-of-parts valuation and MoS determination.
5. Valuation & Margin of Safety
We now bring together our segment valuations and overall financial adjustments to determine AT&S’s intrinsic value range, and from there derive a “Too-Good-to-Pass” (TGTP) buy price that includes a substantial margin of safety. We will use a Sum-of-the-Parts (SOTP) approach given the distinct dynamics of the two segments, then subtract net debt and consider any dilution or other claims.
5.1 Segment Valuation Recap
Electronics Solutions (Advanced PCBs) Segment: Using a conservative Owner Earnings multiple approach, we valued this segment at approximately €1.2 billion EV (based on ~€150m normalized OE and an 8× multiple) in Section 2.1. To cross-verify, we can also do a quick low-growth DCF: assume ~€150m annual free cash flow (post-maintenance) growing ~2% perpetually, discounted at ~9% (some risk but stable) – the value = 150/(0.09-0.02) = €2.14B. That seems high relative to our multiple approach. If we discount at 12%, value = 150/0.10 = €1.5B. Our chosen €1.2B is thus quite cautious (implies ~12.5% cost of capital and zero growth). Given industry challenges, that’s acceptable. We might stick with ~€1.2B for base case. We can consider a slightly higher number (like €1.3–1.4B) in an upside scenario where we allow a 10× multiple or modest growth, and a downside scenario maybe ~€1.0B if margins stay depressed. But the point is, our base case for PCBs is conservative.
For clarity, that €1.2B is enterprise value (before debt). If we later allocate debt, we’ll subtract at total company level.
Microelectronics (IC Substrates) Segment: Our conservative DCF-based valuation yielded roughly €0.93 billion EV for the substrate segment (Section 2.2). This assumed successful ramp to ~€1.2B rev and ~25% margins by FY27/28, but penalized near-term cash burn and used a high discount rate. In a bull case (e.g. including potential Plant 2 or higher margins), this could be much higher – e.g., if we gave it 8× a future EBITDA of €300m, that’s €2.4B. In a bear case (yield issues, only €800m rev at 18% margin), EV might drop to €600–700m. So the range is wide. We choose €1.4B as a base intrinsic value, with the understanding this already bakes in caution (it’s like pricing the segment at ~4.5× its potential EBITDA in 3 years or ~0.5× sales in 2027). This is arguably a margin-of-safety valuation in itself for substrates.
The Microelectronics (IC Substrates) Segment is arguably the hardest to valuate. My initial analysis lead to the following table:
I think 80% fab yield is the most likely outcome by 2028 (and conservative). Which would yield 130 OE. Now, to value this, we need the growth rate which is monstruous for this new segment as they are in a quite growing market together with getting a lot of the new business as customers dual source or diversify their supplier base. Some people say it will grow 12% CAGR, other 15% CAGR. Even with 10% graham’s formula would give 3.7B value in 2028, discounted with a hurdle of 12% that’s 2.6B EV as of 2025. Let’s just look at the bullish case, they get it to 90% and market grows 15%. Then 2028 EV would be 10.5B (0.274* (8,5+2*15)). The upside is stratosferic.
We will later stress test overall by possibly taking the low-end of one segment and base of other, etc., for scenario lows and highs.
Other and Corporate: AT&S has an “Others” segment (which might include some small trading or services, plus corporate costs). It’s not material in revenue (only a few million perhaps) but often shows a negative EBITDA (corporate overhead). We should account for corporate overhead by reducing combined segment value. However, in our segment OE we implicitly took into account some overhead (since segment EBITDA were given presumably before corporate). Actually, in AT&S’s segment reporting, they likely allocate most costs to segments; “Others” might have had -€19m EBITDA in FY22/23 or so (just guessing given difference between sum of segments and group). But we’ll be safe and subtract a lump-sum for the net cost of “Others” business. It might be minor enough to ignore, but for completeness:
The corporate OPEX (R&D not allocated, HQ salaries) could have a capitalized value. If corporate costs net of any “other” revenue are ~€20m per year, at say a 8× multiple (no growth, just cost), that’d be ~€160m negative value.
Alternatively, treat corporate expenses as already reducing the OE of segments in how we estimated them (we somewhat did, as we took net margin after everything for PCBs, etc.). Given potential double counting, we’ll simply keep a note that our segment values might slightly overstate if we haven’t removed corporate overhead. But since we were conservative in margins, it’s probably fine. If needed, we can knock off ~€0.1–0.2B from combined value to cover corporate drag.
Total Segment EV (pre debt): Adding base cases: €1.2B (PCBs) + €1.4B (substrates) = €2.6 billion enterprise value for the operating business. If we subtract a small overhead penalty, maybe ~€2.5B net.
Let’s retain €2.6B as the intrinsic enterprise value (IV) in our base scenario. We will also derive a range:
Low scenario: PCB €1.0B + Substrates €0.8B = €1.8B EV (reflects a situation where PCBs are valued at trough multiples and substrates underperform).
