Zscaler: A Long-Term Owner's Perspective

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

• QHow large is its market ceiling? Is it expanding an existing pie, or creating an entirely new market?6/10

  A

  It is mainly expanding and reshaping an existing pie that is already growing quickly, as enterprise cybersecurity budgets migrate from legacy firewalls/VPNs to cloud-based zero trust; it is not creating an entirely new market from scratch. The ceiling is high enough for the next decade, but it is not unlimited. When Baillie Gifford asks this about LTGG candidates, the real question is whether the market can support a 5x outcome over ten years. For Zscaler, the runway is long, but the opportunity is one of replacement and redistribution, not one where demand is being ignited from zero.

  Start with the size of the market. The report cites Gartner's SASE market forecast: about USD 7 billion in 2022, rising to about USD 25 billion by 2027, a CAGR of about 29%. That is only the narrow SASE definition. Zscaler itself frames its addressable market, combining secure internet access, private application access, cloud workloads, data security, AI/Agentic security, and other areas, as a market worth several tens of billions to over USD 100 billion. Even under the most conservative SASE definition, a USD 25 billion TAM still leaves ample penetration room for a company with FY2025 revenue of USD 2.673 billion and latest ARR of USD 3.525 billion, as of 2026-04-30, up +25% year over year. Its penetration is still in the single digits to low teens, so the runway is clearly not exhausted.

  But it is important to distinguish honestly between expanding a pie and creating one. Enterprises already spend money on cybersecurity by buying firewalls, VPNs, gateways, and SWGs. Zscaler is moving this already existing budget from on-premise hardware boxes to a cloud-native zero trust platform. In substance, this is architecture replacement, budget migration, and a way to thicken spending per customer. That differs from inventing a category where nobody previously spent money. The advantage is that demand is highly certain and the company does not need to educate the market from zero. The cost is that many strong competitors are also chasing the same pie, and share is redistributed rather than exclusively created.

  The part that comes closest to creating a new market is the incremental demand emerging in the AI era: the AI/Agentic security theme emphasized repeatedly in the report. In the latest earnings call, management disclosed that AI Protect had already exceeded USD 100 million in orders within 12 months of its January launch, and data security ARR had surpassed USD 500 million, up more than +30% year over year. CEO Jay Chaudhry positions the company as a cybersecurity platform for the AI era. This does have new-pie characteristics, but it remains a small share today. It is more a way to extend the old runway and raise the ceiling than an independent large market that already supports the valuation.

  Ceiling judgment: high and credible, but not in the unlimited-imagination category. It can support many years of mid-to-high-teens growth, which is also the report's key assumption that ARR growth remains in the mid-to-high teens. It is harder to tell a blue-sky story of a 10x TAM captured exclusively, because this is an expanding market shared by Palo Alto, Netskope, Fortinet, and Cloudflare. Conclusion: Zscaler is expanding a fast-growing existing pie, with a small layer of AI-driven incremental demand. The market ceiling is not a constraint for its current scale of about USD 124.73 per share and about USD 20.2 billion market cap, but this is not the kind of brand-new, unoccupied territory that Baillie Gifford most loves.

  Jun 11, 2026
• QCan its revenue at least double over the next five years? Will growth mainly be driven by volume, price, or new businesses?6/10

  A

  Most likely yes: starting from latest annualized revenue of about USD 3.3 billion, doubling in five years requires only about a 15% CAGR, while the company's current actual growth is clearly higher. The main drivers, however, are existing-customer expansion, mainly volume with some price, plus new product scaling, not simple price increases or pure new-logo growth. Baillie Gifford puts heavy weight on the hard hurdle of at least doubling in five years. Zscaler clears that hurdle and has a cushion.

  Start with the math. In the latest quarter, FY2026 Q3 revenue was USD 850.5 million, up +25% year over year, and the company guided FY2026 full-year revenue to about USD 3.330 billion, up about 24.6–24.7%. Doubling in five years requires a CAGR of about 14.9%. In other words, even if growth decelerates from today's 25% to an average in the teens, a five-year double still holds. It fails only if growth falls below the mid-teens and keeps deteriorating, which is exactly the first reassessment trigger listed in the report. Forward indicators look more optimistic: ARR was USD 3.525 billion, up +25% year over year, and RPO was USD 6.459 billion, up about +30% year over year. RPO is the reservoir of contracted but unrecognized revenue; growth above revenue growth means the next few years of revenue have substantial visible support.

  The driver mix is the crux of this question:

  First, volume, meaning customer expansion, is the main engine. Most of Zscaler's growth comes from existing customers buying more modules and covering more users and traffic. The evidence is a net revenue retention rate of about 115%, stable over several quarters. Existing customers alone naturally expand by about 15% net each year, and new customers add to that to produce mid-to-high-teens growth. Large-customer tiers are also thickening: 748 customers had more than USD 1 million in ARR, up +18%, and 4,000 customers had more than USD 100,000 in ARR, up +19%. The report also lists steady expansion in the number of customers with more than USD 1 million in ARR as a key assumption.

