Coherent: A Deep-Dive Value Investing Study

Hunting ten-year five-baggers among great growth stocks — pressing the upside question: "Can it get much bigger?"

• QHow high is its market ceiling? Is it enlarging a slice of an existing pie, or creating an entirely new market?6/10

  A

  Conclusion first: both, but the main story is "enlarging a slice of an already-exploding existing pie," not conjuring an entirely new market out of nothing. Coherent's ceiling is propped up mainly by the optical-interconnect demand of AI data centers, a pie that already existed but has been re-amplified by the scale-up of AI training/inference clusters; it did not invent the category of "optical communications," but rather lodged itself in the scarcest, fastest-growing link within this pie.

  How big the pie is actually getting can be seen through third-party figures. The report cites LightCounting's April 2026 data: the Ethernet optical transceiver market grew 93% in 2024, is expected to grow another 82% in 2025, and 65% in 2026; Dell'Oro forecast in May 2026 that the optical transport equipment market will grow 16% in 2026, driven mainly by AI data centers. Together these numbers show that this "existing pie" of optical interconnect is in a historic expansion phase; but note that the 82%→65%→16% curve is itself already decelerating, meaning this is high-speed but gradually receding growth, not a brand-new market that has only just sprouted.

  Coherent's own size confirms that it is "enlarging its own share of the pie." In the most recent quarter (Q3 FY2026, ended 2026-03-31), total revenue was $1.806 billion, of which the Datacenter & Communications segment was $1.362 billion, a big ramp from about $1 billion in the same period a year earlier; in the first nine months of FY2026 that segment's revenue was about $3.660 billion, roughly 72% of total revenue. In other words, the company's revenue mix has already been reshaped by this pie: two years ago it looked more like an industrial laser company, and now AI optical interconnect is the absolute main engine.

  Measured with Baillie Gifford's "ceiling" ruler, this is a real but bounded long slope. It fits the "demand structurally amplified" trait that Baillie Gifford favors: hyperscale data centers' near-insatiable hunger for bandwidth is a near-certain direction for many years; but it does not fit Baillie Gifford's most-beloved narrative of "creating a market that did not previously exist" (like e-commerce or streaming back in the day). Coherent is seizing the most critical device and module positions within an existing track, and its ceiling is determined by customers' capex cadence and technology roadmaps rather than defined by itself. The report gives "industry attractiveness" only 3/5, characterizing it as "a strong platform company in a good track plus a slice of traditional cyclical business," which is fair: not low, because AI has truly raised the industry ceiling; not full marks, because what it does is enlarge an existing pie whose growth curve is already retreating, rather than open up a brand-new virgin territory it prices itself.

  Jun 10, 2026
• QOver the next five years, can its revenue at least double? Is growth driven mainly by volume, price, or new businesses?6/10

  A

  Conclusion first: doubling revenue within five years (from about $5.8 billion to $11.6 billion) is possible, but far from a sure thing; it depends heavily on whether the AI optical-interconnect boom can continue, and the growth will be absolutely dominated by "volume," with "price" a phase-specific tailwind and "new businesses" contributing little.

  First, how strong the momentum is. FY2025 revenue was $5.810 billion, up 23.4% year over year; first-nine-months FY2026 revenue was about $5.073 billion, up about another 19% year over year, with Datacenter & Communications segment revenue up +34% and segment profit +39%, while Industrial segment revenue fell -8% year over year but segment profit still rose +10%. Extrapolating this line: if the company as a whole can maintain a low-double-digit compound growth rate, doubling in five years requires only a CAGR of about 15%/year, which, given the current momentum on the AI side, is not an exaggeration. But "doubling" requires the overall growth rate, and the industrial side is still contracting, meaning the burden falls almost entirely on the single leg of Datacenter & Communications.

  The main driver of growth is "volume." The report makes clear this is a B2B hardware/component business, with revenue growing through selling more transceivers, coherent transmission components, pump lasers, and various optical modules. The most direct evidence is the aggressiveness of capacity expansion: first-nine-months FY2026 capital expenditure already reached $547 million (higher than the full FY2025 figure of $441 million), while inventory surged from $1.438 billion at the end of June 2025 to $2.127 billion at the end of March 2026: the company is laying the physical groundwork to "sell more units." The report cites a May 2026 investor presentation: the company expects to double internal InP output within the year and double it again by 2027, with the 6-inch platform mass-producing EMLs, CW lasers, and photodiodes. These are all classic "win by volume" moves.

  "Price" is a tailwind but not a pillar. In both FY2025 and the first nine months of FY2026, the company attributed gross-margin improvement to "pricing optimization" and cost improvement; but the report honestly points out that this is "phase-specific pricing power during the up-cycle, not unlimited pricing power." The historical rule of the optical-module industry is: once standards converge and customer specifications become clear, price pressure will come sooner or later. So relying on price increases to sustain a doubling is unrealistic; price is more about improving unit profitability than driving a doubling of revenue.