High scenario: PCB €1.5B + Substrates €2.4B = €3.9B EV (reflects optimal case with bull multiples and substrate success including expansion). The range is wide, but it frames the uncertainty. Our base ~€2.6B is roughly midpoint to lower half of that range, signifying caution.
5.2 Full Company Valuation (Sum-of-Parts & Adjustments)
Net debt in valuation context
In line with Buffett–Graham valuation logic, net debt is defined strictly as interest-bearing liabilities minus liquid financial assets. It is a necessary deduction when deriving equity value from an enterprise value (EV) calculated via sum-of-the-parts (SOTP) or DCF.
This approach deliberately excludes non-interest liabilities (e.g. trade payables, provisions) and subtracts only highly liquid assets that could be used immediately to extinguish debt.
Total net debt: € 1,491.4 m (as of 31 March 2025)
This figure is explicitly disclosed by AT&S and verified from balance-sheet and note disclosures. It includes:
Gross financial debt (€ 2,091.1 m) • Bank borrowings (incl. EIB, OeKB, IFC, etc.): € 1,524.2 m • Promissory-note loans (Schuldscheine): € 552.0 m • Export loans & bonds: € 35.0 m • Other financing liabilities (e.g. project-specific loans): € 105.3 m • IFRS 16 lease liabilities: € 343.1 m • Derivative financial liabilities (IR hedging): € 5.3 m
Less: liquid assets (€ 599.7 m) • Cash and cash equivalents: € 485.1 m • Other liquid financial assets (short-term securities, deposits): € 114.6 m
Buffett-aligned treatment
Leases included: IFRS 16 lease liabilities are treated as debt equivalents, consistent with Buffett's preference to capitalise operating leases.
No offset from receivables or inventories: Only truly liquid assets are netted; working capital is ignored.
No deduction for pension liabilities: Though unfunded pension obligations exist (€ 42 m), they are not deducted in base case due to size immateriality. Analysts may choose to subtract ~€ 32 m post-tax for ultra-conservatism.
Origin of debt burden
The net debt stems from capex-heavy strategic expansion—primarily IC-substrate facilities in Kulim (Malaysia) and Leoben (Austria). Promissory-note loans and export-credit instruments fund long-term industrial buildouts with maturities matched to asset lives.
Starting from the sum-of-the-parts (SOTP) enterprise value, we deduct non-equity claims—primarily net financial debt—to arrive at equity value, then per-share intrinsic value.
Net Debt Deduction
As of 31 March 2025, AT&S reported gross financial liabilities of €2,091.1 m, composed of:
€1,524.2 m in bank and institutional borrowings (incl. EIB, OeKB, KfW, IFC)
€552.0 m in promissory-note loans (Schuldscheindarlehen)
€343.1 m in IFRS 16 lease liabilities
€105.3 m in project-specific financing-partner liabilities
€35.0 m in public bonds and export/state loans
€5.3 m in derivative liabilities
Offsetting these liabilities:
€485.1 m in cash and cash equivalents
€114.6 m in other liquid financial assets (short-term deposits, marketable securities)
This yields a net debt of €1,491.4 m, fully consistent with Buffett-style definitions (i.e., excluding non-interest-bearing items, including lease obligations, and subtracting only truly liquid assets).
Other Adjustments
Pensions: Minor unfunded pension obligations (€42 m), immaterial in base case. Optional ultra-conservative analysts may deduct ~€32 m post-tax.
Stock-Based Compensation (SBC): Negligible. No active dilution observed; no significant option or RSU programs disclosed. Convertible authorization exists (€150 m), but unused.
Minorities: None. AT&S owns 100 % of all material subsidiaries.
Non-core Assets: No retained interests post-AT&S Korea divestment. No JVs or side-stakes of valuation relevance.
Capex Commitments: Already embedded in segment-level DCFs. No further subtraction required.
Deferred Considerations / Tax Assets: None material. Tax losses (~€37 m) are ignored for conservatism.
Valuation Summary
Based on our base-case sum-of-the-parts (SOTP) analysis, we assign an enterprise value of €2.6 billion to AT&S as of 31 March 2025. From this, we deduct €1.491 billion in net debt, calculated according to Buffett-style principles, which include lease liabilities and subtract only highly liquid assets.
This yields an implied equity value of approximately €1.109 billion. With 38.85 million shares outstanding, the resulting base-case intrinsic value per share is approximately €28.50.
5.3 Margin of Safety & TGTP Price
In keeping with the core tenets of value investing—particularly those of Seth Klarman and Benjamin Graham—we never buy a business simply because it’s "cheap." We buy it only when it’s undeniably cheap, with a wide margin of safety that protects capital even if things go wrong. In AT&S’s case, the level of operational and financial risk demands a particularly cautious approach.
From our valuation above, the base-case intrinsic value per share stands at approximately €28.50. To translate that into an actionable entry point, we must apply a discount that reflects the company’s risk profile.