  Second, new businesses, meaning product breadth, are the second push. Zscaler has expanded from the original ZIA/ZPA into data security, AI security, browser security, and MDR. Data security ARR has already exceeded USD 500 million, up more than +30%, and AI Protect exceeded USD 100 million in orders within 12 months. These products increase wallet share per customer and are the key to offsetting natural deceleration in the old business.

  Third, price, meaning pure unit-price increases, contributes the least. The report explicitly judges Zscaler's pricing power as showing up more through bundling, expansion, suite upgrades, and total cost of ownership savings than through simple large annual price hikes. This point needs to be stated plainly: raising prices is not the main revenue engine.

  The risk that deserves a deduction: management disclosed the departure of two sales leaders at the end of Q3 and adopted a prudent/tempered stance on FY2027 new-logo contribution. That is one reason the stock has fallen from about USD 180 at the time of the report to about USD 124.73 now. If new-customer acquisition keeps slowing and growth relies excessively on existing-customer expansion, the quality of long-term growth will be discounted. But for the binary question of whether revenue can double in five years: yes. Current growth leaves a margin of safety, and the main drivers are volume expansion plus new product scaling.

  Jun 11, 2026
• QFive years from now, what will take over as the next growth engine? Does this second curve exist today?5/10

  A

  The second curve is already emerging in the numbers, not just in a slide deck. It is the set of new engines around AI/Agentic security, data security, and browser security/MDR, which are the most likely candidates to take over from the core secure internet access/private application access businesses five years from now. But today it is still an early second curve: small in scale, partly dependent on acquisitions, and not yet proven as an independent growth engine. Baillie Gifford asks about the second curve to see whether there is another handoff when the main curve slows. Zscaler's answer is that candidates exist and have already become revenue, but the quality of the handoff remains to be verified.

  Start with the first curve. Zscaler's core revenue engine remains secure internet access, ZIA, and private application access, ZPA, around the Zero Trust Exchange. This is its cash cow and growth base. The Baillie Gifford question is: when ZIA/ZPA penetration matures and growth naturally slows, what fills the gap?

  The handoff candidates have moved from narrative to businesses with numbers, which is the key good news:

  First, data security. The latest earnings call disclosed that data security ARR had exceeded USD 500 million, growing more than +30% year over year. That is clearly faster than the overall 25%, meaning it is outperforming the average and has the potential to scale.

  Second, AI/Agentic security. This is the direction on which the CEO is placing the heaviest bet. He positions the company as a cybersecurity platform for the AI era, and says frontier models are discovering vulnerabilities at machine speed. The AI Protect product was launched this January and has already generated more than USD 100 million in cumulative orders over 12 months. The base is still small, but the slope from 0 to USD 100 million is evidence that the second curve has actually lit up.

  Third, Zero Trust Everywhere full-stack penetration plus browser/MDR. Zscaler is extending zero trust from internet access into branches, factories, IoT, and endpoint browsers. More than 700 enterprise customers have adopted Zero Trust Everywhere, up sequentially from 550 in Q2. This is a lever for increasing wallet share per customer and extending the life of the main curve.

  Expectations need to be held down honestly:

  First, the scale is still small. USD 500 million in data security and USD 100 million in AI Protect are limited relative to USD 3.525 billion in ARR. They do not yet justify the word handoff. They look more like incremental extensions of the main curve. If Baillie Gifford strictly distinguishes an independent new growth pole from a core-business extension, these lean toward the latter.

  Second, a large part is bought rather than built. The report notes that in FY2026 the company used about USD 692 million in total for Red Canary + SPLXAI, plus SquareX, to strengthen these new capabilities, and goodwill rose from USD 418 million to about USD 1.094 billion. An acquired second curve carries higher integration risk and makes organic quality harder to judge. The ARR growth figure also needs to strip out acquisition contribution: ARR was up +25% year over year, but +21% excluding Red Canary.

  Third, the investment required for the second curve is pressuring cash flow. To support AI security, the company lowered FY2026 free cash flow margin guidance from about 26.5–27% to 22.8–23.3%, and FY2027 capital expenditure as a share of revenue may rise by up to another 200 basis points. This is the direct catalyst behind the stock's decline from about USD 180 at the time of the report to about USD 124.73 now. The second curve is real, but it is currently in the spending phase. Payback will take time.

  Overall: the second curve does exist today and is already a revenue-generating reality. The direction, AI security, is also right, making it stronger than many growth companies with only a story. But it is still early, small, acquisition-reliant, and cash-consuming. Whether it truly takes over depends on whether its organic growth can keep outpacing the core business over the next few years and whether AI security becomes a large independent engine.