  The increment from "new businesses" is amplified by one key variable: NVIDIA's roughly $2 billion strategic investment (according to Coherent's Q3 FY2026 filing, subscribed in March 2026 for about 7.79 million shares at $256.80 per share) came with a multi-year, several-billion-dollar purchase commitment. This amounts to a semi-locked-in insurance policy for "volume" growth, but the report also cautions that this precisely aggravates customer concentration: the benefit and the risk are two sides of the same coin.

  With Baillie Gifford's ruler: the doubling is achievable, but the "quality" deserves a question mark. The point of Baillie Gifford's "double in five years" question is to test whether a business has enough endogenous growth power. Coherent's growth rate is sufficient, but its doubling is a "capital-intensive volume ramp": it must first pour in capex, build up inventory, and tie up large amounts of working capital to get there. The report points out incisively that it is currently in an expansion phase of "improving profit, strained cash," with first-nine-months FY2026 operating cash flow of only $10.06 million. So the more accurate statement is: revenue doubling can be expected, but it is a kind of "growth bought with capital," far removed from Baillie Gifford's favorite "asset-light, high-cash-conversion compounding doubling."

  Jun 10, 2026
• QFive years from now, what will take over as the next growth engine? Does this "second curve" exist today?4/10

  A

  Conclusion first: Coherent today lacks a clear, independent "second curve." Of its two businesses, one is the current main engine (AI optical interconnect) and the other is a shrinking old engine (industrial); what can truly "take over" five years from now is more likely an extension of the same main curve into higher-value links (systematization, CPO/co-packaged optics) rather than an entirely new growth pole.

  First, clarify the two legs of the status quo. The report discloses that the company is split into two segments, Datacenter & Communications and Industrial, accounting for about 72% and 28% of revenue respectively in the first nine months of FY2026. The problem is: Datacenter & Communications is the "present-tense" first curve, up 34% year over year; the Industrial segment is not only not a second curve but is actually down -8% year over year (the report discloses that SiC and automotive-related end markets were weaker in FY2025 and industrial demand was soft overall). So on the question of "who takes over in five years," one cannot count on Industrial to fill the hole left after the AI tide recedes: it is itself a cyclical, growth-starved legacy business.

  So where is the second curve hidden? The most realistic candidate is "moving up from devices/modules toward systems and toward next-generation optical architectures." The report cites a May 2026 investor presentation and the NVIDIA strategic partnership: the two sides will jointly develop advanced optical technology to scale next-generation data-center architecture, with the cooperation covering advanced laser and optical-networking products. In its risk section the report lists multiple evolving technology routes such as CPO (co-packaged optics), NPO, silicon photonics, InP, VCSEL, and multi-lane transmission. If Coherent can keep occupying a core position within "next-generation interconnect paradigms" like CPO/silicon photonics, that could indeed extend the life of the first curve, and even recreate a higher-value increment; but this is essentially an upgraded version of the main curve rather than a new business started from scratch.

  Measured against Baillie Gifford's "does a second curve exist today" standard, Coherent scores on the weak side. What Baillie Gifford truly admires is a second engine that, like Amazon incubating AWS out of e-commerce, is financially separable from the original curve and can grow independently into a new pillar. Coherent currently has nothing like that: its growth imagination is almost entirely bet on "how long the single AI optical-interconnect curve can last and how high it can climb," while the old industrial leg that could have diversified risk is shrinking. This is precisely the source of the report's core concern: once AI optical interconnect cools, the company has no mature backup engine to hedge, which is also why it lists industrial-side weakness as a key risk and emphasizes that "the AI side being unable to fully hedge the industrial side's sustained weakness" would trigger a reassessment.

  Honestly: Coherent has a real technology reserve and big-customer binding to extend the main curve upward, but "an independent second curve that takes over in five years" does not clearly exist today. This is a structural shortcoming for it as a growth name, and one should not, in order to fit a growth narrative, dress up "the natural extension of the main curve" as a "second curve."

  Jun 10, 2026
• QWhat is its core competitive advantage? Will this moat widen or narrow over the next three to five years?5/10

  A

  Conclusion first: Coherent's core moat is not brand or network effect, but the combination of three things: "process difficulty plus the customer-qualification barrier plus end-to-end vertical integration from materials to systems." Over the next three to five years, this moat will most likely widen amid the AI optical-interconnect boom, but it is a moat that "needs continuous heavy capital investment to maintain," and its width is not impregnable.

  First, what the moat is. In the report's moat scoring table, brand, network effect, and data advantage are all rated "moderate" or "weak"; what is truly rated "moderately strong" is cost advantage (vertical integration, the 6-inch platform, internal InP expansion, the device-to-system chain) and the patent/process barrier (the manufacturing know-how of optical devices, materials, and laser systems is hard to replicate), while scale advantage is "stronger than most pure single-point competitors." The report's overall judgment is a moat strength of 3/5, and it is explicit: replicating Coherent's integrated capability across InP, lasers, modules, and systems typically takes many years and capital on the order of several billion dollars.