At a 35% margin of safety, that would imply a “Too-Good-To-Pass” (TGTP) price of roughly €18.50. But the question is—is 35% enough?
Several factors argue for a steeper discount:
Execution risk remains high. AT&S is still navigating yield ramp-up, customer qualification hurdles, and aggressive scale-ups, all under heavy fixed-cost pressure. These could easily compress margins more than forecast.
Leverage is material. With net debt of nearly €1.5 billion, small variances in free cash flow or interest costs can translate into significant swings in equity value. High leverage compounds risk and warrants deeper protection.
Potential dilution isn’t priced in. If market conditions deteriorate and refinancing becomes difficult, the company might be forced to issue equity—an event we haven’t modeled, but one that would materially reduce shareholder value.
Volatility is a fact. AT&S has traded above €30 at its peak and below €14 during recent troughs. Markets clearly struggle to price this business consistently. A wider margin of safety increases the likelihood of buying near a true bottom.
Taking these factors into account, we believe a 40% margin of safety is more appropriate. That would suggest a TGTP price of approximately €17 per share.
More conservatively, a 50% margin of safety would imply €14.25—a level that would only make sense in a market panic or deeply pessimistic scenario.
Balancing prudence and opportunity, we set our preferred Too-Good-To-Pass price at €17, just below the 40% MoS threshold. At this level, the stock would trade at about 0.6× intrinsic value, offering a potential ~65% upside to our conservative base case—and even more under optimistic scenarios—while still preserving ample downside protection.
In summary:
Intrinsic Value (IV) ≈ €28.50
Chosen Margin of Safety: ~40%
TGTP Price: ~€17 per share
This is the price where the risk/reward skew becomes not just favorable—but compelling.
Part 2: Deeper Dive and Future Outlook
Revisiting Key Assumptions and Verification Points
Maintenance CapEx – Revised Estimate: In the initial analysis, a maintenance capital expenditure (capex) of 5% of revenue was assumed, which appears too low for this business. AT&S’s own guidance suggests a higher ongoing investment need: for example, in FY2021/22 management budgeted up to €100 million for basic maintenance and upgrades. That represented roughly 7–8% of revenue at the time. Given the capital-intensive nature of IC substrate manufacturing (cleanrooms, advanced equipment, etc.), a more prudent estimate is around 10% of revenue for maintenance capex. Using ~10% (instead of 5%) ensures we don’t overstate free cash flow. This adjustment will modestly reduce our FCF estimates and target valuation, but provides a more realistic margin of safety in the model.
Capacity Expansion & New Plant Ramp-Up: A key driver of future growth is the new Kulim, Malaysia IC substrate plant, which was constructed and ramped with remarkable speed. Management completed construction in just two years and achieved high-volume manufacturing one year later – an unusually fast timeline that underscores execution capability. As of May 2025, high-volume production has begun for AMD’s data center processors and other customers at Kulim. The plant already earned “Certified HVM (High-Volume Manufacturing) Site” status from AMD, indicating product qualifications and quality approvals are in place. This de-risks the revenue ramp significantly. Approximately US$1.2 billion has been invested in Kulim so far, and the first phase houses ~500 high-tech machines in 255,000 m² of floor space. The speed and scale of this project are reassuring – it was built on schedule (a potential world record in speed, per management) and positions AT&S to capitalize on surging demand.
Customer Wins and Quality Credentials: Another point from Part 1 was the significance of AT&S’s relationships with top-tier chipmakers. We can verify and take comfort in these credentials: AT&S received Intel’s Preferred Quality Supplier award (PQS) in 2020, an honor given to only 26 suppliers worldwide that year. Intel lauded AT&S for “standout service in a critical area of the supply chain” – a strong endorsement of AT&S’s quality and reliability. Likewise, AMD has effectively validated AT&S as a substrate supplier for its cutting-edge chips by certifying the Kulim plant for production. These achievements support the view that AT&S’s technology meets the highest standards, reducing the risk that the company’s expansion would outpace its ability to attract customers. In fact, AT&S management stated the goal of becoming one of the world’s top-3 IC substrate providers, and with both Intel and AMD in the fold, this is credible.
Market Pricing and Volatility Factors: Our deeper research confirms those “reassuring factors” are real positives – the AMD partnership is expected to grow “due to rising demand for CPUs and GPUs for datacenters, AI, VR and AR”, and new “high-profile clients” beyond AMD are slated to join this year. On the flip side, we must also acknowledge risks like market cyclicality and overcapacity: by Dec 2024 management cited “persistent weakness of the market coupled with overcapacities… and resulting price pressure” in both PCB and substrate segments. Indeed, a surge of industry-wide substrate investments (by AT&S and competitors) following the 2020–2022 boom led to a temporary glut as demand dipped in 2023. We will factor in price pressure and utilization risk in our scenarios to ensure the valuation isn’t overly optimistic on margins.
