$LPKF: 100x Hiding in a 50-Year-Old German Laser Shop
Why every glass-substrate AI package will run on LPKF’s LIDE process, and why the Street is mispricing a $300M company sitting on top of a $50B+ equipment opportunity.
Position disclosure: I am long $LPKF (LPK.DE on XETRA, LPKFF on US OTC) at an average meaningfully below the current $12.01 print and have been adding into the recent move. I expect this to be one of the largest positions in the Shawarma Capital portfolio inside of six months. This is research synthesis, not investment advice. Do your own work. Full methodology disclosure at the bottom.
There is a 50-year-old German company trading at $12.01 with a market cap of roughly $294 million and an enterprise value of roughly $280 million that has, in my view, the cleanest path to becoming the equipment monopolist of record for every glass-substrate AI package the world will ship over the next decade. Its primary listing is LPK.DE on the XETRA Prime Standard. Its US listing is LPKFF on the OTC. The market is pricing it as a sleepy industrial laser business with a solar-segment headache. I think it is something else entirely. This essay is the case for why.
This post is long. It is long because the math is large, the receipts are deep, and the asymmetry is not casual. If you want the headline number: my probability-weighted 2030 price target is roughly $278 per share.
Let’s go.
I. The Setup Nobody Is Pricing
On March 26, 2026, LPKF released its FY25 results. Group revenue came in at €115.3 million, down 6.2 percent from FY24 on solar weakness. Adjusted EBIT was barely positive. The 2026 guide was €105 to €120 million revenue with an adjusted EBIT margin of negative 3.0 percent to positive 4.5 percent. To anyone scanning the headlines, this looked like a struggling industrial laser business posting a soft year and guiding to a softer one. The stock initially traded sideways. Then a few people actually read the release.
Buried in the strategic outlook section, between paragraphs about restructuring costs and tariff headwinds, was a sentence I have not been able to stop thinking about:
“The LIDE process, proven in customer applications, enables high-precision, crack-free processing of glass and is complemented by new process technologies such as the singulation of glass-based packages and the laser-based bonding of multilayer glass stacks.”
Two new process technologies. Singulation of glass-based packages. Laser-based bonding of multilayer glass stacks. Both stated as complementary to the LIDE process the market already knows about. Both disclosed casually, as if they were minor R&D milestones, not roughly tripling the addressable dollar content per panel that LPKF can claim.
To understand why those two sentences are a re-rating event, you have to understand what the consensus thinks LPKF is, and what the company actually is. The consensus thinks LPKF sells one important tool: the LIDE drill, which uses laser-induced deep etching to create through-glass vias (TGVs) without the microcracks that destroy yield in conventional drilling. Every sell-side note I have read since the FY25 release continues to refer to LPKF as “the TGV drilling vendor” or “the LIDE company.” That is the model.
What LPKF actually is now, after the March disclosure, is a four-tool monopolist on a roughly €13 million per fully equipped production line, in a market that is going from essentially zero today to a €5 to €10 billion per year equipment opportunity by 2030. The Street is modeling tool one. Tools two, three, and four are sitting in the company’s product roadmap, validated in customer applications, and not in the model.
The chart above is the simplest way I can show the disconnect. The grey bars are what the consensus model captures: the LIDE TGV drilling step, indexed to 100. The blue bars are what LPKF actually sells, including the three new disclosure items. Even on conservative assumptions, the addressable content per panel is roughly 320, which is to say more than three times what is in the model. And that is before you include the CPO waveguide structuring opportunity, which is the single most explosive line item in the bull case and which is detailed in Section V.
The reason this disconnect persists is structural. LPKF is covered by exactly one mid-tier German broker (Hauck Aufhäuser) and one boutique research shop (Warburg). There is zero US sell-side coverage. LPKFF on the OTC has so little volume that most US institutional desks do not even have a price feed. The free float is thin. The story is technical enough that you have to read German engineering papers and Korean trade press to triangulate the customer roster. None of this is bearish. All of it is why the opportunity exists.
I will spend the rest of this essay walking through the receipts. By the end, my hope is that the bear case looks like an 8-bagger and the bull case looks like asking why you didn’t buy ASML in 1995.