  Why say it will "widen" over the next three to five years? There are three solid pieces of evidence. First, the company is pulling ahead through capacity: the report cites a May 2026 investor presentation in which the company expects to double internal InP output within the year and double it again by 2027, and the 6-inch platform produces EMLs, CW lasers, and photodiodes at yields higher than the 3-inch line: higher yield = lower unit cost = harder to catch up. Second, the moat is being converted into profit: Q3 FY2026 GAAP gross margin rose to 37.7% (+243 basis points year over year), Datacenter & Communications segment revenue was +34% and segment profit +39% in the first nine months of FY2026, and the company attributes the improvement to efficiency, yield, and pricing optimization: a moat that is only wide but does not make money does not count, and here it is indeed monetizing. Third, NVIDIA's $2 billion stake plus multi-year purchase commitment locks it into the core supply position of the next-generation AI optical architecture, which strengthens the qualification barrier and switching cost.

  But why call it "not impregnable"? The report notes three points of fragility with great restraint. One, the price of this kind of moat is continuous heavy capital investment: unlike a consumer brand that "compounds asset-light," it must be sustained through nonstop capex (first-nine-months FY2026 $547 million). Two, the optical-module industry has an iron law: once standards converge and customer specifications become clear, a price war will come sooner or later, so the report prefers to view the current high profit as "a structural advantage and a cyclical dividend existing at the same time" rather than permanent excess profit. Three, the report explicitly acknowledges that competitors include both peers (Lumentum, IPG Photonics, and others) and substitute technologies, and further include "customers who may integrate back upstream": in other words, if the largest customers (such as hyperscalers) accelerate in-house development, the moat will be eroded from the demand side.

  Measured with Baillie Gifford's ruler: this is a moat that is "widening in direction but not top-tier in purity." What Baillie Gifford favors is the kind of moat that widens the more it is used and does not need constant cash-burning to maintain (network effects, platforms, standard-setting power). Coherent's moat is pointed in the right direction and has indeed widened over the past three to five years in the AI tailwind, but it is essentially a "capital-heavy process-and-integration barrier" that needs continuous investment and leading-customer binding to stay alive; once industry standards solidify or big customers integrate upstream, its width will come under pressure. So the honest conclusion is: the moat truly exists and has widened recently, but it does not deserve the imagination of "permanent excess profit," and a score of 3/5 is appropriate.

  Jun 10, 2026
• QIf its core business is disrupted, does it have the genes to reinvent itself? How does it treat mistakes and bad news?5/10

  A

  Conclusion first: Coherent's "self-reinvention genes" are moderately weak. It has indeed gone through one drastic identity transformation (metamorphosing from the industrial-laser II-VI into an AI optical-interconnect platform), but this transformation owes more to catching an external wave plus a large acquisition than to a long-proven endogenous regenerative ability; on treating mistakes and bad news, the new management's approach leans rational, but "the record is still short."

  First, whether it has a historical sample of "rebirth after disruption." The report discloses a key evolution of reporting basis: in 2021-2022 the company was still in the former II-VI phase; 2023 was the first full fiscal year after acquiring the former Coherent (an optics-and-laser giant), clearly affected by purchase-price allocation and inventory fair-value adjustments; and from July 2025 it reorganized its disclosure into the two segments of Datacenter & Communications and Industrial. This line shows the company is capable of reshaping itself from a "materials/industrial-laser company" into a "composite platform of AI optical interconnect plus industrial photonics"; the report's own words are "Coherent is no longer an ordinary industrial laser company." This is a real transformation, proving it is not rigid. But be clear-eyed: the core driver of this reshaping is one giant acquisition plus happening to catch the AI-data-center explosion, rather than the company carving out a new path through its own organizational genes when its core business was disrupted. In its "open questions," the report also candidly admits that the most core incremental value (the margins, capacity, and order fulfillment of the NVIDIA cooperation) still awaits verification in future earnings reports: the reinvention has not yet closed the loop.

  If the core AI optical interconnect is disrupted by a new paradigm in the future, can it hold up? Half glad, half worried. The glad side: the report notes the company has technology reserves across multiple routes such as CPO, NPO, silicon photonics, InP, VCSEL, and multi-lane transmission, and its vertical integration from materials to systems gives it the ammunition to "switch routes." The worried side: the report equally states plainly that "although the company has a technology reserve, that does not mean it can win every route": a reserve does not equal winning. Even more critically, it is now heavily betting on a single boom direction while the old industrial leg is shrinking (Industrial down -8% year over year); once the main curve is disrupted, it lacks a mature backup engine to cushion the blow.