  Jun 11, 2026
• QWhat is its core competitive advantage? Will this moat widen or narrow over the next three to five years?6/10

  A

  The core advantage is a composite moat built from cloud-native zero trust architecture, global distributed scale, switching costs embedded deep in customer traffic, government compliance credentials, and co-selling channels, not any single feature. Over the next three to five years, I expect it to broadly hold, with some marginal narrowing risk. The moat is real, but competitors are using acquisitions to copy the whole platform narrative, shifting Zscaler from structural leadership into an execution contest. This is consistent with the report's conclusion that the moat is real but not overwhelming. On Baillie Gifford's question of whether the moat widens or narrows, I lean toward narrow fluctuation and defense rather than continuous widening.

  Start with the substance of the advantage. Zscaler's killer feature is not a specific function but architectural difference. In the latest call, the CEO repeatedly emphasized that it is not feature A or feature B, but a difference in architecture: a zero trust model where users never connect directly to applications and all access goes through a cloud proxy. This is inherently different from the traditional vendor model of patching firewall boxes. Scale supports that architecture. The report cites company disclosures of more than 160 public exchanges, more than 500 billion requests processed per day, and 130TB of raw logs. This global node footprint and data scale require years and heavy capital to replicate.

  The hardest part of the moat is switching cost. Once the platform is embedded into a customer's traffic, identity, policies, and application access paths, migration is extremely costly. This is the underlying reason net revenue retention remains stable at about 115% and customers with more than USD 1 million in ARR increased to 748, up +18%: once large customers get on the platform, they keep buying more and rarely go back. Add FedRAMP/DoD IL5/CMMC credentials, which are tickets into government and regulated large customers, plus co-selling channels with telecom operators and systems integrators, and you get the composite barrier described in the report: architecture + scale + switching costs + compliance + sales.

  Why I do not think it will keep widening, and why there is even narrowing risk, is the key part of an honest assessment:

  First, competitors are using acquisitions to reproduce the platform narrative at full-system level. Palo Alto, named in the report, has completed its roughly USD 25 billion acquisition of CyberArk, identity security, in February 2026, and has also acquired Chronosphere, about USD 3.35 billion, Protect AI, about USD 700 million, and Talon browser security, about USD 625 million. This means the target is not one Zscaler feature, but a broader integrated security stack. As customers increasingly prefer one vendor to bundle more of the job, the edge of pure zero trust is diluted by a fuller platform.

  Second, new capital is entering the market, and share is still being redistributed. Netskope raised more than USD 900 million in its September 2025 IPO, and Fortinet continues to expand with self-developed ASICs and an integrated architecture. The report's judgment that the market has not formed a winner-take-all profit pool remains valid. The moat can prevent customer churn, but it cannot prevent competitors from taking incremental budget.

  Third, Zscaler itself is also buying to widen the moat. It used acquisitions such as Red Canary/SPLXAI/SquareX to add AI security and browser security to the platform. This strengthens the company, but also indicates that the pure organic moat is not enough by itself to cover the new battlefield. As it expands, it takes on more integration risk. In the report's phrasing, it is moving from a clean core architecture company to a complex platform integrator.

  Three-to-five-year judgment: switching costs and compliance, the two hardest moats, will continue to hold. Customers remain sticky, and the government credentials remain valuable. But the relative advantage of architecture leadership will narrow at the margin as competitors platformize. The moat will show up more as defending the installed base and winning incremental business through execution, rather than collecting monopoly rents. That is why the report's moat score of 4/5, real but not overwhelming, is reasonable. It is good, but not an exclusive barrier that will only get wider over the next three to five years.

  Jun 11, 2026
• QIf the core business is disrupted, does it have the DNA for self-reinvention? How does it treat mistakes and bad news?5/10

  A

  There are above-average signs of self-reinvention: founder leadership, a bet on the self-disruptive new curve of AI security, and a recent quarter in which management chose to proactively disclose bad news, including sales-executive departures and a cash-flow guidance cut, and reset expectations rather than gloss over it. But Zscaler has not yet gone through a true life-or-death moment where the core is disrupted and the whole company must reinvent itself. Its reinvention DNA has been proven more by offensive expansion than by a turnaround from crisis. Baillie Gifford asks this to know whether, if the main channel is overturned, the company will leave its comfort zone and face bad news honestly. Zscaler has stronger evidence on proactive evolution and weaker evidence on reinvention under crisis.