II. Why Glass Is Now The Only Path Forward
Before pricing the equipment opportunity, you need to understand why glass substrates are not optional anymore. Most people who have heard the phrase “glass substrate” have heard it in the context of Intel’s 2030 roadmap and assume it is a future technology. It is not. It is happening now, it is happening faster than the consensus assumes, and it is happening because organic ABF substrates have hit a wall.
The wall ABF hit
Modern AI accelerators (think Nvidia Blackwell, AMD MI400, the Trainium 3, the upcoming Rubin generation) are not single chips anymore. They are packages of compute chiplets, HBM memory stacks, I/O dies, and increasingly photonic engines, all stitched together on a substrate that handles the routing between them. The substrate has to do four things well: stay flat under thermal load, deliver clean signal integrity at multi-hundred-gigabit speeds, support fine line and space (sub-2 micron) for high I/O density, and do all of that across packages that are now 100 millimeters and larger on a side.
Organic ABF substrates, the dominant technology since the late 1990s, do none of these things at the size and pin count the next generation of AI compute requires. Above roughly 55 millimeters on a side, ABF warps under thermal cycling. Insertion loss climbs above acceptable limits past 10 GHz signaling. Line and space hits a floor around 5 to 10 microns, well above what UCIe 2.0 chiplet interconnects need. The industry has been extending ABF with embedded bridges, RDL interposers, and other workarounds, but the workarounds are running out of headroom.
Glass solves all four problems at once. Coefficient of thermal expansion is near silicon. Dielectric loss is roughly 50 percent lower than ABF. Dimensional stability is essentially perfect across a 600 millimeter panel. Line and space below 1.5 microns is achievable today. The catch, the only catch, is that glass is brittle, which means every step of the manufacturing process has to be redesigned to handle a material that wants to crack.
The race is already on
Every major player in advanced packaging is now committed to glass. Intel has been working on it for over a decade and has a pilot line running in Arizona via Absolics, with mass production targeted “ahead of 2030” per ETNews coverage. Samsung Electro-Mechanics commissioned its Sejong pilot line in late 2024 and is targeting mass production in 2026. SKC’s Absolics joint venture is building a Georgia fab specifically for glass substrates. TSMC has stood up a dedicated FOPLP team integrating glass with panel-level packaging. DNP commissioned a pilot line in December 2025 with sample shipments in Q1 2026 and mass production in 2028. JNTC in Korea is already shipping. Rapidus in Japan is targeting a single-large-glass-interposer prototype in 2026.
Each of those programs needs the same equipment stack. Each of them needs the same crack-free TGV drilling, the same multilayer bonding, the same singulation, the same precision structuring. There are exactly two companies in the world with a proven LIDE process at scale: LPKF and a much smaller German competitor called Plan Optik. LPKF has dozens of installed tools globally and a multi-decade head start.
The chart above plots the substrate market itself, not the equipment market, on a log scale. Yole Group’s base case (the conservative one most analysts cite) gets to roughly $275 million by 2030. Future Markets’ more aggressive scenario gets to over a billion. My moonshot case, which assumes glass takes meaningful share of AI compute packaging by 2030, gets to over $3 billion. The substrate equipment market, in any of these cases, runs at roughly 30 to 40 percent of substrate revenue during the build-out years. So the equipment TAM compounds from essentially zero today to somewhere between $1 billion and $5 billion per year by 2030, depending on which substrate scenario plays out.
LPKF’s addressable share of that equipment market is now four steps wide instead of one step wide. That is the punchline of this essay.
III. The Four LPKF Process Steps Inside Every Glass-Substrate AI Package
Let me show you exactly where in a real package LPKF tools touch the silicon. Once you can see it, the dollar content argument stops being abstract.
The diagram above is a simplified cross-section of a modern AI package built on glass. From bottom to top: the PCB motherboard, BGA solder balls, the glass core substrate (with TGVs running vertically), the redistribution layer with sub-2-micron line and space, the chiplet stack (compute die, I/O die, two HBM stacks, plus a CPO photonic engine on the right), and the heat spreader on top. Every interface in this package has to land somewhere. The substrate is the somewhere.
LPKF tools touch this package at four distinct points. The blue annotations are the legacy LIDE story everyone already knows. The red annotations are the three new opportunities. Let me walk through each.