IC Substrates: End Markets and the Product’s Role
What End Markets Do IC Substrates Serve? IC substrates are a critical component in semiconductor packaging, essentially high-density mini circuit boards that sit between a silicon chip and the larger PCB. They are broadly of two types – ABF substrates (using Ajinomoto Build-up Film) for high-density, high-performance chips, and BT substrates (Bismaleimide-Triazine resin based) for lower-density applications. AT&S specializes in high-end substrates (mostly ABF) which are used in CPUs, GPUs, AI accelerators, networking chips, and advanced ASICs – the brains of devices in several end markets. Key end markets include:
Data Center and High-Performance Computing (HPC): This is currently the most exciting segment. ABF substrates are “dominating technology for application in high-performance computers” found in the heart of servers and data centers. Every AI server GPU (e.g. Nvidia, AMD accelerators) or cloud CPU (Intel Xeon, AMD EPYC) requires one or more large, multi-layer substrates. Surging demand for AI training and cloud computing is directly driving demand for these substrates.
Personal Computers and Consumer Electronics: PC processors (for desktops and laptops) also use ABF substrates, as do some high-end smartphone and tablet chips. In the 2020–2022 period, a pandemic-driven spike in PC demand led to substrate shortages; conversely, the 2023 PC downturn caused a substrate glut. 5G mobile devices and tablets contribute to substrate demand as well (often using thinner substrate-like PCBs or high-density interposers). Memory chips (like high-bandwidth memory used in AI systems) often use BT substrates, but as memory moves to advanced packages (e.g. HBM stacks on interposers), there’s overlap in tech requirements.
Automotive & Industrial Electronics: As vehicles adopt more advanced driver-assist and autonomous systems, they need powerful on-board chips (for vision processing, etc.) which may use substrates. AT&S notes that in the future “high-performance computers… in cars” will use ABF substrates. The automotive and industrial segments are smaller contributors today (automotive was only ~16% of AT&S revenue historically), but they are growth areas as chip content in cars and factory equipment increases. The new substrate competence center in Leoben is partly aimed at diversifying into automotive/aerospace applications (they achieved aerospace Gold supplier status recently).
Telecom Infrastructure (5G base stations) & Others: 5G base stations use powerful networking ASICs that also rely on high-density substrates. Additionally, emerging devices like AR/VR headsets (which pack high-performance chips in small form factors) are cited as an area of demand growth for substrates.
In essence, the “end-market of the end-market” for AT&S’s substrates is the digital economy itself. Their immediate customers are chipmakers (like Intel, AMD, Nvidia, Broadcom, Apple, etc.), but the ultimate end-markets are the data centers, communication networks, personal electronics, and smart vehicles that those chips go into. For example, AT&S substrates going into an AMD EPYC CPU ultimately serve the cloud computing and enterprise software users who rely on those servers. Substrates used in an automotive AI chip ultimately enable safer and smarter vehicles for drivers. The final needs being fulfilled are things like faster AI model training, streaming services, advanced mobile apps, and autonomous driving – all of which require powerful semiconductors, which in turn require advanced packaging solutions like AT&S’s substrates.
What “Job” Does the IC Substrate Perform? The IC substrate’s job is to enable the connection and communication between the tiny nanometer-scale world of the silicon chip and the larger world of the electronic system. It provides mechanical support and a fan-out of electrical signals from the chip’s microscopic pads to the much larger pitches of the circuit board. In fulfilling this role, substrates solve several critical needs for the end customer:
Electrical performance: High-end chips may have thousands of I/O connections. The substrate’s multi-layer interconnect design ensures signals and power are delivered with minimal loss and interference. Without an advanced substrate, a modern CPU/GPU could not achieve its performance due to signal integrity issues. As AI and HPC chips push power and I/O limits, substrates are evolving to more layers and finer line/space widths to accommodate this.
Form factor and integration: Substrates allow integration of multiple chips (chiplets) in one package (e.g., AMD’s chiplet architecture or Intel’s EMIB technology) by acting as a high-density interposer. They also mount high-bandwidth memory alongside processors in advanced packages. This is why trends like heterogeneous integration actually increase substrate complexity and content – packages like 2.5D (CoWoS, etc.) still require a large organic substrate as the carrier. AT&S management highlighted that server substrate demand will benefit from the shift to such multi-chip integration in the medium term.
Reliability: The substrate handles the thermal expansion differences between silicon and PCB and protects the chip from physical stress. High-quality substrates improve the overall reliability/lifespan of the device – a crucial need for end users (nobody wants their data center CPU failing due to packaging issues!). The Intel PQS award to AT&S is essentially Intel saying AT&S met extremely high reliability and quality standards in this role.
In summary, IC substrates are indispensable to fulfilling the end customer’s need for advanced computing power in a reliable form. Competing technologies like silicon interposers exist for very high-end uses (e.g., GPU-to-HBM connections), but those are used in addition to organic substrates, not in lieu of them (and glass substrates are a future prospect beyond 2030). So for the foreseeable future, any device that needs high computing performance will rely on state-of-the-art substrates – which is exactly the problem AT&S’s products solve.
