Eoptolink Research from a Long-Term Business Owner's Perspective

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 expanding an existing pie, or creating an entirely new market?6/10

  A

  The ceiling is very high, but Eoptolink is expanding an existing large pie rather than creating a brand-new market from scratch. It is selling the picks and shovels on the long slope of AI compute expansion, not defining the new demand itself.

  The pie itself is indeed expanding quickly. The report cites LightCounting's forecast that the global optical module market will grow at a CAGR of about 22% in 2025–2030, with Ethernet optical modules mainly used in AI compute clusters and cloud data centers growing at a CAGR of about 24%. This can be corroborated by third-party materials: LightCounting expects the global optical module market to grow at a CAGR of about 22% over the next five years, and Goldman Sachs further forecasts that the global optical module market will reach about USD 19.37 billion in 2026, with YoY growth above 50%, driven precisely by volume ramp-up of 800G/1.6T high-end products. The report also cites Cisco's relay of Dell'Oro's view: mainstream AI back-end network switch-port speeds will migrate to 800G in 2025 and to 1.6T in 2027. The height of the ceiling is therefore essentially determined by how many optical modules each generation of AI clusters needs each year, and how expensive they are. Per-rack optical interconnect value jumps from about USD 50,000 in the 400G era to USD 250,000–300,000 at 1.6T, a curve that rises as compute generations advance.

  But it is important to distinguish honestly between a large pie and a new market. Optical modules are a mature component category that has existed in data centers and telecom networks for more than twenty years. Eoptolink is taking share in higher-speed, higher-value segments within an existing category, and enlarging the whole category alongside AI capex. It has not created a demand paradigm that did not previously exist. The creators of that new paradigm are NVIDIA's GPUs and the large models themselves. Eoptolink is a beneficiary and executor of speed upgrades plus volume expansion in this industrial chain. The report's characterization is restrained: this is an excellent manufacturer in a strong track, not the company defining the track. Under Baillie Gifford's LTGG lens, the true blue-sky imagination belongs to companies that can create entirely new markets and control demand definition themselves. Eoptolink's upside is follower-style and determined by customer capex. Its elasticity is enormous, but the initiative is not fully in its own hands.

  A more important layer of risk lies in the shape of the ceiling: as pluggable components, optical modules face potential substitution from architectures such as CPO (co-packaged optics), NPO, and optical switching, which move the optical engine next to the switch chip. The report explicitly lists technology substitution risk as one of the core risks, and notes that in investor communications the company itself only says it has already laid out CPO, with progress depending on customer solutions. In other words, while this pie is expanding, the way it is sliced, meaning the packaging form, may change. If the architecture shifts faster than expected over the next few years, the current largest submarket ceiling for pluggable optical modules could instead be compressed at an inflection point.

  In one sentence: the track ceiling is high enough and long enough, which is Eoptolink's hardest foundation; but it is an expander of an existing large market rather than the creator of a new one, and the boundary of this pie will be redrawn by technology routes. Upside depends on capturing the capex wave and speed generations, not on monopolizing a new demand that no one else can match.

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

  A

  It can probably double, and even several-fold growth within five years would not be exaggerated. But this growth is driven almost entirely by volume × speed-generation upgrades, meaning a jump in unit price. New businesses contribute little, and the further out one looks, the more it depends on the strong assumption that the cycle does not roll over.

  Start with the real base. Eoptolink's 2025 revenue was RMB 24.842 billion, up 187.29% YoY, and net profit attributable to the parent was RMB 9.532 billion, up 235.89% YoY. These figures can be corroborated by the company's annual report and third-party sources (Eoptolink's 2025 revenue was RMB 24.842 billion, up 187.29% YoY, and net profit attributable to the parent was RMB 9.532 billion). 2026Q1 continued the high growth, with quarterly revenue of RMB 8.338 billion and net profit attributable to the parent of RMB 2.780 billion, according to the report. Sell-side forecasts are quite aggressive: the Founder Securities estimates cited by the report put 2025–2027 revenue at about RMB 25.8/52.6/72.4 billion, while Guolian Minsheng estimates about RMB 26.9/47.965/68.151 billion. Under these two consensus sets, company revenue is already close to or above twice the 2025 level by 2027. In other words, a five-year doubling is a fairly conservative floor in sell-side eyes; the market is betting on a steeper curve.

  When the growth engine is broken down, it has two tightly coupled wheels: volume and price.

  First, volume expansion comes from AI compute capex. Each generation of GPU clusters, including GB200 and successors, requires massive optical interconnect. The report relays Dell'Oro's view that mainstream speeds will migrate to 800G in 2025 and 1.6T in 2027, with optical module shipments rising with cluster scale.

  Second, the price jump comes from speed generations. This is a more powerful variable than in ordinary manufacturing. From 400G to 1.6T, per-rack optical interconnect value jumps from about USD 50,000 to USD 250,000–300,000, and 1.6T per-rack value is about 5–6 times that of 400G. The same number of ports can therefore produce multiple times the revenue after a generation upgrade. The report also mentions that the company is among the world's first manufacturers to mass-produce and deliver 1.6T optical modules, which means it can capture front-row benefits from this round of unit-price uplift.