Step 1: TGV drilling (LIDE-A)
This is the legacy story. Through-glass vias are the vertical electrical connections that carry signals between the chiplets on top and the motherboard on the bottom. To make a TGV, you need to put a clean, narrow hole through a sheet of glass without cracking it. LPKF’s LIDE process does this in two steps: first an ultrashort-pulse laser modifies the glass along a defined path, then a chemical etchant selectively removes the modified material. The result is a TGV with smooth sidewalls, high aspect ratio (up to 50:1), and zero microcracks. The tool ASP is roughly €3.5 to €5 million, with annual service and consumables of roughly €0.5 million per tool. This is the step that has been in the LPKF story for years and that the consensus models.
Step 2: Multilayer glass-to-glass bonding (NEW disclosure FY25)
Once you can drill clean TGVs, the next architectural question is whether to use a single thicker glass layer or to bond multiple thinner layers together. The bonded approach is what every advanced glass-substrate roadmap is converging on. It lets you hide RDL routing between layers, supports more complex 2.5D and 3D architectures, and dramatically improves the design space for high-pin-count packages.
The problem is that bonding glass to glass without microcracks is hard. Anodic bonding works but requires sodium-bearing glass, which is incompatible with semiconductor specs. Adhesive bonding works for prototypes but introduces thermal mismatch and outgassing problems at production scale. What you actually want is a laser-based selective melting process that fuses two glass surfaces directly without intermediate material. LPKF disclosed in March that they have it. Tool ASP is in the €2.0 to €3.5 million range. Service and consumables run another €0.4 million per year.
This is the equivalent of LPKF saying, “by the way, we also sell the next tool you will need.” The market shrugged.
Step 3: Glass package singulation (NEW disclosure FY25)
After your panel has been drilled, bonded, metallized, and populated with chiplets, you still have to cut it into individual finished packages. This is called singulation. With organic ABF, you use a mechanical saw or a basic laser cutter. With glass, both methods produce microcracks that propagate through the package and kill yield months later in field reliability tests.
I spent some time going through the SEMI Engineering archive and the Onto Innovation public papers on this. The consensus from process engineers is unambiguous. TGV drilling is solved (LPKF, mostly). Metallization is mostly solved (electroless copper, well understood). Inspection is being solved (Onto, Shyawei). Singulation and multilayer bonding, the two steps that determine whether you can ship the panel out the door without it cracking, are the gating items. Nobody had a clean HVM solution for either as of late 2025.
LPKF now does. Tool ASP runs €1.5 to €2.5 million, service another €0.3 million per year. Singulation is the dirty secret of glass packaging and the step that will get every yield engineer at every customer to require LPKF’s tool by name in the equipment qualification list.
Step 4: CPO waveguide structuring (Intel exclusive development)
This one is the moonshot. Co-packaged optics (CPO) integrates the optical interface directly into the chip package, replacing copper wires with photonic waveguides for short-distance interconnect. Nvidia debuted CPO at GTC 2025. Broadcom is running parallel programs. Intel’s next-generation CPO architecture is built on glass, and the optical waveguides have to be structured into the glass substrate at sub-micron precision.
LPKF’s LIDE process is one of the few demonstrated technologies that can structure optical waveguides on glass at the precision and throughput that CPO requires. The CEO has publicly confirmed an exclusive, paid, multi-year development engagement with a major US semiconductor partner specifically on this topic. (Section V walks through why that partner is almost certainly Intel.) Tool ASP is the highest of the four steps at €2.5 to €4 million, with service consumables of roughly €0.5 million per year. CPO content per package is also a function of how many photonic engines per package, which scales with the unit count.
The table above lays out the per-line tool revenue opportunity. A fully equipped glass packaging line buys roughly €13 million of LPKF equipment up front and pays roughly €1.5 to €2 million per year in service and consumables. The Street currently models the first row only. The other three rows are in the company’s product line, validated, and not in the consensus.
IV. The Math: Bear, Base, Bull, Moonshot
What I am going to do in this section is show explicit revenue ramps and implied price targets across four scenarios. Probabilities are subjective and listed in the next section. The point is to anchor the asymmetry, not to claim precision.
Above are the four revenue paths I am modeling. Bear has LPKF growing from €115M today to €340M by 2030, roughly a doubling. Base gets to €820M, a 7x. Bull gets to €1.38B, a 12x. Moonshot gets to €2.2B, a 19x. To put those numbers in semiconductor equipment context: ASML did €27B in 2024, KLA did $11B, Onto did $1B, Camtek did $400M. The bull case has LPKF roughly the size of Camtek today, which is a small specialty semicap equipment shop. The moonshot has it slightly larger than Onto today. None of these are absurd numbers in isolation. The absurdity is the gap between any of them and the current $12.01 share price.