Future Demand Drivers and Market Outlook
Having verified the current status, we turn to the future. A core part of our thesis is that the market is poised for strong growth, and AT&S is in position to ride that wave. Let’s examine the industry outlook:
Cyclical Bottom Passing – Return to Growth: 2023 was a tough year for the substrate market globally – total IC substrate market value dropped ~26% to ~$13.3 billion due to the sudden cooling of consumer electronics demand and digestion of excess inventory. Notably, ABF substrate sales fell to ~$7.16B (–26% YoY) as PC and general server demand slumped. However, industry forecasts show a sharp rebound from 2024 onward. As inventories normalize and AI-driven demand kicks into high gear, the global substrate market is expected to grow ~15% in 2024 to $15.3B. ABF substrates specifically are forecast to grow ~13.5% in 2024 to ~$8.12B, recovering a good portion of 2023’s dip. In fact, suppliers are openly optimistic that strong AI hardware demand in 2024–2025 will tighten the supply-demand balance again. By some estimates, the ABF substrate market could reach ~$11B by 2025 (approaching or exceeding the 2022 peak) and continue a high-single to double-digit CAGR through the decade.
AI and Cloud as a Growth Engine: The NVIDIA-led AI boom cannot be overstated in its impact on substrate demand. Training a single cutting-edge AI model can require thousands of GPUs – and every GPU module is built on one or more large substrates (for example, Nvidia’s high-end GPUs use sophisticated substrate packages to integrate the GPU die with HBM memory and power delivery). The ramp-up in AI server deployments by cloud providers (e.g. Microsoft, Google, Amazon) and others effectively means exponential growth in high-performance substrate demand in the next few years. Industry commentary indicates that “strong demand driven by AI” is expected to not only drive recovery but push substrate technology to larger sizes and more layers. One report noted that AI servers kept substrate makers’ capacity utilization high even during the 2023 slump, and as that AI penetration widens, it will absorb the capacity that was idle for PC chips. AT&S is already seeing this: its AMD partnership in Kulim is directly tied to data center CPU/GPU demand for AI, VR, AR. Additionally, new AI entrants (startups and cloud in-house chip projects) need substrate suppliers – AT&S’s strategy to diversify its customer base means it has won “additional renowned IC substrate customers in computing/data processing” beyond its historical ones. In short, AI and cloud are secular tailwinds likely to drive multi-year growth for AT&S.
Other Growth Areas: The memory market recovery in 2024 is another positive – as memory chip sales rebound (66% jump forecast in 2024 per Gartner), substrate demand for memory modules (largely BT-based) will rise. While AT&S is less exposed to memory substrates, a general upcycle lifts all boats and could improve pricing power in the substrate industry. 5G infrastructure and the next wave of smartphone upgrades (AI features in phones, etc.) will also contribute. Longer-term, trends like edge computing, IoT, and automotive AI mean new categories of devices requiring high-density substrates. Robotics—especially autonomous vehicles, warehouse bots, and emerging humanoid platforms—may also contribute meaningfully over time, as each new robot class embeds advanced compute packaged on ABF substrates. It’s worth noting that geopolitical initiatives in the US and China to localize semiconductor packaging might create new regional opportunities (and also new competitors) – e.g., the US CHIPS Act funding includes advanced packaging initiatives that treat substrates as strategic. AT&S having a foothold in Europe (Leoben) and now Malaysia/Asia puts it in a strong position to serve multiple regions as this landscape evolves.
Competitive and Supply Considerations: A quick check on competition: the top five players (Unimicron, SEMCO, Ibiden, AT&S, Nan Ya PCB) hold over half the global substrate market. AT&S is currently #4 with ~9.1% share, and it explicitly aims to climb into the top 3 by 2025+. The expansions by AT&S and peers (e.g., Unimicron, Nan Ya, etc. all added capacity in 2021–23) led to the temporary oversupply. However, with demand now rebounding, we expect utilization rates to improve across the industry. There remains a risk of localized gluts (for example, too much capacity for older PC substrate specs), but for leading-edge substrates (needed for AI/heterogeneous chips), the supply is still relatively tight – indeed, any delays in substrates can bottleneck chip deliveries (this happened to some CPU/GPU launches in 2021).
In summary, our deeper research reinforces the upside potential: the mid- and long-term outlook for IC substrate demand is robust, driven by powerful secular trends (AI, cloud, 5G, digitalization of everything). The 2023 hiccup appears to be a cyclical correction, not a structural decline. If anything, it gave AT&S breathing room to complete its expansion before the next wave hits. Now, with new capacity on line, AT&S can capture that growth.
