A-Portfolio

The company in one breath

Zscaler, ticker ZS on the Nasdaq, runs a cloud security checkpoint between people, devices, workloads, AI agents, and the applications they want to reach. Its Zero Trust Exchange checks identity and policy before granting access instead of trusting anything merely because it sits inside a corporate network. The company says the platform handles more than 750 billion transactions each day. That means Zscaler sees an enormous stream of security traffic, which can improve threat detection and make the platform hard to replace. Good would look like that scale producing faster organic growth, better customer outcomes, and more profit without a matching rise in spending.

Why now

Zscaler reported fiscal fourth-quarter results Thursday night, and the first market reaction was drunk. The stock shot as high as $201.21 after hours, then fell back to $173.25 by 3:59 a.m. Eastern on Friday. That round trip means traders loved the quarter and then read the full-year forecast. Good price action would hold above the prior close of $178.02 once regular trading brings real volume, not the thin overnight crowd.

The quarter beat Wall Street. Revenue reached $898.2 million, up 25%, against the $877 million analysts expected. In plain English, sales grew fast and cleared the bar. Good for a company this size is organic growth above 20%, and Zscaler hit that mark after removing the Red Canary acquisition.

The fight is over what comes next. Management guided fiscal 2027 revenue to $3.908 billion to $3.938 billion, about 17% growth at the midpoint. That means the company expects a meaningful slowdown from the latest quarter. Good would look like management beating that forecast while keeping growth near or above 20%, which would show the sales reset is working and the guide was cautious rather than accurate.

The numbers, translated

Annual recurring revenue was $3.771 billion, up 25%. Annual recurring revenue, or ARR, estimates the subscription revenue already contracted for the next year. Red Canary, a private company Zscaler bought, contributed $141 million of that total. Without it, ARR grew 20% to $3.630 billion. Good would look like organic ARR growth staying above 20% as Security for AI and Agentic SecOps become products customers actually pay for.

Net new ARR was $246 million, up 24%. This is the fresh recurring business added during the quarter after churn and contract changes. Excluding Red Canary, net new ARR grew 17%, up from weaker growth earlier in the year. That says the core selling motion improved. Good would look like another two quarters of acceleration, because one quarter is a date, not a marriage.

Full-year revenue was $3.353 billion, up 25%, while free cash flow was $779.1 million. Free cash flow is cash left after operating the business and paying for equipment and internal software. The resulting 23% margin means Zscaler kept about 23 cents of cash from each sales dollar. Good for a mature cloud software company is at least 25 cents, sustained without stretching customer payment terms.

Fourth-quarter free cash flow margin fell to 7% from 24%. The company spent $218.5 million on equipment and internal software during the quarter, up from $78.7 million a year earlier. That means infrastructure spending swallowed most of the quarter's cash production. Good would look like the spending producing capacity now and the margin climbing back above 20% quickly.

GAAP operating loss was $15.5 million, or 2% of revenue. GAAP means the accounting rulebook includes costs such as employee stock awards. The loss narrowed from $32.2 million, so the real income statement moved toward breakeven. Good would look like positive GAAP operating profit through a full year, not just a flattering adjusted number.

Stock-based compensation and related payroll taxes were $841.6 million for the year, equal to 25% of revenue. This is employee pay delivered partly through shares, which can dilute existing owners even when it does not leave the bank account today. Weighted average shares rose about 3.8% from the prior year. Good would look like share pay below 20% of revenue and annual dilution below 2%.

Cash and short-term investments totaled $3.474 billion against $1.696 billion of convertible debt. That leaves roughly $1.778 billion of net cash, meaning liquid assets exceeded debt. Good looks like keeping that cushion while acquisitions start producing organic cross-sell instead of becoming expensive souvenirs.

At the overnight price, the stock carries a market value near $28.25 billion and an enterprise value near $26.47 billion. Enterprise value is the price of the operating business after adding debt and subtracting cash. That equals about 6.7 times the midpoint of fiscal 2027 revenue. Good for a company guiding to roughly 17% growth would be a multiple below 6 times sales, unless growth proves it can reaccelerate.

What I like

The architecture makes sense. Zscaler connects a user or machine only to the application it needs, rather than opening a door onto the whole network. AI agents make that distinction more important because software can now act at machine speed and screw up at machine speed too.

The company still has category weight. About 11,000 customers use the platform, including more than half of the Fortune 500. More than 950 enterprises now buy protection across users, branches, and cloud workloads, up from 700 in the prior quarter. That means larger customers are buying more of the platform. Good would be continued growth in that group without discounting margins into the floor.

The sales engine showed life after two leadership departures helped trigger the stock's worst day in May. Core net new ARR growth improved to 17%. Management also said Security for AI bookings exceeded $100 million over the last year and grew more than 50% sequentially in the quarter. That is early, not proven, but it is a real number attached to the AI pitch.

The founder still runs the place. Jay Chaudhry built Zscaler around cloud-delivered zero trust before the phrase became conference-booth wallpaper. Founder leadership earns points when the founder keeps adapting the original architecture instead of defending yesterday's product.

What worries me

Fiscal 2027 guidance does not confirm the quarter's acceleration. The midpoint implies about 17% revenue growth and roughly the same ARR growth. One analyst calculated the ARR outlook implies only around 4% growth in net new ARR for the year. Good would be management proving that math too conservative by the first half.