  Start with the DNA for reinvention. The most persuasive evidence is that Zscaler has not sat on the old ZIA/ZPA rent stream. It has actively pushed into a direction that could disrupt itself: AI security. In the latest call, CEO Jay Chaudhry treated AI as both opportunity and threat, stating explicitly that AI is changing the nature of cybersecurity in real time, frontier models are discovering vulnerabilities at machine speed, and the company is being repositioned as a cybersecurity platform for the AI era. The bet is backed by real money: AI Protect exceeded USD 100 million in orders over 12 months, and the report records acquisitions such as Red Canary/SPLXAI/SquareX that bring browser security and AI security into the platform. A company willing to fight the next battle while its current business is still highly profitable has the will to reinvent.

  Now look at how it treats mistakes and bad news. This is the part of the question that best reveals corporate character, and the most recent quarter provided a real stress test:

  First, management chose to put bad news on the table rather than hide it. Two sales leaders left at the end of Q3. Management proactively disclosed this on the call and explained that this was why it adopted a prudent / tempered stance on FY2027 new logos. It explained the personnel disruption and the impact on guidance, instead of waving it away with a claim that everything was normal. That is evidence of facing bad news honestly.

  Second, management was willing to take the hit to the share price and lower expectations to reflect reality. The CFO cut FY2026 free cash flow margin guidance from about 26.5–27% to 22.8–23.3%, acknowledging higher capital expenditure for AI data centers, including memory, storage, and processors. This cut helped drive the stock from about USD 180 at the time of the report to about USD 124.73, roughly 60% below its 52-week high of 336.99. Management knew the market would punish it but still revised guidance honestly rather than forcing attractive numbers. This matches the report's favorable view that management's balance between growth and profitability is not out of control and that it emphasizes disciplined execution. Sacrificing the look of current cash flow to position for AI five to ten years out also supports the long-term perspective discussed in Q6833.

  The deductions also need to be stated honestly:

  First, it has not experienced a true core-business disruption. Zscaler's reinvention so far has been proactive expansion in favorable conditions. It has not suffered the kind of core-business collapse that forces a full transformation. Its reinvention DNA has not been tested under fire, so its ability to rescue itself from crisis remains unknown in the Baillie Gifford sense.

  Second, the governance structure weakens checks and balances for error correction. The report notes that the company has a combined chairman and CEO, no lead independent director, a classified board, and public opposition to declassification. A strong founder can help the company turn quickly, but it also means self-correction depends heavily on one person's judgment and external checks are weak. If the founder is wrong one day, reinvention may be slower.

  Overall: on willingness to evolve proactively and tell the truth about bad news, Zscaler scores well, thanks to its active bet on a self-disruptive direction and its honest recent reset around personnel and cash flow. On the question of whether it can reinvent itself under existential pressure, the evidence is blank. For long-term holders, the picture is an ambitious and candid company whose correction mechanism depends heavily on the founder and has not yet been truly tested.

  Jun 11, 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 the next five to ten years?7/10

  A

  Yes, and the alignment is deep: founder Jay Chaudhry has served as both CEO and chairman since 2007, owns about 17% of the company, and takes almost no cash compensation. On the dimension of economic alignment and long-term perspective, this is top tier. He has also just proved willingness to sacrifice current profit for a five-to-ten-year position by actively cutting cash-flow guidance to invest in AI, despite the hit to the stock. The one deduction is that the governance structure is not especially friendly to minority shareholders. This is the core tension behind the report's management score of 3/5: operating credibility, but governance discount. In this Baillie Gifford question, the people are a strength, while the governance framework is the weakness.

  Start with alignment and long-term orientation. The evidence is hard:

  First, the founder remains in charge and owns a large stake. Chaudhry has been CEO and chairman since founding the company. The report cites the 2025 proxy statement showing his stake at about 16.8%, while public figures show about 17.26% direct ownership and about 38% total voting power, worth roughly USD 8 billion. This means his personal wealth is almost entirely tied to the same share-price outcome as common shareholders. This is exactly the kind of deep company alignment Baillie Gifford values.

  Second, he takes almost no cash and earns mainly through equity value creation. The report records that in FY2025 his base salary was only USD 23,660 and that he did not participate in the annual cash bonus plan. The company also has CEO share ownership guidelines requiring shares worth at least 5x annual salary. A founder whose compensation is almost entirely tied to the stock price is naturally incentivized to run for long-term value rather than short-term accounting profit.

  Third, willingness to sacrifice current profit for the next five to ten years has just been tested in reality. This is the latest and most important evidence for this question. To position for the AI era, management proactively lowered FY2026 free cash flow margin guidance from about 26.5–27% to 22.8–23.3%, and said FY2027 capital expenditure as a share of revenue may rise by up to another 200 basis points, citing AI data-center memory, storage, and processor costs. This decision directly drove the stock from about USD 180 at the time of the report to about USD 124.73. Management knew the stock would be hit hard and still chose to spend more now to secure a place in AI security over the next few years. That is an affirmative answer to Baillie Gifford's question about sacrificing current profit for five to ten years out. Historically, the company has also poured money into R&D and sales, with the report showing FY2025 R&D of USD 672 million and sales and marketing of USD 1.259 billion, accepting long-running GAAP losses. The operating style has consistently prioritized long-term investment.