  How does it treat mistakes and bad news? This is exactly the window for observing an "honest culture," and the evidence leans positive but is limited. First, it does not cover up bad regulatory news: the report discloses that after receiving an inquiry from the U.S. BIS in January 2025 regarding past sales to Huawei, the company stopped shipping to Huawei and candidly admitted it could not yet predict the outcome: faithful disclosure rather than downplaying. Second, it does not whitewash historical baggage: over the past two years the current management (per the report, CEO Jim Anderson took office in June 2024 and CFO Sherri Luther joined in October 2024) has sold non-core assets, deleveraged (total debt fell from $4.310 billion at the end of June 2023 to $3.194 billion at the end of March 2026), brought in NVIDIA strategic capital, and pushed through the preferred-stock conversion; the report characterizes this as "cleaning up the past rather than having already entered an easy harvesting period," a posture of acknowledging and correcting historical mistakes.

  But two places call for vigilance, constituting deductions from a "healthy error-correction culture." One, the company uses non-GAAP adjustments heavily: in FY2025 it added back about $161 million of stock-based compensation, $303 million of acquisition amortization, $160 million of restructuring, and $85 million of impairment, among others, for non-GAAP net income attributable to the parent of $693 million versus GAAP of only $49.36 million; the report cautions that this means the gap between "adjusted profit" and truly distributable profit is large, and long-term investors must beware of using pretty metrics to mask problems. Two, per the report's citation of the proxy statement, in 2024 the non-binding shareholder vote on executive compensation drew about 81.58 million votes in favor and about 73.55 million against, close to a 50-50 split, showing the market does not fully buy into its incentive scheme.

  With Baillie Gifford's ruler: Baillie Gifford places special weight on "whether it can reinvent itself through organizational genes when the core business is disrupted" and "whether management faces bad news calmly." On the former, Coherent is "has had one successful transformation, but through acquisition plus a wave rather than purely endogenous genes"; on the latter, it is "the new team handles bad news relatively honestly, but the tenure record is short, and the non-GAAP adjustments need continuous watching." On balance, it is not the kind of organization that has repeatedly proven itself "unkillable and able to be reborn," but rather a company that has just completed one externally driven transformation and whose error-correction culture still awaits the test of time: its genes are acceptable, but far from the benchmark level of Baillie Gifford's ideal of "self-revolution in adversity."

  Jun 10, 2026
• QDoes management (especially the founder) have long-term vision and interests deeply aligned with the company? Are they willing to sacrifice current profit for five to ten years out?4/10

  A

  Conclusion first: this is one of Coherent's most obvious shortcomings as a Baillie-Gifford-style name. It has no founder at the helm, the current professional-manager team's tenure record is still short, and management's direct shareholding is extremely low (all directors and officers combined hold less than 1%), so "deep real-money alignment with shareholders" is not strong; however, on "willingness to sacrifice current profit for the long term," the company has given a positive answer through concrete action.

  First, how weak the "alignment" is. Per the report's citation of the 2025-08-31 data disclosed in the 2025 proxy statement: CEO Jim Anderson beneficially held about 25,998 shares, CFO Sherri Luther only 285 shares, and all directors and officers combined about 706,906 shares, less than 1% of total shares. Against the current total share count of about 195.6 million shares and market cap of about $69.6 billion, this level of shareholding is indeed thin as interest alignment. This does not mean management is unreliable, but it is far removed from Baillie Gifford's favored sample of "founder holding a heavy stake, personal wealth rising and falling with the company's fate" (such as the growth stocks it heavily holds where founders own 20-30%). Coherent is a company with highly dispersed ownership run by professional managers, lacking the kind of head who "bets his entire net worth on ten years out."

  Next, "long-term vision and tenure record." Per the report, CEO Jim Anderson took office on June 3, 2024 (previously CEO of Lattice Semiconductor), and CFO Sherri Luther joined in October 2024. In other words, this "new Coherent era" core team has an operating record of less than two years so far; the report explicitly notes that "the current core management team's operating record for the new Coherent era is not yet long" and "the verifiable long-term excellent capital-allocation record is not yet long enough." What Baillie Gifford wants is a leader who can see clearly "what this company will become in five to ten years," and here we hold only a very short sample, unable either to falsify or to confirm their long-term vision.

  "Whether willing to sacrifice current profit for five to ten years out": on this item, the company's answer is affirmative, and it is where it most resembles growth-oriented management. The strongest evidence: in the first nine months of FY2026 the company was willing to squeeze operating cash flow down to only $10.06 million (from $503 million in the same period a year earlier), build inventory from $1.438 billion to $2.127 billion, and raise capital expenditure to $547 million, all to lay capacity for a higher future revenue level: a classic "sacrifice current cash for long-term share." Management has also publicly planned to double internal InP output within the year and double it again by 2027. The report's assessment of capital allocation over the past two years leans positive: no token buybacks during a high-volatility period (no common-stock buybacks in 2025), but prioritizing deleveraging (total debt $4.310 billion → $3.194 billion), divesting non-core businesses (selling the aerospace and defense business in September 2025 for about $400 million in cash), bringing in NVIDIA's $2 billion strategic capital, and lowering financing costs; the report characterizes this as "clearly more rational than before, but not yet excellent."