  Third, new businesses such as silicon photonics, coherent products, and CPO are currently more like technical reserves for defending the main speed race, rather than an independent second revenue pole. The 2022 acquisition of Alpine added silicon photonics/coherent capabilities, and the direction is right, but in the short term it still serves the purpose of selling pluggable/LPO optical modules at higher speed and higher prices. It does not constitute a new growth source distinct from the core business.

  But long-term owners must apply three discounts. First, this is a pro-cyclical doubling from simultaneous volume and price gains, not a consumer-products style doubling through naturally compounding repeat purchases. Its premise is that AI capex does not fall materially after 2027, which the report lists as one of the largest uncertainties. Second, the unit-price jump is a double-edged sword: after each generation's upgrade dividend, every speed generation tends to face a price war. The report explicitly warns that 800G/1.6T competition will intensify after the prosperity peak, and prices and gross margins may fall. Volume can rise while price falls. Third, revenue doubling must be fed by working capital. 2026Q1 inventory had already risen to RMB 9.026 billion and accounts receivable to RMB 5.567 billion, according to the report, so growth is not light.

  Conclusion: the probability of revenue doubling over five years is high, and it is mainly driven by shipments plus the unit-price jump from speed upgrades, with limited contribution from new businesses. This is the most solid part of Eoptolink's growth profile. But it is a strongly cyclical doubling that follows customer capex and technology generations, not a locked-in subscription-style deterministic doubling. Whether the doubling turns into shareholder value will depend on gross margin and cash flow at that time, not just the revenue number itself.

  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

  Frankly, Eoptolink does not yet show a truly independent second curve today. Its work in silicon photonics, coherent products, LPO/LRO, CPO/NPO, and related areas is essentially speed extension and form-factor evolution along the same main curve of high-speed optical interconnect, not a new growth pole distinct from the optical module core business. This is precisely the feature of an excellent manufacturer rather than a platform company.

  Start with the reserves the company already holds. The report records that the company says it has built a technology system covering traditional pluggables, LPO/LRO, XPO, NPO, CPO, and other interconnect forms. It is one of the few global manufacturers with scaled mass-production and delivery capability for optical modules above 800G, and one of the world's first companies to mass-produce and deliver 1.6T optical modules. At OFC 2026, it also released a 1.6T DR4 solution based on IMDD 400G per-lambda. But note that all of this sits in the single basket of optical interconnect. These efforts answer how to build the next speed generation and how to reduce power consumption and cost, not what independent money-making engine the company has beyond selling optical modules. After 1.6T comes 3.2T, then CPO-style optical engines. This is the natural extension of the main curve, making the same business faster, not a second business.

  The only thing that may count as half a second curve is the value-chain reshaping caused by a shift in technology form. If CPO (co-packaged optics) becomes mainstream in the future and optical engines are moved next to switch chips, Eoptolink could move from selling pluggable modules to selling optical engines/optical subsystems, changing both value content and customer positioning. This is its opportunity to move upstream and toward deeper integration from a component supplier. Silicon photonics, strengthened through the 2022 acquisition of Alpine, is the technical foundation of this path. But the report is restrained on this point: in investor relations, the company itself only says it has already laid out CPO, with progress depending on market solutions and customer demand. In other words, the initiative for this potential second curve is not in the company's own hands; timing is determined by customers and industry standards. It is far from already existing and contributing revenue today.

  Under Baillie Gifford's LTGG standard, this is a relatively weak link in Eoptolink's growth narrative. The great growth stocks Baillie Gifford prefers often already have second and third independent curves while the core business is still strong, such as extending from hardware into software/services/platforms and forming a multi-engine relay. Eoptolink is currently a single engine with increasingly higher-powered pistons. The engine itself is powerful, but the baton is still the same one: it always depends on the next speed generation plus continued AI capex expansion. If AI optical interconnect demand peaks at some inflection point, the company does not have an obvious independent business that can step in and offset the core cycle. The report lists whether high order and delivery growth can continue after 2027 as one of the biggest uncertainties precisely because there is no second curve to smooth this cycle.

  There is also a reverse risk: a technology-form shift may be an opportunity, but it may also be disruption. If CPO/optical switching causes more of the optical value to converge into switch-chip vendors such as Broadcom or into large customers' in-house systems, the current main curve of pluggable optical modules will be eroded. Whether the company can then successfully transform into an optical engine/subsystem supplier remains uncertain. This is why the report lists the switching speed of CPO/NPO/optical-switching routes as a core technology substitution risk.

  Conclusion: the second curve does not exist today. What exists is multi-generation technical reserve along the main curve. The company has a potential path to upgrade from component supplier to optical engine/subsystem supplier, but that path is not yet contributing independent revenue and the initiative is not in its own hands. For long-term owners, this means Eoptolink's growth height depends on a single engine not stalling. That is the key weakness separating it from true ten-year five-bagger platform companies.

  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

  Eoptolink's core competitive advantage is a dynamic moat formed by the overlap of high-speed mass-production and delivery capability, certification positioning with top customers, and scale/process cost. But it is not a wide moat that deepens by itself and can be defended without running. The report gives moat strength only 3/5. Whether it widens or narrows over the next three to five years depends on one core variable: whether the pluggable architecture remains mainstream. The short to medium term looks stable, perhaps slightly wider; the long term is highly uncertain.