  The governance weaknesses must be called out because they are the source of the report's discount:

  First, power is highly concentrated and checks are weak. The company has a combined chairman and CEO, no lead independent director, a classified board, and public opposition to board declassification proposals. This structure helps a founder enforce a long-term will, but it also weakens outside shareholders' ability to correct him. For conservative investors, this is a real governance discount.

  Second, dilution control is weak and not especially friendly to minority shareholders. The report notes that shares outstanding rose from 138.7 million shares in July 2021 to about 161.7 million shares recently, with no systematic buyback offset for continuing stock-based compensation dilution. The founder's long-term alignment is real, but the company has not done enough to protect per-share shareholders from dilution.

  Overall: on whether the founder has a long-term view, deep alignment, and willingness to sacrifice the present for the future, Zscaler has near-full-score evidence: founder still in charge, large ownership, almost no cash pay, and a fresh decision to sacrifice cash-flow guidance for AI. The governance framework, including concentrated power and weak dilution control, is what pulls the score down. This is a management profile where the person is right and the system has flaws.

  Jun 11, 2026
• QIf it disappeared tomorrow, how much would customers miss it? Is its growth model sustainable and not dependent on social harm or regulatory arbitrage?6/10

  A

  Customers would miss it a lot, and in the short term it would be hard to replace seamlessly. Zscaler guards critical traffic paths for enterprise internet access, private application access, identity, and policy enforcement. It is infrastructure embedded in the operating skeleton, not a nice-to-have tool. Its growth model is also highly healthy: it depends almost not at all on social harm or regulatory arbitrage, and instead helps enterprises and governments raise their security and compliance standards. Baillie Gifford's question tests two things: indispensability, or how painful its disappearance would be, and the social/regulatory sustainability of growth. Zscaler scores high on both.

  Start with indispensability. Zscaler is not decorative SaaS. It is a security platform sitting on the main artery of enterprise traffic. The report cites company disclosure that it processes more than 500 billion requests and 130TB of raw logs per day, carrying customers' user internet access, private application access, identity, and policy enforcement. If it disappeared tomorrow, the problem would not be losing a feature. It would be questions like how employees securely access applications, how traffic is filtered, and how zero trust policies are enforced. Migrating back to legacy firewall/VPN architecture would take months and materially increase security exposure in the meantime.

  This level of being missed is objectively supported by data:

  First, retention and expansion are extremely sticky. Net revenue retention is stable at about 115%. Customers not only stay; they buy about 15% more on a net basis each year. Customers with more than USD 1 million in ARR reached 748, up +18%, and customers with more than USD 100,000 reached 4,000, up +19%. Large customers keep embedding deeper, which is market evidence that they cannot easily leave.

  Second, it has penetrated critical institutions. The report discloses that about 40% of the Forbes Global 2000 and more than 45% of the Fortune 500 are customers, and it holds FedRAMP, DoD IL5, and CMMC Level 2 credentials. It serves large and regulated organizations for which downtime is extremely costly, further raising indispensability.

  But the boundary around miss it needs to be stated honestly: Zscaler is strong because of switching costs, not because it is globally irreplaceable. If it disappeared, customers could eventually migrate to Palo Alto, Netskope, Fortinet, and others. The report explicitly describes this as a market full of strong competitors where share is still being redistributed. So the precise statement is short-term switching is extremely painful, while long-term alternatives exist, not permanent monopoly indispensability. This is consistent with the report's view that the moat is real but not overwhelming.

  Now look at the second dimension: social and regulatory sustainability. On this, Zscaler is almost a positive asset:

  First, the business itself helps make society more secure. It sells cybersecurity and zero trust protection. Faster growth means more companies, governments, and critical infrastructure raise their security standards. This is the opposite of models that rely on addiction, regulatory arbitrage, or harming users. The report also classifies it as a legitimate market with rigid demand and relatively rigid customer budgets.

  Second, there is almost no regulatory sword of Damocles over the business. It does not profit from privacy arbitrage or evading regulation. The only related risk noted in the report is that under the new EU Data Act, customers may in certain circumstances cancel subscriptions without cause, usually with early termination fees, along with compliance complexity around data sovereignty and localization. These are operating costs or friction, not existential regulation that kills the business model. In fact, compliance credentials such as FedRAMP/IL5 are a source of moat, and tighter regulation is more tailwind than headwind.

  Third, data usage is defensive rather than extractive. Zscaler processes massive traffic and logs for threat research and product improvement, what the report calls scale learning effects. This serves the customers' own security rather than monetizing user data through advertising. Social perception and regulatory risk are both low.