  But "investing for the long term" also has a dilution side, which must be honestly counted in. The report notes: the company just issued 7,788,161 shares to NVIDIA in March 2026, and earlier there was the dividend and dilution burden of the Series B convertible preferred stock (fully converted by the end of March 2026). For long-term shareholders, if growth relies mainly on continually issuing shares to raise capital, per-share value growth may not keep pace with the growth in company scale: this is also why the report separately lists "capital-allocation and dilution risk."

  An integrated judgment with Baillie Gifford's ruler: the weaknesses are clear and the highlights are limited. The core of this Baillie Gifford question is "founder-level long-termism plus deep interest alignment plus daring to sacrifice the present for the distant future." Coherent scores on "daring to sacrifice the present for the distant future" (heavy investment in capacity, squeezing cash flow for growth), but on "founder at the helm" it is entirely absent, on "deep interest alignment" it is nearly failing (shareholding <1%), and on "verifiable record of long-term vision" the sample is too short. The report gives management and capital allocation 3/5, and judges "is management trustworthy" as a pass (no obvious integrity red flags for now) and "is capital allocation rational" as a pass (recently rational but with heavy historical baggage), which is fair: this is a professional-manager team that is pointed in the right direction but has not yet proven itself over a long enough time and is loosely bound to shareholders, falling short of Baillie Gifford's ideal management picture.

  Jun 10, 2026
• QIf it disappeared tomorrow, how much would customers miss it? Is its way of growing sustainable and free of reliance on harming society and regulation?5/10

  A

  Conclusion first: if Coherent disappeared tomorrow, a few core customers (especially hyperscale data centers and partners like NVIDIA) would "miss it quite a bit," because it has short-term-hard-to-replace scarcity in the key devices and capacity of AI optical interconnect; but it is not the "only indispensable" sole-source supplier, and there are peers like Lumentum in the industry that can partly substitute for it. On the dual test of "whether growth harms society and regulation," the company is fundamentally healthy and positive (it sells key devices for AI infrastructure, benefiting the expansion of computing power), but there is one real, unresolved regulatory matter (the BIS inquiry into sales to Huawei) that needs continuous watching.

  First, how strong the "indispensability" is. The evidence leans strong but is not absolute. One, demand-side scarcity is written into the revenue: Q3 FY2026 Datacenter & Communications segment revenue was $1.362 billion, a big ramp from about $1 billion in the same period a year earlier, showing customers increasingly depend on its optical modules and coherent devices. Two, the strongest "miss-it" signal is that NVIDIA was willing to put in about $2 billion for a stake along with a multi-year, several-billion-dollar purchase commitment; per Coherent's Q3 FY2026 filing, that investment was 7.79 million shares at $256.80 per share, totaling about $2 billion, completed on March 2, 2026. A chip giant willing to use real money to bind a supplier into its own next-generation architecture is itself a market endorsement that "this supplier is indispensable." Three, the report notes that replicating Coherent's integrated capability from materials, InP, and lasers to modules and systems takes many years and several billion dollars, and customers cannot find a fully equivalent substitute in the short term.

  But "indispensable" has a ceiling and cannot be inflated into sole-source. The report explicitly rates its customer "switching cost" as "moderate" (not extremely high) and candidly acknowledges that competitors include both peers like Lumentum and IPG Photonics and substitute technologies, and further include "customers who may integrate back upstream": in other words, if hyperscalers accelerate in-house development or split orders among peers, Coherent can be partly substituted. In the peer group the report cites, Lumentum's current size is likewise huge and it is also growing fast under AI demand, proving this is not an exclusively monopolized link. So customers would "miss it a lot," but would not be "unable to carry on."

  "Whether growth is sustainable and free of reliance on harming society and regulation": this is the dual test Baillie Gifford adds to this question, viewed from two sides. The positive side: the company's growth comes from supplying key optoelectronic devices for AI data centers, telecom networks, semiconductor capital equipment, and display and precision manufacturing; it essentially supports the expansion of computing and communications infrastructure, its social utility is positive, and there is no business model of "growing by harming users or society" (not gambling, not addictive products, not regulatory arbitrage). The report also confirms that FY2025 internal controls were deemed effective by management and the auditor, and FY2026 Q3 controls and procedures remained effective, with no strong signs of fraud at the accounting level: the "texture" of the growth is clean.

  The cautionary side: there is one real regulatory uncertainty that must be honestly counted in. Per the report, in January 2025 the company received an inquiry from the U.S. BIS (Bureau of Industry and Security) regarding past sales to Huawei; the company has stopped shipping to Huawei but cannot yet predict the outcome; it also faces uncertainty from tariffs, rare earths, and the global trade environment. This means part of its historical growth once touched the sensitive zone of export controls: although it has proactively stopped supply and this is "past" conduct, if the investigation evolves into a substantive penalty, it would hit both reputation and the compliance narrative. The report lists "the regulatory event evolving into a substantive penalty" as one of the signals that trigger reassessment.