  First, what exactly is strong about the advantage? The report records that the company is one of the few global manufacturers with scaled mass-production and delivery capability for optical modules above 800G, and one of the world's first companies to mass-produce and deliver 1.6T optical modules. Capacity was running near full utilization in 2025. The ability to deliver next-generation speed products in the quantities, quality, and certification windows customers require is exactly the scarce resource cloud vendors value most. Scaled mass production of 800G/1.6T is not something a small factory can achieve immediately by spending money. Combined with its entry into the supply chains of mainstream global internet companies and telecom equipment vendors, according to the report, being on the list itself is a barrier.

  But the quality of the moat must be dissected honestly. The report's item-by-item characterization is useful: scale advantage is relatively obvious; switching cost is medium, because customer certification, introduction cycles, and reliability verification raise switching costs but not to the extremely high level of enterprise software; cost advantage exists but is not impregnable, because components and chip procurement remain constrained by upstream suppliers; brand advantage is medium to strong, reflected in quality/delivery/certification track record rather than consumer mindshare; network effect is almost absent; regulatory license barrier is weak; data advantage is weak. Add these together and the conclusion is clear: this is a moat dynamically maintained by technology + certification + delivery + scale + customer relationships. No single item is unshakable on its own; the strength comes from the combination and from continuous execution.

  Pricing power is the sharpest test of moat depth. The report's judgment is that the company has only periodic pricing power, not absolute pricing power. Gross margin rose significantly in 2024–2025 (2025 consolidated gross margin rose to 47.81%), but this looks more like bargaining power granted by product-generation upgrades plus strong supply and demand, not Coca-Cola-style end-brand pricing power. Peer comparison also shows this is an industry-wide dividend rather than something unique to the company: InnoLight's 2025 gross margin rose from 34.65% to 42.61%, so the entire track was lifting margins in a boom. Once supply and demand move from shortage to balance, this pricing power will recede as well.

  Will it widen or narrow over the next three to five years? Split the answer into two periods:

  Short to medium term, the next 1–2 years: stable, with a slight tendency to widen. As long as 800G/1.6T pluggables remain mainstream, the company's existing mass-production positioning, certification relationships, and first-mover position in 1.6T should further consolidate as shipments ramp, reinforcing scale advantage.

  Long term, 3–5 years and beyond: highly uncertain, and it may narrow. The report's core warning is technology-route switching. If CPO, NPO, optical switching, and related architecture changes move faster than expected and shift optical engines next to switch chips, the foundation of the moat, namely pluggable-module mass-production and delivery capability, will be changed. Or if overseas top customers reallocate procurement share, the moat can deteriorate from one that keeps deepening into a position that must be defended by constant running. In investor relations, the company itself only says it has already laid out CPO and that progress depends on customer demand, effectively acknowledging that the future of this moat is not entirely under its own control.

  Conclusion: the core advantages are real and scarce, namely high-speed mass production, top-customer certification, and scale. They are Eoptolink's hard foundation. But this is a dynamic moat of medium depth, with periodic pricing power and long-term erosion risk from architecture shifts. Over three to five years, the bias is short-term stability and long-term risk. This is also why the report gives the moat 3/5 and classifies the company as an excellent manufacturer rather than a wide-moat compounder.

  Jun 10, 2026
• QIf the core business is disrupted, does it have the DNA to reinvent itself? How does it handle mistakes and bad news?4/10

  A

  This is one of the hardest questions in Eoptolink's growth narrative to score highly. It has engineering iteration capability to keep catching up and avoid falling behind on the same technology track, but it lacks a verified gene for reinventing itself across boundaries after the core business is truly disrupted. Its record in handling mistakes and bad news is mixed: candid and pragmatic at the business level, but with one concrete governance blemish.

  First, the reinvention gene has two levels.

  The first is generational self-renewal within the same track, and the company has proved itself here. The report records that over many years, the company has continuously caught up with higher-speed products, moving from early speeds to 800G and becoming one of the world's first manufacturers to mass-produce and deliver 1.6T. In 2022, it also acquired Alpine Optoelectronics to add silicon photonics, coherent, and silicon photonics capabilities, proactively reserving ammunition for possible technology shifts. This shows that on the main line of continuously upgrading optical interconnect speed, the company has strong iteration capability and a sense of crisis. It is not trapped by the previous speed generation and can keep jumping to higher speeds. The report characterizes its culture and operating capability as relatively strong.

  The second level, which is what this question is really asking, is whether the company can move beyond its identity as a component supplier and reshape itself into another form if the core business, namely pluggable optical modules, is broadly disrupted by CPO/optical switching and related architectures. This has not been verified. The report is restrained: in investor relations, the company itself only says CPO has already been laid out and progress depends on market solutions and customer demand. This is equivalent to admitting that in the face of disruptive architecture, it is a follower and preparer, not a definer or reinventer. It has no second curve to offset the risk that the core business is disrupted, as discussed in this framework's judgment on the second curve. If the optical value chain is reshaped by switch-chip vendors or large customers' in-house systems, whether the company can successfully transform into an optical engine/subsystem supplier is currently an open question. The great growth stocks Baillie Gifford favors often have a record of disrupting themselves and changing how they live. Eoptolink is currently closer to a top student running very fast on a defined track, with a reinvention gene that has not yet faced a true life-or-death test.