Management’s Perspective and Strategy Through Time
We also examined how AT&S management’s communication and strategy have evolved, to gauge their credibility and whether they may be over- or under-stating prospects:
Early Bullish Targets (2021–2022): Management was very bullish when initiating the big expansion. In March 2021, as the Chongqing plant III was coming online, AT&S raised its medium-term guidance – expecting to exceed €2 billion revenue by FY2023/24 (a year earlier than initially planned) and aiming for a 25–30% EBITDA margin. They cited unabated demand for substrates and even launched additional expansions (adding €200M to fully equip more substrate capacity). The CEO’s stance: “We must be visionary... constantly working on technologies that will drive digitalization in the future”. This led to the massive €~2+ billion capex program (Kulim + Leoben). By Nov 2021, they forecast ~€3.0B revenue in 2025/26. Subsequently, during the 2022 boom, they briefly upped the target to €3.5B – reflecting confidence amid the chip shortage. For example, a press release in mid-2022 indicated AT&S foresaw €3.5B revenue in 2025/26 (raised from €3.0B).
Adjusting to Reality (2023–2024): When the market weakened in late 2022, management did temper their short-term optimism but maintained long-term conviction. In early 2023, AT&S acknowledged the “extreme increase” in pandemic-era demand had reversed with a PC/NB saturation, and they pushed their €3.5B target out by one year to 2026/27. They clearly communicated the reasons (war in Ukraine, inflation, inventory correction) and took action by trimming near-term capex plans (cut ~€450M from 2023/24–24/25 spending by delaying the second Kulim facility’s equipment until a customer demand warrants it). Importantly, even at that time, the long-term perspective was “unchanged” – “it does not change anything about the long-term perspective of our markets and positioning”, said then-CEO Gerstenmayer. He emphasized that they were using the downturn to diversify customers and would emerge stronger. This seems to have played out: they did add new substrate customers (the R&D center in Leoben was expanded into a small production line financed by new customers in computing). By May 2023, management still “confirmed” the medium-term guidance of €3.5B in 26/27 with 27–32% EBITDA margin, showing confidence despite the rough second half that year.
Recent Guidance and Tone (late 2024–2025): As the slump persisted a bit longer, AT&S took a conservative step in Dec 2024: they cut the 2026/27 outlook to €2.1–2.4B revenue and 24–28% EBITDA margin. This was a significant downward revision from €3.5B, reflecting the deeper/longer downturn and pricing pressure. The wording was frank about “overcapacities” and the need to intensify cost cuts. Essentially, management reset the base-case to a more prudent scenario given what they knew at end of 2024. It’s worth noting, however, that this revised forecast explicitly excludes any upside from the second Kulim plant (which is built but not yet equipped). In other words, €2.4B is what they expect with one Kulim plant + Leoben new line, assuming market remains only modestly growing. If AI demand or other factors call for bringing that second plant online, that could add hundreds of millions of revenue beyond 2026/27 – effectively an embedded call option on an industry upcycle.
New CEO and Strategy: We should mention that as of May 2025, a new CEO (Dr. Michael Mertin) took the helm, as long-time CEO Gerstenmayer stepped down. The strategic direction remains consistent – focus on high-end substrates and interconnect solutions – but we will watch if any change in tone occurs. Early signs are that the new leadership continues the transparent communication (e.g., giving only quarterly guidance for FY25/26 due to uncertainty, which shows discipline in not over-promising). They also reiterate aiming for “profitable growth and increase in company value” as the expansion projects complete. The management quality appears solid: they navigated the downturn by cutting costs, yet kept strategic investments in place, and were honest with the market about adjustments. This boosts our confidence that the bullish long-term projections have a realistic basis (they weren’t simply blind optimism – when conditions changed, they adjusted accordingly).
In conclusion, management was indeed very bullish during the boom – and those lofty goals had to be reined in – but importantly they did not waver on the thesis that digitalization and electrification trends (IoT, 5G, AI, etc.) will yield significant growth for substrates in the long run. With the worst of the downturn likely over, their earlier bullishness (e.g., 17% CAGR outlook) may start to look more realistic again, especially if AI-related demand accelerates. Our analysis will incorporate a range of scenarios to capture both the tempered guidance and the potential re-acceleration.