  An integrated judgment with Baillie Gifford's ruler: "miss-it degree" is upper-middle, and "social/regulatory sustainability" is broadly acceptable but with one mine still to be defused. The essence of this Baillie Gifford question is "indispensable to the point that customers cannot do without it, and a way of growing that can withstand the long-term test of society and regulation." Coherent scores decently on the first half (core customers are hard to replace in the short term, NVIDIA endorses it with a stake), but does not reach "absolute sole-source"; on the second half, its business is essentially a healthy, positive infrastructure supplier, but on regulation the BIS/Huawei unresolved matter constitutes a real risk point. The conclusion: customers would miss it a lot, its social value is positive, but it is neither an irreplaceable exclusive chokepoint nor fully free of the shadow of export controls: it is a key supplier that is "important but replaceable, positive but needing compliance watching."

  Jun 10, 2026
• QHow are this business's unit economics (gross margin, incremental returns)? Do they get better or worse as it scales? Where is the money it earns spent?5/10

  A

  Conclusion first: Coherent's unit economics are "clearly improving but still not excellent." Gross margin is rising with scale and yield (Q3 FY2026 GAAP gross margin 37.7%), and getting bigger does bring positive leverage; but it is a high-manufacturing, high-R&D, high-working-capital-consuming heavy-asset business, and the money it earns is currently almost all reinvested into capacity and inventory, still far from the mature phase of "high cash conversion."

  First, the trend of gross margin and incremental returns: this is the most direct thermometer of unit economics. The report discloses that GAAP gross margin rose from 31.4% in FY2023 and 30.9% in FY2024 to 35.2% in FY2025 and about 37.1% in the first nine months of FY2026 (with the latest single quarter, Q3 FY2026, reaching 37.7%, +243 basis points year over year). More telling is the segment-level incremental return: in the first nine months of FY2026, Datacenter & Communications segment revenue was +34% year over year and segment profit +39%; profit growing faster than revenue means that scale expansion is producing positive operating leverage. The company attributes this to efficiency gains, yield improvements, and pricing optimization.

  "Better or worse as it scales": the answer is structurally better, but the degree of improvement is capped by an industry iron law. The physical basis for the improvement is solid: the report cites a May 2026 investor presentation in which the 6-inch platform mass-produces EMLs, CW lasers, and photodiodes at yields higher than the 3-inch line, and internal InP output is set to double within the year and double again by 2027; larger scale plus higher yield equals lower unit cost, a classic "bigger is cheaper." But the report cautions with restraint: in the optical-module industry "once standards converge and customer specifications become clear, price pressure will come sooner or later," so it prefers to view the current high gross margin as "a structural advantage and a cyclical dividend existing at the same time" rather than excess profit that can be extrapolated forever. In other words, the direction of unit-economics improvement is real, but the ceiling is sealed by competition and price wars, and it will not rise indefinitely.

  "Where is the money it earns spent": this is currently the most critical point, and the one that most makes value investors wary: the money goes almost entirely into growth investment rather than back to shareholders. Three sets of numbers best illustrate this: First, R&D is a big chunk, FY2025 R&D was about $582 million, roughly 10% of revenue, and first-nine-months FY2026 was about $507 million, again about 10%. Second, capital expenditure is being ramped up, with first-nine-months FY2026 at $547 million, already exceeding the full FY2025 figure of $441 million. Third, inventory heavily ties up working capital, surging from $1.438 billion at the end of June 2025 to $2.127 billion at the end of March 2026. The direct consequence of these three combined is: first-nine-months FY2026 operating cash flow of only $10.06 million and simple free cash flow of -$537 million. The report puts it incisively: "this looks more like expansion-phase profit than mature-phase cash-cow profit." At the same time the company did not use the money for buybacks (no common-stock buybacks in 2025) but prioritized deleveraging (total debt down to $3.194 billion): the money goes to "capacity plus R&D plus debt repayment," not "dividends plus buybacks."

  One must also honestly puncture a metric trap: the huge gap between GAAP and non-GAAP. The report emphasizes that FY2025 non-GAAP net income attributable to the parent was $693 million while GAAP was only $49.36 million, with add-backs including about $161 million of stock-based compensation, $303 million of acquisition amortization, and $160 million of restructuring. This means that when assessing true unit economics, one cannot simply take the pretty non-GAAP EPS at face value; the conservative "owner earnings" range the report gives is $550 million to $700 million, and it explicitly does not treat all stock-based compensation as freely addable back.

