Zscaler beat Q4 FY26 and traded up roughly 8% after hours; Ciena beat the same day and was sold roughly 10%. The growth rates differ. The regime is identical. The only variable is where each stock’s embedded expectation sat.
Figures as reported by Zscaler and same-day coverage; not independently audited. The ~+8% after-hours move is an initial reaction — not a verdict; it can and does change by the open and beyond. Not investment advice. No buy/sell recommendation. No price targets.
Per Zscaler’s release on GlobeNewswire and same-day coverage, Zscaler reported Q4 FY26 revenue of approximately $898.2 million, up roughly 25% year over year against a consensus estimate of around $877 million. Non-GAAP EPS came in at approximately $1.19 versus the roughly $1.09 expected. Annual recurring revenue reached approximately $3.771 billion, up 25% year over year, crossing the psychologically significant $3 billion threshold, with net new ARR of roughly $246 million and record operating margin.
For fiscal 2027, the company guided revenue to roughly $3.908 to $3.938 billion — approximately 17% growth — with non-GAAP EPS of $4.86 to $4.90. The stock moved up roughly 8% in after-hours trading. That initial move is a price reaction; it is not a settled verdict on valuation, and it can change materially by the open and in the days that follow.
The same day, Ciena reported its own beat — roughly 37% revenuegrowth — raised its guide, and was sold approximately 10%. Same earnings-beat pattern. Opposite market reaction. That divergence is the story, and the explanation has nothing to do with which company grew faster.
The Bar Reset — Context Per Reporting
May 2026
Zscaler issues cautious FY27 frame — ARR growth guided to mid-teens. Single-session drop of roughly 31%, per reporting — the worst in company history. The bar is reset to the floor.
Sep 3, 2026 — Pre-market
Ciena beats Q3 at ~37% revenuegrowth, raises guide. Stock sold roughly 10% — it had grown into a sky-high embedded expectation. A guide that merely met the bar in the price landed as a hold.
Sep 3, 2026 — Same Day
OpenAI designates GPT-6 Astra its first ‘Critical’ cybersecurity model — AI cyber-offense that reportedly found and chained real zero-day vulnerabilities. The attack surface expands.
Sep 3, 2026 — After Hours
Zscaler Q4 FY26 beat. FY27 guide firms the ~17% frame rather than re-cutting it. A guide that would disappoint against a high bar clears a low one easily. Initial after-hours move: roughly +8%.
The key insight: The 17% FY27 guide is the same number that triggered a roughly 31% collapse in May. Today it triggered a roughly 8% after-hours gain. The guide did not change. The embedded expectation did. That asymmetry is the entire trade — and it is the cleanest confirmation yet of the expectations-regime framework that the Ciena read established earlier the same day.
The Structural Read
The contrast between Ciena and Zscaler on September 3 is not a story about growth rates. Ciena grew 37%; Zscaler grew 25%. By any conventional read, Ciena’s quarter is better. Yet the market bought Zscaler and sold Ciena. The explanation is that the market is pricing the second derivative — not the level of growth, but whether the print beats what the price had already embedded. This is the expectations-regime framework, and both names confirmed it simultaneously.
Ciena’s problem is that it grew into a sky-high embedded bar. Its stock had re-rated aggressively on AI-infrastructure tailwinds; by the time the print arrived, a merely in-line guide was already baked. Meeting the bar reads as a hold. Zscaler’s advantage is the mirror image: May’s collapse did the work of resetting expectations to a pessimistic floor. A guide that firms rather than re-cuts clears that floor easily — the same approximately 17% growth frame that caused a roughly 31% single-session drop in May becomes a relief print in September. Same regime, opposite outcome. The deciding variable is not growth velocity; it is the distance between the reported number and the number the price assumed.