  Overall: customers would miss it greatly because it is hard to replace in the short term, embedded in critical paths, and increasingly deep inside large customers, even though long-term competitors exist. Its growth model is clean and sustainable, almost not dependent on social harm or regulatory arbitrage, and aligned with the global rise in security and compliance requirements. For this question, Zscaler is one of the rare companies with both high indispensability and positive social value.

  Jun 11, 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 money it earns go?6/10

  A

  The unit economics are excellent and improve with scale: roughly 80% high gross margin, subscription cash collected upfront and recognized over time, and record-high operating margin on a non-GAAP basis after scaling. These are classic traits of a high-quality software business. The money it earns mainly goes into R&D, sales expansion, acquisitions to strengthen the platform, and, as the latest change, AI data-center capital expenditure. But one leakage must be counted honestly: ongoing stock-based compensation dilution means reported cash flow cannot be treated 1:1 as real shareholder return. Baillie Gifford asks about unit economics to know whether the business becomes more profitable as it grows or more strained. Zscaler is in the former camp, with solid evidence.

  Start with the unit-economic base. All three layers are strong:

  First, gross margin is high and stable. Latest FY2026 Q3 non-GAAP gross margin was 80.7%, versus 80.3% in the same period last year. Under the report's definition, GAAP gross margin has long stayed in the high 70s, with FY2025 at 76.9%. A cloud-native multi-tenant architecture means the marginal cost of serving one more customer is low. That is the structural source of high incremental returns.

  Second, operating margin clearly improves after scaling. This is the most direct evidence that the business becomes more profitable as it gets bigger: FY2026 Q3 non-GAAP operating margin reached 23%, a record high. Compared with the historical GAAP operating margins disclosed in the report, from -30.9% in 2021 to about -4.8% in FY2025, losses have steadily narrowed, and adjusted margins have repeatedly reached new highs. Scale effects clearly exist.

  Third, cash flow thickens with scale. The report records FY2025 revenue of USD 2.673 billion and free cash flow of USD 727 million, with an FCF margin of 27%. From 2021 to 2025, revenue nearly quadrupled and FCF scaled with it, exactly the pattern of cash release after scale. Add the upfront-cash, deferred-revenue model shown by deferred revenue of USD 2.477 billion and RPO of USD 6.459 billion, and cash visibility is high.

  Now look at where the money goes: the capital allocation destinations.

  First, most of it is reinvestment in R&D and sales. The report shows FY2025 R&D of USD 672 million and sales and marketing of USD 1.259 billion, still ahead of GAAP profit release. This is appropriate for a growth software company, and it appears alongside ARR and large-customer expansion, so it is not blind scale-building.

  Second, acquisitions are used to strengthen the platform. In FY2026, the company had already spent about USD 692 million acquiring Red Canary + SPLXAI, plus SquareX, and goodwill rose to about USD 1.094 billion. Capital allocation is shifting from purely organic build to buying capabilities. The logic is sound, but integration still needs to be proven.

  Third, the new cash destination is AI capital expenditure. This is the latest and important change. To support AI security, the company lowered FY2026 free cash flow margin guidance from about 26.5–27% to 22.8–23.3%, lifted FY2026 capital expenditure to the high single digits as a percentage of revenue, and said FY2027 may rise by up to another 200 basis points. In other words, the unit economics are good, but the company is actively reinvesting part of cash flow into AI data centers, reducing the near-term FCF optics. This is also a reason the stock fell from about USD 180 at the time of the report to about USD 124.73 now.

  The final required deduction is the true cost that unit economics can understate:

  The report repeatedly emphasizes that Zscaler is still GAAP-loss-making, with FY2025 net margin of about -1.6%. Attractive cash flow comes substantially from upfront billing and large non-cash expenses. The key leakage is ongoing dilution from stock-based compensation: shares outstanding rose from 138.7 million shares in 2021 to about 161.7 million shares recently, without systematic buybacks to offset it. On this basis, the report estimates conservative Owner Earnings at USD 450–550 million per year, clearly below headline FCF of about USD 870 million. So the honest conclusion is: unit economics are excellent at the company level and improve with scale, but they deserve a discount at the per-share shareholder level. Incremental returns are real, but some are consumed by dilution and some are being deliberately reinvested into AI.

  Jun 11, 2026
• QWhat conditions must hold simultaneously for it to rise fivefold over ten years? Are those conditions realistic? What expectations are implied in today's share price?3/10

  A

  A fivefold return over ten years from the current price of about USD 124.73, or about USD 20.2 billion market cap, requires the market cap to reach about USD 100 billion, implying about 17.5% annualized returns. That is much easier than starting from about USD 180 at the time of the report, which required a market cap of about USD 145 billion for a fivefold return. But it still requires several things to be true at the same time: revenue roughly triples or more over ten years, margins and cash flow truly release, dilution converges, and valuation multiples do not compress materially. This is not fantasy, but it needs nearly everything to go right, leaving little room for mistakes. This Baillie Gifford question is the full accounting of valuation and odds. The key is to lay out what must be true and what today's price already implies.