  An integrated judgment with Baillie Gifford's ruler: the unit economics are "right in direction, medium in texture." The essence of this Baillie Gifford question is "whether gross margin and incremental returns are excellent, whether they get better with scale, and whether the money earned is efficiently reinvested for excess returns." Coherent scores on "gross margin improving with scale" and "segment profit growing faster than revenue," proving it has real operating leverage; but its unit economics are dragged down by three things: the heavy-capital nature leads to poor cash conversion (operating cash flow near zero), industry price wars cap the profit ceiling, and the huge gap between GAAP and non-GAAP discounts "truly distributable profit." The report judges "can it generate stable free cash flow" as a fail and "is return on capital excellent" as uncertain, which is appropriate. This is a medium-texture business that is "getting better but still does not deserve a 99-126x owner-earnings valuation," definitely not Baillie Gifford's favorite compounding machine of "high margin, light capital, cash gushing out."

  Jun 10, 2026
• QTo make it rise five-fold in ten years, what conditions must hold simultaneously? Are those conditions realistic? What expectations does today's share price imply?2/10

  A

  Conclusion first: making Coherent rise several-fold over ten years requires several high-difficulty conditions to hold "simultaneously," and at the current price of about $355.94 (2026-06-09 close), a market cap of about $69.6 billion, and a TTM P/E of nearly 149x, the market has in fact already priced in a good chunk of that "ideal script" in advance. In other words, buying today is not betting on "whether it will happen," but on "whether it can be even better than this already very high expectation": the margin of safety is nearly zero.

  First, list the conditions that a "ten-year five-bagger/several-bagger" needs to hold simultaneously, checking each against reality:

  First, the AI optical-interconnect boom must last longer than the market expects. The third-party data the report cites is the Ethernet optical transceiver market at +93% in 2024, +82% in 2025, and +65% in 2026; the growth rate itself is decelerating, which means "how many years the high-speed growth can last" is the biggest variable, not "whether it will grow." Reality check: the direction is certain, but requiring it to maintain high growth for ten years carries a heavy dose of optimism.

  Second, Coherent must successfully convert its enormous capacity expansion and inventory into high-quality, sustainable free cash flow. This is currently the weakest link: first-nine-months FY2026 operating cash flow was only $10.06 million, simple free cash flow was -$537 million, and inventory piled up to $2.127 billion. Reality check: today there is not even the shadow of a "cash cow" yet, and this condition means turning things around starting from a negative, which is highly difficult.

  Third, the competitive landscape must not deteriorate noticeably, and no price war can break out. The report repeatedly warns that in the optical-module industry "after standards converge, price pressure will come sooner or later," and that big customers may integrate upstream through in-house development. Reality check: this is a condition that is "harder to maintain the longer time goes on."

  Fourth, customer-concentration risk must not blow up. Per the report, in FY2025 two customers accounted for 12% and 10% of revenue respectively, and after the NVIDIA cooperation grows in the future, concentration could rise further; a design switch or in-house development by any single large customer would badly hurt results.

  Fifth, the market must be willing, over the long term, to give it a valuation center far above that of a traditional hardware manufacturer. This is the most hidden and most dangerous one: even if the first four all come true, as long as the valuation multiple reverts to normal, the share price could tread water or even fall.

  The probability of these five holding "simultaneously" determines why none of the report's three valuation tiers can explain the current price. Based on discounting owner earnings, the report gives: conservative-scenario intrinsic value of about $45-70/share, neutral about $80-130/share, optimistic about $160-220/share; even the upper edge of the most optimistic tier, $220, is still nearly 40% below the current price of $355.94. The report states plainly that "even in a somewhat optimistic scenario, it is hard to interpret today's price as offering room." From the owner-earnings angle it is even more glaring: at a market cap of about $69.6 billion, it equals about 126x conservative owner earnings (the $550 million basis), and even using the loose $700 million basis it is close to 99x.

  "What expectations does today's share price imply": this is the landing point of the question, and the answer is: the current price already implies a near-perfect script in which "almost all of the above high-difficulty conditions are realized, and the valuation center stays high over the long term." The report's phrasing is that "investors are effectively paying for an idealized ten-year script rather than buying an asset at a clear discount" and that "the current price demands that the next ten years be near-perfect." Corroborating from another angle: the long-term annualized return expectations the report gives based on the current price are conservative -8% to -3% per year, neutral -1% to +4% per year, and optimistic only +5% to +9% per year. That is, even if the optimistic scenario is realized, the return may not beat the roughly 4.61% ten-year Treasury and the broad-market index; and once any one of the five conditions falls through, the report considers a long-term drawdown of 50%-70% "not an exaggeration."

  An honest wrap-up with Baillie Gifford's ruler: the direction is one Baillie Gifford would like, but the price is not. Baillie Gifford is willing to tolerate a high valuation for a "great growth stock that five-bags in ten years," provided the upside imagination (blue-sky) is big enough and the current price has not wrung out all the good expectations. Coherent's track imagination is big enough, but its "ten-year surge" needs five high-difficulty conditions to hold simultaneously, and the current price already prices those conditions as accomplished facts; this is exactly why the report judges both "is the valuation below intrinsic value" and "is the margin of safety sufficient" as fails, with a final rating of "Watch." Can it five-bag in ten years? In theory it is possible; but today's share price gives you no room for error, and looks more like a "dream bid" than an "acquisition bid."