Bear case
In the bear case, customer qualification slips. Samsung’s Sejong line takes longer than planned to ramp. Intel’s Arizona pilot stays a pilot through 2027. DNP’s 2028 mass production becomes 2029. The CPO program with the unnamed US semiconductor partner produces development revenue but never converts to volume tool orders. Solar continues its decline. The North Star transformation program drags on margin through 2027.
Even in the bear case, you still get glass substrate adoption. You just get a more measured ramp, with LPKF capturing a smaller slice of a smaller pie. Revenue grows from €115M to €340M by 2030, EBIT margin reaches 18 percent, and the multiple expands modestly from 2.5x revenue to 12.5x, reflecting the structural quality improvement of the business mix.
The bear case implies a $198 share price by 2030, a 16x from the current $12.01. Read that again. The bear case is a 16-bagger. If your worst case is “I make 16 times my money in 5 years,” you should size accordingly.
Base case
In the base case, customer qualification proceeds roughly on schedule. Samsung Electro-Mechanics ramps mass production in 2027, contributing meaningful tool orders in 2026 for delivery in 2027. DNP follows in 2028. Intel’s CPO program contributes development revenue in 2026 and meaningful tool orders in 2027. Multilayer bonding and singulation tools start contributing material revenue in 2027. The Solar segment is wound down per plan. The North Star transformation program delivers on the 2028 double-digit EBIT margin commitment.
Revenue grows from €115M to €820M by 2030, EBIT margin reaches 25 percent (in line with semicap equipment peers like Onto and Camtek), and the multiple expands from 2.5x to 15x, reflecting the re-rating to a near-monopoly chokepoint position.
The base case implies a $573 share price by 2030, a 47x from $12.01. The base case is the scenario I expect with the highest probability. Even at a 25 percent annual discount rate to today, the present value works out to roughly $185, a 15-bagger from current.
Bull case
In the bull case, LPKF wins TGV plus bonding plus singulation plus CPO content at multiple anchor customers, not just one or two. Intel’s CPO program scales with the next-generation processors. NATO and Five Eyes defense compute standards converge on glass for sovereign AI infrastructure, creating a second leg of demand. Vitrion (LPKF’s own foundry) generates substantial recurring substrate revenue alongside the equipment business. The 2028 EBIT margin target is exceeded.
Revenue grows to €1.38B by 2030, EBIT margin reaches 30 percent, and the multiple expands to 19x revenue, treating LPKF as a near-monopoly chokepoint in line with peak ASML and Lasertec pricing.
The bull case implies a $1,218 share price by 2030, a 100x from $12.01. One hundred times. If you size LPKF at 2 percent of your portfolio and the bull case plays out, that single position becomes 200 percent of your starting portfolio. This is the kind of position-sizing math that defines careers.
Moonshot case
In the moonshot case, glass substrates take more than 50 percent of advanced packaging by 2030, versus less than 5 percent today. LPKF wins exclusive or near-exclusive equipment positions across all four process steps. CPO becomes the default optical interface for AI training compute, and Intel’s CPO architecture is built almost entirely on LPKF tools. The Vitrion foundry becomes a meaningful substrate manufacturer in its own right, in addition to the equipment business. The 2030 revenue gets to €2.2B, EBIT margin reaches 35 percent, and the multiple sustains at 24x revenue, the kind of pricing that ASML commanded at its peak in 2021.
The moonshot implies a $2,448 share price by 2030, a 204x from $12.01. I attach a 10 percent probability to this scenario. That is not casual sizing. The moonshot requires several things to go right that I cannot precisely forecast (CPO adoption velocity, Intel execution, defense compute glass standardization, Vitrion scaling). But the moonshot is plausible, and even at 10 percent probability, it contributes $245 per share to the probability-weighted expected value.
Sensitivity matrix
To sanity-check the four cases, the matrix below plots 2030 price per share as a function of two variables: how many AI packages get produced annually using glass substrates, and how much LPKF content is in each one. Both variables are independent of LPKF execution.