Future Financial Scenarios and Model Adjustments
With all the above in mind, we will now model the coming years’ financials under several scenarios to gauge the valuation range and ensure our thesis holds water under different outcomes. The key variables we will flex include revenue growth (volume ramp, pricing), EBITDA margins, capex intensity, and working capital needs. We will also correct any prior modeling errors (like the maintenance capex ratio) to refine our valuation. Here are the scenarios:
1. Base Case (Management-Guided Scenario): This scenario aligns closely with the current guidance (which we consider relatively conservative post-reset). We assume FY2025/26 will be roughly flat to modest growth (since Q1 25/26 is guided +15% YoY but the company is cautious on full-year outlook), and then a pickup in FY26/27 toward ~€2.3B revenue (midpoint of €2.1–2.4B). EBITDA margin in this base case we take around 26% by 26/27 (mid-range of 24–28% guide), up from ~25% adjusted in FY24/25. The improvement comes as Kulim ramps and start-up costs abate by then. Depreciation will be high (we project D&A peaking around €350M by 26/27, up from €328M in FY24/25, as the new assets are fully depreciated), so EBIT margins might be 10–12%. We’ll use 10% of revenue as maintenance capex in this scenario (€230M by 26/27) and assume that by 2026 the major expansion capex winds down, so total capex falls sharply after FY25/26. Working capital might normalize (some one-time hits like factoring program changes affected FY24/25 cash flow). The result: by FY26/27, AT&S could be generating solid positive free cash flow (FCF) again, as its growth capex drops while EBITDA grows. We expect net debt/EBITDA to comfortably fall <3x by then (maybe ~2x), and ROCE to improve (though likely still single-digit until volumes ramp more). Valuation: In this base case, if we assume ~€2.3B revenue, €600M EBITDA, ~€200M net profit in 2026/27, the stock at ~€18 (market cap ~€760M) is trading at very undemanding multiples (e.g. ~3.8x FY27e EBITDA, ~4× PE!). Even accounting for some execution risk, a reasonable fair value could be in the mid-€30s per share as the company exits this investment phase and the market starts pricing it on normalized earnings. We will refine the exact target after outlining other scenarios, but clearly the base case indicates substantial upside if AT&S simply meets its (already lowered) guidance in the next two years.
2. Bull Case (High-Demand/Upside Scenario): This scenario examines the potential if AI/cloud demand is even “crazier” than currently baked in – effectively a return closer to the earlier growth trajectory. Here we assume AT&S not only reaches the €2.4B upper end of guidance by FY26/27, but perhaps exceeds it by FY27/28 by activating the second Kulim plant earlier. If, for instance, Nvidia, Google, or others come knocking for more substrate capacity in 2025, AT&S could start installing equipment in Kulim Plant 2 (the shell is ready). That could add perhaps ~€500M+ annual capacity when fully ramped (since one plant was ~€1.2B investment for presumably ~€600-800M annual output at scale, very rough estimate). So our bull case might see revenues approaching €3.0B by FY27/28, effectively back to the original plan by a one-year delay. CAGR from now would be ~20%+. We also assume in this scenario that high utilization and a richer product mix (more high-end AI substrates) push EBITDA margins toward 30% (peak of original range). There could even be some pricing power if capacity gets tight for cutting-edge substrates in late 2025 (some industry observers don’t expect substrate prices to peak in 2024, implying room in 2025+ for increases). Maintenance capex would rise with revenue, but as a percentage might fall slightly if they gain efficiency (perhaps 8–10% still, we keep 10% to be safe). In this bull case, AT&S would be throwing off significant cash by 2027/28 – likely enabling rapid de-leveraging or even dividends/resumption of higher payouts. Valuation: Under these rosy conditions, the intrinsic value would be markedly higher. DCF analysis suggests a fair value possibly above €50 per share if we discount those higher cash flows back at a reasonable rate – meaning the stock could triple in a bull scenario over a 2–3 year horizon. Even on comparables: a €3B revenue, high-margin AT&S would deserve a multiple similar to peers (which historically trade ~1.5–2× PEG given growth, or EV/EBITDA 8–10×). That would yield a market cap well into the €2–3B range (vs ~€0.76B now). Of course, we wouldn’t set our official target price to bull case levels, but this shows the potential upside if things go right.
3. Bear Case (Downside/Prudent Scenario): We also must stress-test the downside to ensure capital is protected. In a bear case, perhaps the market remains soft longer or competition eats into opportunities. We assume maybe only €2.0–2.1B revenue by FY26/27 (the low end of guide), or even flat ~€1.6–1.8B if another recession hits, etc. Perhaps PC demand stays weak and AI uptake is slower than expected, leading to continued overcapacity. In such a scenario, margins might stay at ~20–22% EBITDA (due to pricing pressure and suboptimal utilization). AT&S could also face higher interest costs if rates stay elevated, and would be more leveraged with less cash inflow. We’d still assume they cover maintenance capex (~10% rev) and can service debt, but FCF would be minimal until the market improves. We also consider a risk that yields at the new plant could disappoint or a major customer might in-source or shift orders (though there’s no sign of this now – if anything, more customers are coming). In this conservative case, AT&S’s equity might languish as book value slowly grows but ROCE stays low. Valuation: The stock already trades at ~0.8× book, implying the market is somewhat pricing a mediocre outcome. Even if earnings are lackluster near-term, the tangible assets and tech know-how provide a margin of safety. In a downside scenario our DCF yields maybe ~€15–€18/share, suggesting limited further downside unless things really deteriorate. One could also take comfort in the fact that AT&S might itself become a takeover target or strategic partner in a weak scenario – it’s one of only a few advanced substrate makers in the world; a larger player or a state-backed fund (from Europe’s IPCEI program or elsewhere) could see value in its assets.
In short, the deeper research gave us more confidence in the quality of the earnings and cash flows (even if a bit delayed), which supports a high-conviction view that AT&S is undervalued.
Part 3: Final review
Basically what all of this tells us is that AT&S has a profitable mature business (with some growth) and a high growth business across China, Malaysia and Austria.