  Jun 10, 2026
• QWhy has the market not yet realized all this? Is it that it cannot understand, looks down on it, or cannot see far enough? What will become the "narrative inflection point"?3/10

  A

  Conclusion first: for Coherent, this Baillie Gifford question must be asked in reverse: the market is not "not yet aware of how good it is," but has already fully, even excessively, become aware of it. What is truly not fully priced by the market is its "fragility": cash flow not yet realized, highly concentrated customers, industry price-war risk, and the valuation's over-reliance on a perfect script. So the "narrative inflection point" here is not "the market finally understands the good news and rallies another leg," but more likely "a downward re-rating the day the market discovers the script is not perfect enough."

  First, let me make clear why the market's perception of this name is "seeing through it too clearly, giving it too much" rather than "cannot understand / looks down on / cannot see far." The evidence is written in the valuation: the current price is about $355.94 (2026-06-09 close), market cap about $69.6 billion, and TTM P/E nearly 149x, with the report recording its 2026-05-20 static P/E as high as 167.6x, price-to-sales about 10.5x, and EV/Adjusted EBITDA above 50x. The report also points out how exaggerated the market-cap comparison is: Coherent's market cap of about $69.6 billion is far higher than IPG Photonics's about $7.05 billion, and even higher than Lumentum's about $56.57 billion. A company the market "looks down on" would not have this kind of valuation; the market has clearly already priced it as "an AI optical-interconnect core asset plus a NVIDIA strategic partner." The report's own words: "the market has already given it an extremely high valuation as an AI optical-interconnect core asset in advance."

  So in which cognitive blind spots does the market's "over-optimism" show? First, the market prefers to stare at the pretty non-GAAP numbers: FY2025 non-GAAP net income attributable to the parent was $693 million, while GAAP was only $49.36 million, a huge gap. Second, the market tends to overlook the weakness of cash flow: first-nine-months FY2026 operating cash flow was only $10.06 million and inventory piled up to $2.127 billion; the profit improvement has not yet turned into real money. Third, the market treats customer concentration (in FY2025 two customers accounted for 12% and 10% of revenue) and price-war risk as distant problems that can be ignored for now. The report's strongest bear case is based precisely on this: "Coherent may indeed become stronger in the AI-optics era, but shareholders will not necessarily make money because of it, because the market has already given it an extremely high valuation as a core asset in advance"; this bear case does not require the company to fail operationally, only requires it to be "not good enough to match the current price."

  "What will become the narrative inflection point": this is the landing point of the question, and for Coherent the inflection is two-way, but the downward probability is more worth watching.

  Downward inflections (signals more likely to trigger a re-rating, which the report has listed as conditions for reassessment): One, Datacenter & Communications revenue growth clearly stalls, which is the lifeblood of the current valuation; Two, inventory keeps climbing but operating cash flow keeps failing to improve, proving that "expansion-phase profit" cannot convert into a cash cow; Three, gross-margin improvement stops and turns into decline, breaking the "bigger is more profitable" narrative; Four, the January 2025 BIS inquiry into sales to Huawei disclosed by the report evolves into a substantive penalty; Five, large equity financing or intense dilution appears again (following the issuance of about 7.79 million shares to NVIDIA). Any one being confirmed could make the market switch from the "core asset" narrative back to the "capital-heavy manufacturer" valuation center; the report considers a long-term drawdown of 50%-70% at that point "not an exaggeration."

  Upward inflections (conditions that would make the report willing to raise the intrinsic value): Datacenter & Communications maintains healthy cash conversion under high growth, with inventory turnover not deteriorating; after capex intensity falls, free cash flow turns significantly and sustainably positive; the NVIDIA cooperation and other hyperscale orders prove it is a long-term high-share core supplier rather than a "one-off boom beneficiary"; management continues to deleverage and raise per-share owner earnings without obvious dilution. Once these are realized, they would firm it up from a "dream bid" into "growth backed by results."

  An honest wrap-up with Baillie Gifford's ruler. The essence of this Baillie Gifford question is to look for "a great growth stock mispriced by the market because it cannot understand, looks down on, or cannot see far enough." Coherent does not belong to this category: it is a hot name the market understands very well and gives very fully, and the direction of the perception gap is "over-optimism" rather than "undervaluation." Therefore, for it, the more realistic thing is not to wait for the market to "suddenly wake up and then rally," but to beware of "a downward re-rating after the narrative inflection point appears." This is exactly the underlying logic of the report giving a "Watch" rating and defining the best move as "waiting for the market to offer a price that looks more like an acquisition bid than a dream bid."

  Jun 10, 2026