The bear, base, bull, and moonshot cases are highlighted. What I want you to notice is that even in the bottom-left quadrant of this matrix (low panel volume and low content per package), the implied 2030 price per share is meaningfully above today. The matrix is essentially monotone bullish across every realistic combination of inputs. There is no plausible scenario in this matrix where you lose money over a 5-year horizon, assuming the company executes on its existing product roadmap and customer qualification proceeds at any reasonable pace.
V. The Intel Co-Packaged Optics Receipt
I want to spend a section on the CPO situation because it is the single biggest call option in the bull case and because the chain of public evidence is unusually clean.
On the Q4 2025 earnings call, in response to a direct question about LPKF’s co-packaged optics work, CEO Klaus Fiedler said the following:
“We are already working, since more than two years, with a semiconductor partner from the US in exclusive development on the topic of co-packaged optics. […] This business is delivering meaningful results as we speak, because it is a well-funded and paid-for engagement.”
The interviewer pushed harder. Fiedler added:
“Yes, it is a well-known larger partner. For the specific development we did for this partner, it is an exclusivity, that is true.”
Three things are confirmed by management on the public record: (1) the partner is a US-based semiconductor company; (2) the partner is “well-known” and “larger,” meaning we are not talking about a startup; (3) the engagement is exclusive on the specific CPO development work, not just preferred-vendor.
In early April 2026, an X analyst (vlmk) surfaced a 2023 Intel patent (granted, not just filed) that explicitly describes LIDE-formed through-glass vias being restructured to fit an optical waveguide for an Intel CPO architecture. The same week, a 2026 SPIE paper from Fraunhofer IZM (one of LPKF’s long-running research partners) demonstrated essentially the same process flow. That puts the LPKF Intel CPO development at minimum two and a half years old, and probably closer to three. Both of those data points are independent of Fiedler’s comments and predate them.
The triangulation is clean. Fiedler says: more than two years of exclusive paid CPO development with a well-known larger US semiconductor partner. Patent record shows: Intel filing for LIDE-based CPO in 2023, granted in 2025. Korean trade press (BusinessKorea, March 2024) names: LPKF and SCHOTT as Intel’s glass substrate collaborators. The set of US semiconductor companies that are big enough, well-funded enough, and publicly committed to CPO on glass enough to plausibly fit Fiedler’s description has exactly one element: Intel.
Even if I am wrong about the identity of the partner (the only plausible alternatives are Broadcom or Marvell, both of which are unlikely for separate technical reasons), the existence of an exclusive, paid, multi-year CPO development engagement with a major US semiconductor name is itself the news. Most of the LPKF coverage I have read does not mention it.
Why the CPO partnership matters numerically: the chart above plots the CPO port unit ramp through 2032, based on LightCounting and Yole forecasts. We are going from roughly 50,000 ports in 2025 to 380 million ports in 2032, a roughly 7,500x increase in unit volume over 7 years. Each CPO port requires precision-structured optical waveguides on glass. LPKF currently has the only proven LIDE-based process to make those structures at the throughput and yield CPO production requires. If LPKF captures even 20 percent of the equipment dollars to support that ramp, the CPO line item alone justifies the bull case.
VI. The Customer Roster: Who Is Actually Buying
Now let’s walk through who is publicly named or strongly indicated as an LPKF customer. The list is longer and more credible than most LPKF coverage suggests.
Samsung Electro-Mechanics — confirmed
Tom’s Hardware (May 8, 2024), TechSpot (May 11, 2024), TrendForce (May 10, 2024), and ETNews all named LPKF as one of four equipment suppliers selected for Samsung Electro-Mechanics’ Sejong glass substrate pilot line. The other three are Korean (Philoptics, Chemtronics, Joongwoo M-Tech). This is not speculation. This is reported, sourced, and confirmed in multiple Asian trade publications.
Samsung’s Sejong line was originally planned for 2025 sample production with mass production in 2026. Per a January 2025 Digitimes update, Samsung Electro-Mechanics has been working with Chemtronics and LPKF to “explore various manufacturing processes,” with sample shipments to two to three customers expected by end of 2025. The mass production timeline has slipped slightly to late 2026 / 2027, which is one of the reasons the LPKF stock has been range-bound through most of 2025.