  Next, how does it handle mistakes and bad news? This is a window into management character, and the company shows two sides:

  Business/accounting level: relatively candid, without concealment. The report notes that net realizable value of inventory has been listed as a key audit matter for two consecutive years, 2026Q1 asset impairment losses expanded YoY, and management proactively explained that the increase was due to provisions under the inventory write-down policy. The company also truthfully disclosed partial raw-material tightness in Q1 and expected easing from Q2. It did not evade accounting sensitivity and operating fluctuations under high growth, which is a positive. This point is also corroborated by a third party: in 2025, net realizable value of inventory was listed as a key audit matter for the second consecutive year.

  Governance level: one hard flaw that cannot be ignored. The report explicitly records that a 2025 administrative penalty decision by the China Securities Regulatory Commission showed that chairman Gao Guangrong transferred company shares through his own account and family trust account, involving violations of restrictive rules on share transfers and illegal disclosure of shareholding information, and the CSRC penalized him. Although this occurred at the level of personal share reduction and disclosure, not financial fraud, it directly tests management's attitude toward rules. A problem in precisely the place where rules most need to be followed creates a real discount to the item of candidly handling one's own mistakes. The report therefore gives management and capital allocation only 3/5 and states plainly that without the chairman's penalty, the score would be closer to 4.

  Conclusion: Eoptolink has strong renewal capability to keep iterating within the same track and not fall behind, but lacks a verified gene for cross-boundary reinvention after the core business is disrupted. It is relatively candid about operating and accounting bad news, but the governance-level administrative penalty is a clear character deduction. For long-term owners, this means one can trust it to keep making optical modules better, but cannot assume that it can gracefully transform when the paradigm is disrupted. That is precisely where it differs from great growth stocks with genuinely antifragile reinvention genes.

  Jun 10, 2026
• QDoes management, especially the founder, have a long-term view and deep alignment with the company? Is it willing to sacrifice current profit for five to ten years from now?5/10

  A

  Alignment is medium to high, but flawed. The founding team holds a meaningful stake and is broadly aligned with the company, operating execution is strong, and management is willing to make upfront capacity investments for the future. But long-term vision and character trust are materially discounted by the chairman's CSRC administrative penalty. The founder-level long-termism credit that Baillie Gifford values most is not perfect here.

  Start with the hard data on alignment. The report records that under the 2024 year-end/2025 first-quarter reporting basis, the company's actual controllers Gao Guangrong and Huang Xiaolei directly held about 14.5% in aggregate and acted in concert. This qualifies as deep alignment of interests with the company. It is not a professional-manager structure with single-digit ownership and reliance on excess compensation incentives. The founding team's wealth rises and falls with the share price. This is a positive and is the basis for the report giving management 3/5 rather than a lower score.

  Next, whether management is willing to sacrifice current profit for five to ten years from now. In behavior, the company is indeed making upfront investments for the future: 2025 capacity ran near full utilization, 2026 continues to expand capacity and improve delivery capability, according to the report, and in 2025 cash paid for the purchase and construction of fixed assets, intangible assets, and other long-term assets was about RMB 1.320 billion for capacity ramp-up. The 2022 acquisition of Alpine Optoelectronics added silicon photonics/coherent/silicon photonics capabilities, and the report judges that this was aligned with the company's later expansion into high-speed products and silicon photonics routes, with the right direction. R&D investment has also continued. These are actions to prepare early for the 1.6T and higher-speed era. Still, objectively, these investments occurred during a boom and amount to pro-cyclical expansion with the tide. They do not really show active sacrifice of current profitability for the long term, since current profitability is excellent and expansion has not depressed profit. The real test of long-termism is whether the company would still invest counter-cyclically when the industry turns cold. That has not yet been verified in a downcycle.

  Capital allocation is generally rational, but one signal deserves long-term owners' attention: the dividend pace has clearly increased. The report records that in 2024 the company proposed RMB 4.5 per 10 shares in cash dividends plus 4 bonus shares for every 10 shares, and in 2025 proposed RMB 10 per 10 shares in cash dividends plus 4 bonus shares for every 10 shares. For the 2025 profit distribution and 10-for-4 bonus issue, the record date was June 10, 2026, and the ex-rights/ex-dividend date was June 11. Raising dividends is shareholder-friendly in itself, but for a company that claims to still be in high growth and whose cash flow is heavily consumed by working capital, bonus-share issuance plus larger cash dividends may also be catering to market sentiment at a cyclical peak and enlarging the free float. This is not fully consistent with extreme long-termism that sacrifices the present for five to ten years from now and retains every penny for reinvestment. Baillie Gifford would prefer a founder who keeps cash inside the business and invests it into higher-return projects.

  The most important deduction is character trust. The report explicitly records that a 2025 administrative penalty decision by the CSRC showed that chairman Gao Guangrong had transferred company shares through his own account and family trust account, involving violations of restrictive rules on share transfers and illegal disclosure of shareholding information, and the CSRC imposed a penalty. This occurred at the level of personal share reduction and disclosure, not financial fraud, but it directly weakens the core question here: whether one fully trusts this founder's long-term vision and rule consciousness. The answer cannot be full marks. The report therefore states plainly that operating execution is strong and capital allocation is broadly rational, but governance character cannot be scored highly; without the chairman's penalty, the score would be closer to 4.