Zscaler plans to cut about 3% of its workforce while redirecting money toward AI and growth work. A focused reshuffle can help. A reshuffle after sales-leadership disruption can also mean the machine still needs parts. Good would look like higher sales productivity and lower employee turnover without another executive surprise.

Competition is vicious. Palo Alto Networks (PANW) can bundle cloud access into a broad security relationship. Cloudflare (NET) can attack with its network, self-service motion, and public pricing. Zscaler has depth, but it still asks customers to tolerate a sales process, a complex rollout, and occasional performance headaches discussed by users in r/Zscaler.

The accounting gap remains too wide. Adjusted operating margin hit a record 24%, while GAAP operations still lost money. The biggest bridge is share pay. I do not call employees free because the accountants use a different drawer.

What would make this a buy

I want one of two doors.

Door one is price. From $140 to $150, the enterprise value would fall to roughly 5.4 to 5.8 times fiscal 2027 sales. That means I would pay a saner price for a company growing in the high teens. Good would be a calm base in that zone rather than a straight knife through it.

Door two is proof. I would pay closer to today's price if organic ARR growth stays above 20%, full-year free cash flow margin reaches at least 25%, and GAAP operating profit turns positive. I also want Security for AI and Agentic SecOps to add enough business that management separates the numbers instead of serving one large bowl of AI soup.

No chasing Friday's first print. The stock traveled from $201.21 to $173.25 overnight. Let the adults arrive, let price discovery happen, and then judge the setup.

What makes me walk away

I walk if organic ARR growth falls below 15% for two quarters while management keeps talking about reacceleration. That would mean the story improved faster than the business.

I walk if free cash flow margin stays below 20% after the current infrastructure build, or if dilution remains above 3% a year. Cash generation that depends on paying employees with ever more stock is not clean cash generation.

I walk if the sales reset produces another senior departure, another guide cut, or weak large-deal activity. Cybersecurity demand is not the excuse. Palo Alto Networks and Cloudflare will happily cash the checks Zscaler misses.

IPO Watch

Siyata PTT (private, proposed ticker PTT), expected September 8, reference price $7 with no traditional range. The company sells rugged push-to-talk handsets and vehicle communications gear. This is a direct listing of up to 6.1 million existing and warrant-linked shares, which means selling holders get liquidity and the company gets no sale proceeds. Its filing shows $11.1 million of 2025 revenue, down 4.2%, a 16.3% gross margin, and substantial doubt about its ability to continue as a going concern. I would write it up only if the listing comes with enough funding visibility for at least a year and gross margin starts moving above 25%. Right now, this is not a debut. It is an escape hatch with a ticker.

Earnings On Deck

Oracle (ORCL), September 10 after the close. Watch cloud infrastructure growth and the cash cost of building AI capacity. A good print means backlog converts into revenue without capital spending crushing free cash flow. A bad print means giant AI contracts require giant checks long before the customer money arrives.

Adobe (ADBE), September 10 after the close. Watch whether generative AI lifts paid users and annual recurring revenue rather than merely lifting demo applause. A good print means stronger Digital Media growth and stable margins. A bad print means AI usage grows while pricing power and customer retention weaken.

The verdict

Score: 78 out of 100. Bucket: watch. Zscaler earned points for founder-led category leadership, 20% organic ARR growth, a deep enterprise footprint, net cash, and a credible AI-security extension. It lost points for a fiscal 2027 slowdown to roughly 17%, a fourth-quarter cash margin collapse, stock pay equal to 25% of revenue, sales-team disruption, and a valuation that still assumes plenty goes right.

A candidate starts at 80. Zscaler misses by two because the quarter says acceleration and the guide says patience. I believe the platform. I want the numbers to stop arguing.

The DBTech read

Organic growth measures how much a company expanded from its own products and customers, excluding businesses it bought. Zscaler's headline ARR grew 25%, but growth was 20% after removing Red Canary. The lower number tells us what the original machine produced without acquisition help. Good organic growth stays strong enough that acquisitions add fuel rather than hide a slowdown. When two growth rates appear, ask which one management had to buy.

Sources and as-of date

• Price: $173.25 at 3:59 a.m. Eastern on September 4, 2026, Alpaca overnight feed. Prior regular-session close: $178.02 on September 3, 2026, Alpaca IEX feed.
• Zscaler fiscal fourth-quarter and fiscal 2026 results, September 3, 2026.
• Zscaler fiscal 2026 Form 10-K, filed September 3, 2026.
• CNBC earnings report and Jay Chaudhry interview, September 3, 2026.
• TIKR earnings-call analysis, September 4, 2026.
• Siyata PTT Form F-1, as amended through August 2026.
• Oracle earnings date announcement, September 2, 2026.
• Earnings dates checked against the September 7 to September 11, 2026 calendars from Earnings Whispers and StockTi.
• Community and product pulse checked across 13 Reddit threads, 4 YouTube videos, 5 GitHub items, and 364 public job listings collected from August 5 through September 4, 2026. No relevant Hacker News stories or prediction markets surfaced.
• All company financials are as of July 31, 2026, unless stated otherwise. Market-derived calculations use the overnight price above. Research and commentary only, never a buy signal.