They made substantial investments in growth expecting to be making bank by 23/24 or slightly later but a supply glut of IC Substrate is leading to overcapacity, however it seems their investment are working out as expected and demand is coming back.
The issue is the valuation. While it’s easy to say the net debt is € 1,491 million and the mature business’s EV is about € 1,200 million, it’s much harder to value the IC substrate segment. Some parts of it (Chongquing) are producing, some are starting up (Leoben and Kulim) and some are idle (Kulim 2). Meaning they are ready for huge production, they just need the demand. Demand there is (AMD) but I expect to grow from the cyclicality (supply glut ending, cyclical restartà but also the general demand.
Let’s estimate that we value the business on just PCB-net debt we get an equity value of € 1,200 million - € 1,491 million = -€ 291 million.
The equity is around € 762 million as of my buy price (€19.63 - July 14, 2025).
This means the EV of the IC Substrate segment must be at least €1,053 million.
Let’s look into it.
1 | Strict‑minimum valuation (the “what if they only muddle through” case)
Assumptions
Yield stalls around 75–78 % for the next three years (below management guidance, comparable to first‐year yields at Chongqing).
Utilised revenue plateaus in the €1.0 billion area.
EBITDA margin hangs at 15 % (low end of peer history).
That spits out €180 million EBITDA, hence ~€1.1 billion EV. That is almost exactly the €1.053 billion the market is already crediting. In other words, €19.63 prices in nothing more than a barely cash‑break‑even substrate business. Capital preservation in this scenario comes from the PCB cash flow; you don’t lose money, but you don’t win either.
2 | Base‑case valuation (what seems probable given the evidence)
Assumptions
Yield climbs steadily to 85 % by FY‑27 (Kulim replicates late‑cycle Chongqing metrics).
Revenues reach the €1.2 billion midpoint of installed capacity.
EBITDA margin reaches ~27 % (line with Ibiden at similar yields).
Multiple rises a little to 8× once start‑up risk fades.
Now EBITDA is c. €320 million and EV rounds to €2.6 billion so, €2.3 billion total equity, ~€59 per share.
3 | Upside runway (high‑yield, no hiccup)
Assumptions
Yields press to 90 %,
Revenue nudges €1.3‑1.4 billion,
Margin hits 30–32 %,
Market is willing to pay 9–10× EBITDA for a now de‑risked AI infrastructure supplier.
EBITDA jumps beyond €420 million, EV breaks €3.9 billion, €3.6 billion total equity, ~€92 per share. That is the right‑tail of the simulation but still anchored in peer precedents (Shinko, Unimicron, Ibiden during tight cycles).
Take‑away
Even if you demand a Graham‑level haircut, the substrate division’s EV comfortably clears the €1.053 billion hurdle already embedded in the stock. Treating that as the “strict minimum,” every incremental tick of yield converts almost linearly into shareholder value because the debt is fixed and the fixed‐cost base is already paid for.
Hence the investment thesis in one sentence: at €20 you pay for the cash cow and get a high‑probability substrate call option for free; if that option merely executes to the mid‑80 % yield that peers have shown is achievable, intrinsic value triple; if it excels, it more than quadruples.
Bonus: Monte Carlo Analysis – Understanding the Probabilistic Payoff
I ran a Monte Carlo simulation to quantify the range of possible equity values stemming solely from the IC Substrate segment, using conservative ranges grounded in industry precedent. The setup was simple: for each scenario, we randomly selected values for four key drivers — yield, revenue utilization, EBITDA margin, and valuation multiple. These reflect how efficiently the plants run, how much capacity they fill, how much profit they retain, and what price the market might eventually assign once risk fades.
We ran 100,000 simulations. For each run, we calculated the EV of the IC Substrate segment and then subtracted €291 million (or €7.49 per share) — which is the effective “cost” of owning the mature PCB business and the company’s debt. The result gives us the equity value per share attributable to the IC substrate business alone.
What did we find?
The mean value was ~€51.33/share, meaning the most likely outcome — considering all ranges — delivers more than 2.5× upside from today’s €19–20 range.
Even the 10th percentile scenario lands around €27.56, implying the odds of losing capital are low, even if execution disappoints.
On the other hand, the right tail reaches over €90/share, reflecting the kind of upside that comes if AT&S hits near-perfect yields and margins.
The histogram shows a clear skew toward large gains. And crucially, we didn’t need blue-sky assumptions to get there — we just assumed AT&S performs like its peers (e.g. Ibiden, Shinko, Unimicron) with plants that are already built and already awarded by customers like AMD.
Conclusion: This is not just a value stock — it’s a mispriced call option on AI (and more) infrastructure, with its downside covered by a solid legacy business. If the substrate ramp merely delivers average execution, the share price should at least double. If they overdeliver, it’s a rare multi-bagger hiding in plain sight.
Thanks for reading! Subscribe for free to receive new posts and support my work.
Thanks for reading! This post is public so feel free to share it.