  Quantify the target first. At about USD 124.73 per share and about USD 20.2 billion market cap, a fivefold outcome means about USD 101.0 billion in market cap, or about 17.5% annualized over ten years. One important change needs to be stated honestly: when the report was written, the stock was about USD 180 and the market cap about USD 29.0 billion, so a fivefold outcome required about USD 145.0 billion. Now that the stock has fallen from its 52-week high of 336.99 by about 60% to about USD 124.73, the hurdle has moved down. This matters: for the same company, lower price improves the odds.

  For a fivefold return to work, the following conditions need to be broadly true at the same time:

  First, revenue needs to be roughly 3x or more over ten years. Starting from latest FY2026 guidance of about USD 3.33 billion, if market cap rises fivefold and valuation multiples remain unchanged, revenue needs to rise roughly fivefold as well. If modest multiple expansion is allowed, a threefold revenue increase, about 12% annualized, combined with cash-flow release could also get there. Current growth provides a cushion: Q3 revenue was +25%, ARR was USD 3.525 billion and +25%, and RPO was USD 6.459 billion and about +30%. But it requires the company not to slide into low-to-mid teens growth over ten years, the report's first reassessment trigger.

  Second, profit and cash flow need to truly release, and dilution must converge. Non-GAAP operating margin has reached a record high of 23%, but GAAP is still loss-making, with FY2025 net margin of about -1.6%, and shares outstanding have still climbed from 138.7 million shares to about 161.7 million shares. For a fivefold outcome to show up per share, dilution must slow or buybacks must offset it. Otherwise market cap can rise fivefold while per-share returns are discounted. The report's key assumption is exactly that dilution gradually declines or buybacks begin to offset it.

  Third, this round of AI capital expenditure must produce returns rather than structurally consume cash flow. The company has already lowered FY2026 free cash flow margin guidance from about 26.5–27% to 22.8–23.3%, with FY2027 capital expenditure rising by up to another 200bps. The fivefold scenario requires these AI investments to convert into new high-gross-margin revenue, including data security and AI Protect, rather than permanently lower cash flow.

  Fourth, competition must not become fatal, and share must be defended. The platformization push from Palo Alto, which has completed its roughly USD 25 billion CyberArk acquisition, Fortinet, Netskope, September 2025 IPO, and others must not materially capture Zscaler's budget pool.

  Are these conditions realistic? Each condition is individually reasonable, but together they require almost everything to go right. The most fragile combination is growth not falling a tier for ten years, dilution converging, and valuation not compressing. If any one of these fails, a fivefold return becomes difficult. The report's judgment matches mine: it can be underwritten, but it is not a high-probability glide path.

  Finally, what does today's price imply? This is the soul of the question, and the picture has changed because the stock has fallen sharply:

  At the time of the report, around USD 180, the intrinsic value ranges were conservative 90–115, reasonable 120–155, and optimistic 160–195 dollars. The report judged the then-current price as close to the lower end of the optimistic scenario and clearly above the reasonable midpoint, with insufficient margin of safety. But today's roughly USD 124.73 is already at the lower end of the report's reasonable intrinsic value range of 120–155 dollars and near the upper boundary of its ideal purchase range of 95–125 dollars. In other words, a large part of the over-optimistic expectation the report disliked has been digested by the market's sharp correction. The current price no longer implies that the optimistic case must happen; it is closer to requiring the reasonable case to play out. Based on the current price and the report's conservative Owner Earnings estimate of USD 450–550 million, the roughly USD 20.2 billion market cap implies about 37–45x conservative owner earnings and about 23x headline FCF, using FY2026 guided FCF margin of about 23%, or about USD 760 million. That is still not table-pounding cheap, but it has improved from clearly expensive at the time of the report to within a reasonable range.

  Overall: a fivefold return over ten years needs a chain of conditions to hold simultaneously, with difficulty skewed moderately high and little room for errors. But because the stock has fallen sharply from the report's price to the lower end of the reasonable range/upper boundary of the ideal purchase range, today's implied expectations are much more moderate than when the report was written. The odds have improved, but the company still needs to turn high-quality growth into real per-share cash returns over a long period.