  Conclusion: management's alignment with the company is sufficient, with the actual controllers holding about 14.5% in aggregate and acting in concert. Execution is strong, and the directions of capacity expansion and M&A are right, providing substantive support for Eoptolink. But it lacks a verified record of sacrificing the present for the long term counter-cyclically, the stepped-up dividends/bonus shares carry a hint of catering to the market, and the chairman's administrative penalty leaves an unavoidable flaw in the founder long-termism credit that Baillie Gifford values most. Overall: alignment is qualified, but long-termism and character trust are discounted. Medium to high, not a high score.

  Jun 10, 2026
• QIf it disappeared tomorrow, how much would customers miss it? Is its growth model sustainable and not dependent on harming society or regulatory privilege?5/10

  A

  If Eoptolink disappeared tomorrow, a handful of leading cloud vendors and equipment makers would be quite uncomfortable in the short term, because only a few global manufacturers can mass-produce and deliver 800G/1.6T at scale, and a supply interruption would slow their AI cluster deployment cadence. But its indispensability is at the level of a key supplier in the supply chain that can be replaced, not a level where no one else can fill the gap. Its growth model itself is compliant and does not harm society, but extreme dependence on overseas large customers creates geopolitical and regulatory fragility rather than moral unsustainability.

  Start with the real weight of indispensability. The report records that the company is one of the few global manufacturers with scaled mass-production and delivery capability for optical modules above 800G, one of the world's first to mass-produce and deliver 1.6T, ran near full capacity in 2025, and has entered the supply chains of mainstream global internet companies and telecom equipment vendors. In the AI compute arms race, optical modules are a bottleneck component where missing one piece can prevent a rack from being fully populated. Next-generation speed capacity is even scarcer. So if it suddenly disappeared, customers relying on it would immediately face the practical question of who can replace that portion of 800G/1.6T capacity and how long recertification would take. They would certainly miss it in the short term; that is real.

  But the degree of missing it must be calibrated honestly. It is not the kind of indispensable company that customers cannot live without and no one can replace:

  Substitutes exist and are strong. InnoLight, in the same track, is larger: InnoLight's 2025 revenue was RMB 38.24 billion, and net profit attributable to the parent was RMB 10.797 billion, larger than Eoptolink. The report also lists comparable companies such as TFC Communication, Accelink Technologies, and Cambridge Technology. In other words, cloud vendors already have multiple qualified suppliers. The disappearance of a single supplier would create switching costs and a short-term capacity gap, but the industrial chain could rebalance within several quarters.

  Customers are sophisticated large buyers, not locked-in users. The report emphasizes that the company's switching cost is only medium. Large customers' refined procurement systems naturally follow multi-supplier strategies precisely to avoid irreversible dependence on any one party. This is the normal state of a B2B component business: customers miss you, but they are not unable to live without you.

  So the honest answer to how much customers would miss it is: high, but capped. It would be pain at the level of a key supplier outage, not paralysis from the disappearance of a platform or standard. Under Baillie Gifford's standard, truly great growth stocks are often those whose disappearance would leave a structural void for customers and society, such as a de facto standard or an unavoidable node. Eoptolink is an excellent but replaceable key supplier. This directly reflects the quality of its moat, rated 3/5 by the report, on the indispensability dimension.

  Next, whether growth is sustainable and whether it relies on harming society and regulation:

  Moral/social level: clean. Selling high-speed optical modules for AI compute interconnect in data centers is hard-tech manufacturing that meets genuine demand. It does not harm consumers or rely on regulatory arbitrage or negative externalities to make money. The report also points out that the company's regulatory/license barrier is weak, with main barriers coming from technology, process, and certification. This in turn shows that it does not grow through any privilege obtained by harming the public interest. This item is sustainable.

  Geopolitical/regulatory fragility: high, and a real risk. This is the real deduction in this question for Eoptolink. The report records that in 2025, the company's overseas revenue was about RMB 23.888 billion, accounting for as much as 96.16%; its top five customers accounted for about 72.34%, and the largest customer about 22.97%. This means its growth rests on extreme dependence on overseas, mainly North American, large customers. If international trade frictions, export controls, customer-location regulation, or geopolitical conditions change, growth sustainability will be directly hit. The report explicitly lists exchange-rate and trade-policy risk and customer concentration risk as core risks. In 2026Q1, finance expenses swung from RMB -0.033 billion in the same period last year to RMB 0.522 billion due to FX losses, according to the report, which was a preview of this fragility. This is not unsustainability from harming society; it is fragile sustainability rooted in an uncontrollable external environment.

  Conclusion: customers would genuinely miss it, because a supply interruption would slow AI deployment, but the degree is capped. It is an excellent yet replaceable key supplier, not an irreplaceable standard or node. The growth model itself is compliant and does not harm society. The real threat to sustainability comes from 96% overseas revenue plus highly concentrated customers, which creates geopolitical and regulatory fragility. From a long-term owner's perspective, the answer is important but not unique, compliant but not robust: medium indispensability, acceptable social sustainability, and fragile external-environment sustainability.