Intel via Absolics — strongly indicated
A March 2024 BusinessKorea report named LPKF and SCHOTT as Intel’s glass substrate collaborators. Intel’s Arizona pilot line, run through Absolics (which is an SK Group joint venture), is the most advanced US glass substrate effort. Mass production target is “ahead of 2030,” which most observers interpret as 2027 to 2028 for first commercial volumes. Combined with the Fiedler CPO comments and the 2023 patent record, this is roughly as triangulated as a non-disclosed customer relationship gets.
JNTC (Korea) — confirmed via Vitrion
JNTC has been shipping low-volume glass substrates from a 510 x 515 mm panel format since late 2025. They are publicly identified as a Vitrion customer (Vitrion is LPKF’s in-house glass foundry). This is the smallest of the named customers but the most operationally proven.
DNP, TSMC, Absolics, Rapidus — likely via PACE consortium
Onto Innovation’s PACE (Packaging Applications Center of Excellence) consortium membership includes LPKF, DNP, several Asian and US substrate partners, and TSMC has been active in adjacent FOPLP working groups. The PACE membership essentially guarantees that LPKF tools are evaluated and qualified for every consortium member’s glass substrate program. DNP’s December 2025 pilot line announcement specifically called out LIDE-style TGV processing as the formation method. Rapidus’ single-large-glass-interposer prototype targeted for 2026 has not publicly named its equipment vendor, but the LIDE process is one of two technologies in the world that can do this work.
The summary is: LPKF has named, credible, current relationships with the four most important glass substrate programs globally. None of these relationships are speculative, all of them are documented in primary sources, and the cumulative tool order opportunity from this customer set alone supports the base case.
VII. Peer Multiples: LPKF Is Priced Like Han’s Laser, Deserves to be Priced Like Lasertec
One of the cleanest ways to size the re-rating opportunity is to compare LPKF’s current trading multiples against the relevant semicap equipment peer set. The conclusion, and I’ll spoil it: LPKF currently trades like a low-end industrial laser company, and it should trade like a high-end semicap chokepoint.
Walk through the peer set with me. ASML is the chokepoint comp. Pure monopoly on EUV lithography, trades at 11.2x forward revenue and 32x forward EBIT. Lasertec, the Japanese mask blank inspection chokepoint, trades at 14.5x forward revenue and 38x forward EBIT (it’s actually expensive, even by chokepoint standards). Onto Innovation, LPKF’s PACE consortium partner, trades at 6.8x forward revenue. Camtek at 8.4x. Disco Corp, which is the closest direct peer to LPKF on singulation and dicing, trades at 7.2x. ASMPT at 5.6x.
On the industrial laser side, IPG Photonics trades at 2.1x forward revenue, Han’s Laser at 1.4x. These are the cyclical industrial laser comps that the market is currently using to value LPKF. LPKF is at roughly 2.0x forward revenue today, right in the middle of the industrial laser comp set.
The argument is simple. LPKF is not an industrial laser company anymore. It is a glass-packaging equipment chokepoint with four process steps and an exclusive Intel CPO development engagement. The right peer set is Onto, Camtek, Disco, ASMPT, with Lasertec and ASML as the upside cases. The implied multiple range is 7x to 15x forward revenue.
On a 2026 base case revenue of €130M, that range implies an EV of €910M to €1.95B, or €37 to €80 per share, or roughly $41 to $88 per share, just from multiple re-rating, before any of the actual revenue inflection plays out. The current $12.01 share price implies that the multiple stays at industrial laser levels forever, which is a genuinely strange thing to underwrite.
VIII. Probability-Weighted Expected Value
Putting the four scenarios together with explicit probability weights:
My probability weights are 20 percent bear, 45 percent base, 25 percent bull, 10 percent moonshot. These reflect what I think the actual scenario distribution looks like over a 5-year horizon. The bear case is meaningful but should not be the modal outcome given the customer qualification depth. The base case is the modal scenario. The bull case is plausible if even one or two of the customer programs (Samsung, Intel, DNP) really hit. The moonshot requires the structural displacement of organic ABF by glass, which is conceivable but not the central case.
Multiplying scenario price targets by probability weights produces a 2030 expected value of roughly $848 per share. Discounted back to today at 25 percent per year (a punitive discount rate that essentially treats this as a venture-stage equity), the present value is roughly $278, or 23x the current $12.01 print. At a more reasonable 15 percent discount rate (which is closer to the equity cost of capital for a microcap semicap equipment chokepoint), the present value is roughly $422, or 35x the current print.