  Jun 11, 2026
• QWhy has the market not realized all this yet? Does it fail to understand it, look down on it, or not look far enough ahead? What will be the narrative inflection point?3/10

  A

  For Zscaler, the question needs to be reversed: the market has not failed to recognize that it is a good business. Quite the opposite, it was once fully and even excessively priced, with the stock reaching about USD 337 within the past year. The current fall of about 60% from the high to about USD 124.73 reflects a swing in market sentiment from AI security winner to growth deceleration, AI spending consuming cash flow, and sales disruption. So the real perception gap is not that the market does not understand this is a good business. It is whether, amid pessimism, the market is underestimating Zscaler's ability to turn current AI capital expenditure into long-term returns. That is closer to not looking far enough ahead. Baillie Gifford's question asks whether the market fails to understand, looks down on it, or does not look far enough ahead. For Zscaler, the answer is that the good business was understood long ago; the current issue is short-termism plus an emotional overshoot.

  Start by correcting the premise. Zscaler is not an obscure, neglected stock. When the report was written, it was about USD 180 per share and about USD 29.0 billion in market cap, with dense institutional coverage, and it was a star name in SSE/zero trust. Its problem was never that nobody had discovered it. The question was whether the price was too high. Now that the stock has fallen by about 60% from its 52-week high of 336.99 to about USD 124.73, the market has already squeezed out part of the optimism. So the question for Zscaler is why the market is discounting it now, not why the market has not found it.

  There are three specific reasons for the current discount. None is about failing to understand; all reflect a short-term lens magnifying bad news:

  First, the free cash flow guidance cut hit the core of the valuation. To bet on AI, the company lowered FY2026 free cash flow margin guidance from about 26.5–27% to 22.8–23.3%, and FY2027 capital expenditure as a share of revenue may rise by up to another 200 basis points. For a software stock whose valuation is supported by cash flow, a cut to FCF guidance is a direct negative. The market voted with its feet. This is the core recent catalyst for the decline.

  Second, sales disruption triggered concern about growth deceleration. Two sales leaders left at the end of Q3, and management therefore took a prudent/tempered stance on FY2027 new logos. The market has very low tolerance for growth deceleration in growth stocks. Personnel change plus conservative language was amplified into a fear that the growth story is stepping down.

  Third, AI is both the positive and negative side of the narrative. On one side, the market worries that AI may lower security barriers and let new entrants disrupt the sector. The report and news both mention that intensified AI competition hurt sector sentiment. On the other side, investors doubt whether Zscaler's heavy AI data-center spending will show near-term returns. Sentiment flipped from AI beneficiary to AI victim/spender.

  So where is the perception gap? This is what Baillie Gifford is really after:

  The possible market short-sightedness is that current pessimism treats AI capital expenditure as pure cash-flow loss, while possibly underestimating the other side of the ledger. If these investments create an AI security lead, as shown by AI Protect orders above USD 100 million over 12 months and data security ARR above USD 500 million and +30%, then Zscaler is buying the foundation for the next stage of growth. At the same time, the fundamentals have not collapsed: Q3 revenue was +25%, ARR was USD 3.525 billion and +25%, RPO was USD 6.459 billion and about +30%, and net revenue retention was about 115%. The market is discounting a company whose long-term platform value is still accumulating because the near-term FCF optics have worsened. If you believe the report's judgment that the business is excellent and the moat is real, this correction has moved the price from the report's expensive zone into the report's reasonable lower bound/ideal purchase upper boundary of USD 95–125. The direction of the perception gap is favorable for long-term investors.

  But the opposing case must be preserved honestly: the market may be seeing more clearly rather than failing to look far enough ahead. If sales disruption foreshadows structural growth deceleration, AI investment suppresses cash flow for a long time, and competition, including Palo Alto's roughly USD 25 billion completed CyberArk acquisition and other platformization moves, starts taking budget, then the current discount is a fair repricing rather than an overreaction. This is the mirror image of the report's repeated warning that permanent capital loss comes from buying at too high a price and then suffering both valuation and fundamental compression. The difference is that the price has already come down substantially.

  What would be the narrative inflection point?

  Upside inflection: AI investment starts paying off and growth stabilizes. AI Protect/data security keep scaling, moving from USD 100 million and USD 500 million to larger bases; FCF margin rises again after the capital-expenditure peak; the sales team stabilizes and new logos reaccelerate; organic ARR growth holds in the mid-to-high teens. Any combination of these could shift the narrative from burning cash and slowing to AI-era platform winner.

  Downside inflection: bad news is confirmed. Organic ARR growth falls below the mid-teens, growth in customers with more than USD 1 million in ARR slows materially, gross margin falls below the mid-70s, FCF margin keeps deteriorating, or competitors clearly take budget in key product areas. These are the report's reassessment triggers.

  Overall: Zscaler is not currently buried and undiscovered. It has been repriced from the highs by pessimism. The real perception gap is that the market may not be looking far enough ahead and may be underestimating the long-term payoff from AI investment and the resilience of the fundamentals. But unlike many neglected stocks, the price has already fallen into the report's reasonable/ideal purchase range. The narrative inflection point depends on which happens first: AI investment paying off as expected, or growth stabilizing.

  Jun 11, 2026