  Jun 10, 2026
• QWhat are the unit economics of this business, including gross margin and incremental returns? Do they improve or deteriorate with scale? Where does the earned money go?6/10

  A

  The unit economics look extremely attractive on the accounting statements: nearly 50% gross margin, nearly 40% net margin, and margins rising as scale expands. But on a real-cash basis they are materially discounted. Incremental revenue consumes a lot of working capital, and a considerable portion of the money earned is tied up in inventory, receivables, and prepayments rather than becoming free cash flow. This is the key duality of Eoptolink's unit economics: strong accounting profit, weak cash conversion.

  Start with accounting unit economics, which are indeed outstanding. The report and third parties are consistent: 2024 gross margin was about 44.72% and net margin about 32.82%; 2025 consolidated gross margin rose to 47.81%, with net margin about 38.35% (95.32/248.42); 2026Q1 gross margin further rose to about 49.16%, according to the Eastmoney source cited by the report. More importantly, the company shows a positive slope of greater scale and higher margins. This usually implies good operating leverage and incremental returns: a higher share of high-end speed products, Thailand capacity ramp-up, and process improvements allow each additional RMB of sales to carry a higher marginal profit rate than the existing base. On incremental returns, the company has almost no interest-bearing debt, with a 2026Q1 asset-liability ratio of about 31% and short- and long-term borrowing items empty according to the report, yet it can deliver nearly 40% net margin and extremely high returns on capital over the past two years. This is what the report means when it marks return on capital as a pass in the Checklist.

  But the quality of the accounting margin must be explained fully: such high margins cannot be treated as naturally perpetual. The report's judgment is important. Part of it comes from structural upgrade, including high-end product mix, capacity ramp-up, and process, while another part clearly carries cyclical dividends from the AI capex boom, high industry prosperity, and price increases caused by shortages of some raw materials. Linearly extrapolating nearly 50% gross margin for ten years is one of the most dangerous things in the current market, according to the report. Peer comparison also confirms this is a track-wide dividend rather than a company-exclusive one: InnoLight's gross margin rose from 34.65% to 42.61% over the same period, and the whole chain was lifting margins in the boom. Once supply and demand move from shortage to balance, gross margin will fall. The report lists gross margin falling below 35%–38% and failing to recover for two consecutive quarters as one of the signals that would invalidate the investment thesis.

  Do unit economics improve or deteriorate as scale expands? The honest answer is two-sided:

  Margin dimension: they improve with scale, through operating leverage + product mix upgrade, but this is pro-cyclical improvement and includes a dividend.

  Cash conversion dimension: they deteriorate with scale, which is the real soft spot in Eoptolink's unit economics. The report's core evidence is that operating cash flow coverage of net profit is highly volatile and low-leaning: OCF/net profit attributable to the parent was only about 22.6% in 2024, recovered to about 80.8% in 2025, and fell again to about 24.6% in 2026Q1. The reason is that growth consumes working capital. Accounts receivable rose from about RMB 2.539 billion at 2024 year-end and inventory of about RMB 4.132 billion to RMB 4.438 billion and RMB 7.234 billion at 2025 year-end, then to RMB 5.567 billion and RMB 9.026 billion in 2026Q1. The report puts it sharply: for the same RMB 10 billion in profit, a company that needs little additional inventory and receivables is far more valuable than one that must keep feeding growth with more inventory and receivables, and Eoptolink is clearly closer to the latter.

  This directly determines where the earned money goes and how much is left. The report records that in 2025, net profit attributable to the parent was about RMB 9.532 billion, net operating cash flow about RMB 7.701 billion, and cash paid for purchasing and constructing fixed assets, intangible assets, and other long-term assets about RMB 1.320 billion for capacity ramp-up. On this basis, strict free cash flow was about RMB 6.38 billion. In other words, accounting profit was RMB 9.5 billion, but ① about RMB 1.8 billion of the gap was swallowed by working capital and related items, because net profit did not convert into equal operating cash flow, and ② another RMB 1.3 billion was spent on capacity investment, leaving truly free cash of about RMB 6.4 billion. The report further makes a conservative Owner Earnings estimate, treating about RMB 0.8 billion as maintenance capex, giving owner earnings of about RMB 6.9 billion. The conclusion is that accounting profit has moved far ahead of distributable cash. This is also why the report repeatedly stresses that on 2025 FCF, the free cash flow yield is only about 1.1%. Under the current higher market-cap basis, this yield would be even lower. The money earned mainly goes to feeding working capital, including inventory, receivables/prepayments, expanding capacity, and raising dividends, with the 2025 proposal of RMB 10 per 10 shares in cash dividends plus 4 bonus shares per 10 shares.

  Conclusion: the unit economics are a machine with strong accounting profitability but cash conversion efficiency dragged by cycles and working capital. Gross margin/net margin rise with scale, though with cyclical dividends and no basis for perpetual extrapolation; cash conversion comes under pressure as scale grows, and incremental revenue consumes large amounts of working capital. For long-term owners, this means one cannot just look at the beautiful income statement. What truly determines value is how much of the money earned becomes free cash distributable to shareholders, and Eoptolink is currently clearly weak on this point.