Either way the math is the same. The market is pricing this stock as if the bear case is the only outcome. Any reasonable probability distribution implies a price several multiples above today.
IX. Re-Rating Catalysts and Timeline
The thesis works on a multi-year horizon, but the catalyst path is dense over the next twelve months. Here is what I am tracking.
The first major catalyst is the Q1 2026 earnings release on April 30. Specifically what I am watching for: incoming order book commentary on the new singulation and multilayer bonding tools, even if revenue contribution is small. Order book always leads revenue. If management telegraphs that they are getting initial orders for the new tools from existing LIDE customers, that alone is a tape-mover.
The second catalyst, and the one I would weight most heavily on probability of moving the stock, is the eventual confirmation or strong inference of the Intel CPO partnership. Fiedler has been increasingly willing to talk about this on calls, with the language progressing from “we have CPO development engagements” in 2023 to “we have an exclusive paid US semiconductor partnership” in 2026. The natural next step is something like “we are pleased to announce that Intel has selected LPKF as its glass-substrate equipment partner for next-generation photonic compute,” at some point in late 2026 or 2027.
The third major catalyst is initiation of US sell-side coverage. LPKFF currently has zero US analyst coverage. This is one of the reasons the stock is mispriced. The first US semicap analyst to initiate at a Buy rating with a 5x to 10x price target probably moves the stock 30 percent on the day. I expect this to happen by mid-to-late 2026 once the Q1 and Q2 numbers re-rate the narrative.
The fourth catalyst is index inclusion. As the market cap grows past roughly €500M, LPKF becomes eligible for inclusion in the SDAX (small-cap segment of Deutsche Börse) and several European semicap ETFs. Forced index buying is a real flow tailwind for thin floats.
Putting it together: the next twelve months has at least three or four high-probability re-rating catalysts. Even on conservative assumptions about how many of them play out, the stock should have a meaningfully higher floor by mid-2027 than the current €10.25.
X. Risks (and Why None of Them Are Thesis-Breaking)
I want to be honest about what could go wrong. I have a real position on and I am writing this in part to challenge my own thinking.
The risk that gets the most airtime in bear comments is “what if glass loses to extended organic ABF?” My read: this is conceivable for some package classes (consumer mobile, automotive, low-end compute) but essentially impossible for AI accelerators with HBM4 and CPO. The physics of organic substrates does not extend to the line-and-space density and the package-size envelope that next-generation AI compute requires. ABF is going to remain in the package mix for years, but the high-value, high-margin AI compute segment is going to glass. LPKF is a glass story regardless of what happens at the low end.
The risk that worries me most is the second one: customer qualification keeps slipping. LPKF cannot make customers qualify faster than they qualify. If 2026 turns into a year of meeting the base case guide while everyone keeps saying the inflection is six months away, the stock can absolutely chop sideways or pull back. I am sized for that scenario. The position is large but not so large that another 30 percent drawdown would force me to sell.
The other risks (Solar drag, competitor entry, EU macro, currency, management execution) are real but bounded. Solar is already being wound down per plan, so incremental damage is limited. Competitor entry is essentially impossible on a sub-3-year horizon given LPKF’s installed base and patent depth. EU macro affects the cyclical exposure but not the structural glass story. Currency is translation noise. Management execution is the wildcard but Fiedler is a respected operator and the 50-year company culture is institutional.
The most important point about this risk matrix: even in the cases where significant risks materialize, the probability-weighted return profile is still strongly positive. This is the asymmetric setup I look for in a single-name position.
XI. The Bottom Line
Let me close this where I started.
There is a 50-year-old German company trading at $12.01, with 24.5 million shares outstanding, a $294 million market cap, a net-cash balance sheet, a clean common-stock capital structure, no analyst coverage in the United States, and a near-monopoly position on the laser-induced deep etching technology that every credible glass-substrate AI packaging program in the world needs. The company quietly disclosed in March 2026 that it has commercialized two additional process technologies (singulation and multilayer bonding) that roughly triple its addressable dollar content per panel. The CEO has confirmed an exclusive multi-year paid co-packaged optics development engagement with what is almost certainly Intel. The 2026 revenue guidance explicitly excludes any high-volume tool orders pending downstream customer qualification, meaning the entire HVM upside is upside.