The deeper thread is the demand driver underneath the beat. Zscaler sells zero-trust and secure-access security — SSE architecture that defends enterprise perimeters. Its structural tailwind is that AI, and specifically agentic AI that browses, calls tools, and acts autonomously, expands every enterprise’s attack surface. More agents, more endpoints, more lateral movement risk, more demand for zero-trust enforcement. That connection is not a claim that agentic AI shows up in this specific quarter’s ARR; Zscaler’s growth predates today’s agent deployments by years. It is an analytical framing of the direction of the demand curve.
Business Engineer — Expectations Regime
“It is not how fast you grow. It is whether you beat what the price already assumed. A bar on the floor is easier to clear than a bar at the ceiling — even if the ceiling belongs to the faster-growing business. The market is always pricing the second derivative against the embedded number, and the only question worth asking before earnings is: where did the bar sit?”
What closed a loop today is that both ends of the AI cybersecurity flywheel reported simultaneously. On the offense side, OpenAI designated GPT-6 Astra its first ‘Critical’ cybersecurity model — a system it says found and chained real zero-day vulnerabilities. On the defense side, Zscaler printed a Q4 beat driven by the zero-trust and SSE products that exist precisely to contain the kind of threat Astra represents. That co-occurrence on September 3 is an analytical connection between two events, not a claim that one caused the other or that Astra’s capabilities appear anywhere in Zscaler’s current financials. The connection is structural: more capable AI offense manufactures demand for AI defense. The security layer is where the AI capability race converts into recurring enterpriserevenue — and today both ends of that flywheel reported in the same 24-hour window.
Three Implications
IMPLICATION 1 — The Reset-Low-Bar Asymmetry Is Now the Primary Signal
For AI-adjacent software names, the most important question going into an earnings print is not the consensus growth estimate — it is whether the stock has been punished to a pessimistic expectation or bid up to an optimistic one. Zscaler was punished in May. That punishment was the setup for today. Names priced against their own disaster scenario have the lowest bar to clear; names that have re-rated aggressively have the highest. The analysis framework here is identical to the Ciena read — same regime, opposite position on the bar.
IMPLICATION 2 — The AI Cyber Offense/Defense Flywheel Is Structurally Self-Reinforcing
GPT-6 Astra’s ‘Critical’ designation is not a curiosity for security researchers — it is a demand signal for zero-trust and SSE vendors. Every increment of AI offense capability raises the floor of what enterprises must spend on AI defense. Zscaler, Palo Alto, CrowdStrike, and any name that sits at the enforcement layer of the enterprise perimeter inherits that demand mechanically. The flywheel does not require a direct causal link between Astra and any specific ARR cohort today; it operates on the structural direction of enterprise security spend as AI agents proliferate.
IMPLICATION 3 — The Security Layer Is Where AI Capability Converts to Recurring Revenue
In the Map of AI stack, the security layer sits above infrastructure and below application — it is the enforcement plane that every enterprise must maintain regardless of which foundation model or agent framework they adopt. That position makes security revenue more durable and less winner-take-all than the application layer above it. Zscaler’s ARR crossing $3.771 billion on a 25% growth rate, with record operating margin, is a data point in favor of that structural positioning — not a prediction about future growth, which the FY27 guide already frames conservatively at approximately 17%.
The Map of AI Redrawn tracks 200+ companies across 9 layers of the AI stack. The security enforcement layer — where Zscaler, Palo Alto, and CrowdStrike operate — is positioned between infrastructure and application: it inherits demand from every layer above and below it as agentic AI expands the attack surface. The expectations-regime read and the offense/defense flywheel both map directly onto where these names sit in the stack. Today’s Zscaler/Ciena divergence is a live case study in how the market prices AI-layer positioning against embedded expectations — not absolute growth rates.
91,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
This is businessanalysis, not investment advice — no buy/sell recommendation and no price targets. Earnings figures are as reported by Zscaler and same-day outlets, not independently audited; the after-hours move is an initial reaction that can change. The GPT-6 Astra cyber connection is an analytical link between two same-day events, not a driver of this quarter’s results.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