  Jun 10, 2026
• QWhat conditions must hold simultaneously for it to rise fivefold over ten years? Are these conditions realistic? What expectations are embedded in today's share price?2/10

  A

  For Eoptolink to rise another fivefold from today's price, a set of conditions must all hold at the same time, and every one of them is optimistic. The realism is low. Today's share price has already bought in most of the optimistic scenario calculated in the report, and even stronger assumptions. It is not an entry ticket with a margin of safety, but a high-priced ticket betting that the super-cycle does not stall.

  First, clarify today's price anchor. This is crucial because the share price has risen sharply relative to the report snapshot. The report uses the May 21, 2026 baseline of RMB 567.11/share, total market cap about RMB 563.7 billion, and TTM PE about 52.49x. But as of the June 10, 2026 close, the share price had risen to about RMB 772.5, with the previous trading day closing at about RMB 785.73, total market cap about RMB 767.5 billion, and PE about 71.7x (cn.investing.com, 2026-06-10), up about 36% from the report baseline. In other words, the report already judged that there was no margin of safety, and the current price is more than 30% higher than then, with valuation further re-rated above 70x. Note: the record date for the 10-for-4 bonus issue was June 10 and the ex-rights date June 11. After ex-rights, share count increased from about 994 million shares to about 1.392 billion shares, and the per-share price was correspondingly divided by 1.4, but the total market-cap basis was unchanged. The above comparison is on the pre-ex-rights basis.

  What conditions must hold simultaneously for a ten-year fivefold rise? Expanding the current market cap of about RMB 767.5 billion fivefold gives about RMB 3.8 trillion. To support this scale, the following items must almost all be realized:

  First, the super-cycle in AI optical interconnect must continue beyond 2027 and not peak midway. The report lists whether high order and delivery growth can continue after 2027 as the top uncertainty. This is the main premise; if the capex cycle rolls over, everything that follows fails.

  Second, the company must keep winning mainstream verified share in 1.6T, 3.2T, and even the CPO era, rather than falling behind during architecture shifts. The report warns that CPO/NPO/optical-switching route shifts may erode the moat of pluggable modules, and the company itself only says CPO has already been laid out.

  Third, gross margin must remain above 45% after several years of volume and price gains, and cash flow must continue to approach profit. This means the prosperity dividend does not recede and working capital no longer swallows it. The report explicitly lists maintaining gross margin above 45% with cash flow continuously close to profit as one of the conditions under which it would admit it had been too conservative, effectively acknowledging this is a high bar.

  Fourth, customer concentration must decline rather than rise, and the geopolitical risk from overseas dependence must not erupt. Current overseas revenue accounts for about 96.16%, and the top five customers account for about 72.34%, according to the report. This is fragility rather than certainty.

  Fifth, the valuation multiple must not contract sharply. Even if all the above are realized and profit truly grows fivefold, if PE falls from 70x back to the 20–30x range of mature manufacturing, the share price still cannot rise fivefold. This is the hidden killer of a reverse Davis double play.

  Are these conditions realistic? Each condition individually has a reasonable probability, but for all of them to hold simultaneously and continuously for ten years is a low-probability joint event. The report's three-scenario Owner Earnings discount model gives a hard constraint. Starting from about RMB 7.0 billion in conservative Owner Earnings, the conservative scenario corresponds to about RMB 139/share, the neutral scenario about RMB 221/share, and the optimistic scenario about RMB 327/share, with estimated equity values of about RMB 138.4/219.5/324.8 billion. Note that even the optimistic scenario's about RMB 327/share is materially below the report's then RMB 567/share and far below today's about RMB 772.5/share. This means:

  What expectations are embedded in today's share price? The market has already priced in assumptions more optimistic than the report's optimistic scenario. Based on the current market cap of about RMB 767.5 billion and 2025 free cash flow of about RMB 6.4 billion, the FCF yield is only about 0.83%. The report calculated about 1.1% at the lower RMB 563.7 billion market cap, and it is lower now. At about 71x TTM PE, the earnings yield is only about 1.4%. Against the report's reference set, the CSI 300 PE is about 14.40x with an earnings yield of about 6.9%, and China's 10-year government bond yield is about 1.76%. Eoptolink's earnings yield is already below the risk-free government bond yield, and the FCF yield is far below the bond yield. Put differently, today's price embeds a perfect script in which high prosperity is almost uninterrupted for the next ten years, gross margin does not fall, share is not lost, and the market is always willing to grant a high valuation. This directly reflects the report's conclusion that the current price has no margin of safety and the market has at least priced in most of the optimistic scenario, or even stronger assumptions. Because the share price has risen another 36% since the report snapshot, this conclusion is even more valid now than it was then.

  Conclusion: a ten-year fivefold rise requires a long list of conditions to be met simultaneously: super-cycle continuation + sustained share leadership + high gross margin not receding + cash flow catching up + valuation not contracting. This is a low-probability joint event. Today's price of about RMB 772.5/about RMB 767.5 billion market cap has already priced expectations above the report's optimistic scenario, with FCF yield below 1% and earnings yield below government bonds. From a long-term owner's perspective, this is not a favorable-odds starting point. Upside has been heavily pre-spent by high expectations, leaving almost no margin of safety for error.