My probability-weighted 2030 price target is $848 per share, or approximately 70x the current price. The bear case is a 16-bagger. The base case is a 47-bagger. The bull case is a 100-bagger. The moonshot is over 200x.
You will not see another setup like this in your investment career very often. The last time I saw an equivalent risk-return profile in a publicly traded equity was IQE plc in summer 2024, which I wrote about and which has roughly tripled since.
The setup is rare because the conditions required to produce it are rare. You need a real technology monopoly. You need a customer set that is actively qualifying. You need a valuation that has been depressed by an unrelated business segment (in this case, Solar). You need management to be sandbagging guidance. You need zero US sell-side coverage. You need a Mittelstand cap stack with no convertible overhang. You need a 50-year company culture that produces institutional reliability. LPKF has all of these things at the same moment.
The stock has already moved 92 percent in nine months and most of the move is the last six weeks. That is uncomfortable. It is also still cheap on the only set of numbers that will matter in 2027 and 2028. The risk-reward at $12.01 is, in my view, the cleanest asymmetric long I have seen in this market.
The bottleneck people are talking about is glass itself. The bottleneck I think actually matters is what you do with the glass after you have drilled it. LPKF just told us they own that step too. Position accordingly.
Disclosure & Methodology
Position: Shawarma Capital is long $LPKF (LPK.DE on XETRA, LPKFF on US OTC) at an average entry price meaningfully below the current $12.01 print. This is one of the largest positions in the portfolio and I expect to add on weakness. I have been adding into the recent strength as well. I have no plans to trim before any of the catalysts identified in Section IX play out.
Sourcing: The factual claims in this essay are drawn from primary sources where possible, including LPKF’s FY25 annual report, FY25 press release, Q4 2025 earnings call transcript, and Vitrion product documentation. Customer relationship triangulation draws from Tom’s Hardware (May 2024), TechSpot (May 2024), TrendForce (May 2024 and December 2025), ETNews, Digitimes (January 2025), BusinessKorea (March 2024), Onto Innovation’s PACE consortium publications, and SEMI Engineering archive content. Industry sizing draws from Yole Group, Future Markets Inc., LightCounting, and IEEE IRDS. Patent record analysis includes the 2023 Intel CPO patent surfaced by vlmk on X. Where data is estimated rather than sourced, I have flagged it as such in the chart caption or table footnote.
Modeling: All revenue scenario projections, EBIT margin paths, multiple expansion paths, and price target calculations are Shawarma Capital estimates. They are bottoms-up models built from the disclosed customer set, the indicative tool ASP assumptions, and the public glass substrate market sizing forecasts. They are not forecasts in any predictive sense. They are conditional pricing exercises that ask: if scenario X plays out, what does the stock look like? You should treat them as a framework for thinking about the asymmetry, not as precise targets.
Conflicts of interest: I have no relationship with LPKF, Intel, Samsung, DNP, Absolics, JNTC, Onto Innovation, or any of the other companies named in this essay. I have no consulting relationship, advisory role, or board seat with any party that would benefit from the publication of this thesis. I do not receive compensation from any source related to LPKF coverage. The only economic interest I have in this essay is that I am long the stock.
Not investment advice. This is research synthesis. Do your own work. If you are not comfortable doing your own valuation work on a German microcap industrial laser company with a glass-substrate equipment monopoly thesis, do not buy this stock. If you are comfortable, position size appropriately to your own risk tolerance and time horizon. The asymmetry described in this essay is real, but the path to realizing it could include 30 to 50 percent drawdowns along the way. If you cannot stomach those, this is not the right position for you.
If you found this useful, please share it. The Shawarma Capital free tier publishes the thesis-level work, the paid tier publishes the position-sizing, entry, and catalyst tracking. Subscribe at shawarmacapital.substack.com.
More in this series
The Glass Monopoly: The Complete Series
The Glass Monopoly, Part 4: The Standoff Went To A Vote, Management Won, An
The Glass Monopoly, Part 3: There is an activist already in the boardroom
Incredible write-up. And LPKF might have the potential runway that is described. But backlog levels fell. And guidance for 2026 is lower in revenue than 2025. So do we have information from primary sources that underwrites this potential? I added it to my watchlist.
You made me buy this