  Jun 10, 2026
• QWhy has the market not realized all this yet? Is it failing to understand, dismissing it, or not looking far enough? What would become the narrative inflection point?3/10

  A

  For Eoptolink, this question must be reversed. Baillie Gifford's "why has the market not realized it" assumes a good company that is undervalued and overlooked. Eoptolink is the opposite: the market has not only fully recognized its growth, but has also bought it up to a TTM PE of about 71x and to a level about 36% higher than the report baseline from just over one month earlier. The real perception gap is therefore not whether the market understands how good the company is, but whether the market has fully priced its cyclicality, cash-flow weakness, and valuation risk. In other words, what is overlooked is not the upside, but the downside.

  Use price first to show that the market has already understood and embraced it. The report used RMB 567.11 on 2026-05-21, a market cap of about RMB 563.7 billion, and TTM PE of about 52.49x as its baseline, and already concluded that there was no margin of safety. By the June 10, 2026 close, the share price was about RMB 772.5, with the previous close about RMB 785.73, market cap about RMB 767.5 billion, and PE about 71.7x (cn.investing.com, 2026-06-10). A stock that the market fails to understand, dismisses, or does not look far enough into would not rise another 30% in a month and be re-rated above 70x. Sell-side expectations are also fully stretched. The report cites Founder Securities' expected 2025–2027 revenue of about RMB 25.8/52.6/72.4 billion and Guolian Minsheng's about RMB 26.9/47.965/68.151 billion. The market sees the main line of high AI optical interconnect prosperity + 1.6T volume ramp very clearly and has embraced it enthusiastically. So a Baillie Gifford-style positive perception gap basically does not exist here.

  Then what might the market not look far enough into, or be unwilling to look at? This is the real reverse perception gap:

  First, it may treat cyclical dividends as structurally perpetual. The report repeatedly stresses that the nearly 50% gross margin in 2024–2026 comes partly from structural upgrade and partly, clearly, from AI capex explosion + raw-material shortages, and that linearly extrapolating it for ten years is one of the most dangerous things in the current market. Current pricing embeds the assumption that high gross margin does not recede. That is where the market may be too optimistic.

  Second, it may focus on the income statement and ignore the real quality of cash flow. Accounting profit is extremely bright, with 2025 net profit attributable to the parent of RMB 9.532 billion, but operating cash flow coverage of net profit is highly volatile, at about 22.6% in 2024, about 80.8% in 2025, and about 24.6% in 2026Q1 according to the report. Growth is heavily occupied by inventory, which rose to RMB 9.026 billion in 2026Q1, and receivables of RMB 5.567 billion. Based on the current market cap of about RMB 767.5 billion and 2025 free cash flow of about RMB 6.4 billion, FCF yield is only about 0.83%. The market has paid for high growth in accounting profit, but may be underestimating how poor the efficiency is when these profits are converted into shareholder cash.

  Third, it may underprice the risk of valuation multiple compression. Even if fundamentals remain excellent, PE moving from 70x toward 25–30x is enough by itself to cause a large drawdown. The report separately lists valuation risk. In a boom, markets often price only the upside scenario and leave no discount for a reverse double hit.

  Now to the conclusion of this question: what would become the narrative inflection point? For a stock that has fully priced the optimistic scenario, the inflection point is more likely to be the type that breaks the narrative downward. The report has effectively listed the checklist, which can be summarized as:

  Order/prosperity inflection: AI optical interconnect demand moves from shortage to balance around 2027, and North American cloud-vendor capex cadence slows. This is the single largest inflection point, listed by the report as the top uncertainty.

  Gross-margin inflection: gross margin stays below 38% and fails to recover for more than two consecutive quarters, which the report explicitly names as an investment-thesis invalidation signal.

  Cash-flow inflection: operating cash flow remains materially below net profit for a long period, while receivables and inventory continue to grow faster than revenue, proving that cash-consuming growth is structural rather than one-off.

  Technology/share inflection: the company loses mainstream verified position during the 1.6T/CPO generation transition, or top customers reallocate share and orders from the largest customer, about 22.97%, weaken.

  Governance/valuation inflection: another major governance flaw appears, given the chairman's prior CSRC administrative penalty, or a fall in market risk appetite triggers broad valuation compression across high-multiple optical module stocks.

  Conversely, for an upward narrative inflection point that proves today's high price reasonable, the report gives the conditions: after 1.6T and higher-speed products scale, the company still maintains gross margin above 45% and cash flow continuously close to profit; customer concentration declines; AI demand does not fall after 2027; and the company continues to win mainstream share in the CPO era. If these three items all hold simultaneously, high expectations may be truly digested by performance. But as discussed above, this is a low-probability joint event.

  Conclusion: the market has not failed to realize Eoptolink's strengths. It has realized them fully and has expressed that enthusiasm with a 70x PE and another 36% price rise. What is relatively overlooked is the company's cyclicality, cash-flow weakness, and valuation fragility. Therefore, for Eoptolink, the greater concern is not an upward inflection where good news is discovered, but a downward inflection where prosperity, margin, cash flow, or share is disproved. At the current price, narrative risk is asymmetrically tilted to the downside.

  Jun 10, 2026