Following CrowdStrike and Palo Alto, another cybersecurity winner emerges in the era of AI inference! Zscaler (ZS.US) enters the trillion-dollar race with a sales pipeline surging by 75%.

According to Zhitong Finance APP, shares of the top U.S. cybersecurity technology company Zscaler (ZS.US) surged over 11% after-hours on Thursday, following the release of a strong quarterly earnings report and forward-looking guidance that both exceeded market expectations. This performance underscores how, as generative AI moves from experimental deployment to the era of large-scale deployment of AI agents, cybersecurity budgets are shifting from discretionary spending to a prerequisite for AI deployment. Furthermore, cybersecurity product lines are seeing a growth momentum in orders driven by the explosive expansion of the AI reasoning market. As artificial intelligence moves toward massive-scale inference and Agentic AI workflow phases, cybersecurity needs are no longer simply "growing in lockstep with AI technology updates," but are likely to experience structural incremental expansion far above that of traditional IT spending.

Zscaler's Q4 revenue was $898.2 million, up 25% year-on-year, surpassing the market expectation of $877 million; adjusted earnings per share were $1.19, higher than the expected $1.09; annual recurring revenue (ARR) grew 25% to $3.771 billion. More importantly, "Security for AI" bookings grew over 50% quarter-on-quarter, and the sales pipeline grew 75%—the former reflects orders already signed, while the latter reflects a strong reserve of future orders. The more frequently models call data, applications, and tools, the greater the needs for zero trust access, identity governance, and data loss prevention become.

Zscaler's management expects first quarter 2027 revenue to be between $935 million and $939 million, above the market expectation of $927 million; adjusted earnings per share are projected at $1.15-$1.16, also above the expected $1.08. The overall guidance for full-year revenue is between $3.908 billion and $3.938 billion, slightly above the expected $3.9 billion; adjusted earnings per share guidance is $4.86-$4.90, significantly ahead of the expected $4.60, but ARR is forecasted to grow by only 16.6%-17.4%, which is a clear slowdown from the 25% seen in fiscal year 2026.

After the earnings and forward guidance were released, Zscaler's stock price initially spiked more than 11% after-hours, but later slipped into a slight decline of 1% after-hours. So far this year, the stock has still fallen about 20%. This round of strong rebound is essentially the market re-evaluating the tremendous expansion prospects and stronger monetization capability of its cybersecurity business in the AI reasoning era, rather than a complete disappearance of high-valuation risk and sustainable growth concerns.

Peer performances in cybersecurity further verify the strength of demand. CrowdStrike's Q2 revenue was $1.47 billion, up 26% year-on-year; ARR grew 25% to $5.84 billion, with net new ARR hitting a record $333 million, up 51% year-on-year, and free cash flow reached $377 million. After the report, its shares jumped as much as 10.4% post-market. Another cybersecurity giant, Palo Alto Networks, reported Q4 revenue of $3.41 billion, up 34% year-on-year and better than the expected $3.35 billion; adjusted earnings per share were $1.02, exceeding the expected $0.98. Next-generation security ARR rose 63% to $9.1 billion, and fiscal year 2027 revenue guidance of $14.1-$14.2 billion also beat the market's $13.79 billion expectation. Both CrowdStrike and Palo Alto Networks have rallied over 80% this year, forming a sharp contrast with Zscaler’s previous underperformance.

The underlying logic is clear: AI inference, especially Agentic AI-dominated autonomous workflow, is no longer just about generating text — it involves continuous model invocation, calling APIs, databases, and enterprise tools, and accessing sensitive data and performing operations at machine speed. New risks such as machine identities, privilege escalation, prompt injection, data leakage, agent lateral movement, model supply chain pollution, and autonomous attacks have emerged. Security architecture must shift from traditional boundary protection to continuous identity authentication, least-privilege access, inline traffic inspection, and runtime governance. Though compute demand for training can fluctuate in projects, every inference, tool call, and data access by agents requires secure controls and AI security tools, providing sustained revenue growth for cybersecurity product lines as AI usage expands.

Palo Alto Networks management, in the Q&A session, cited a future scenario of $5 trillion of AI capital expenditure over the next five years and pointed to about $1 trillion in global cybersecurity technical debt in urgent need of modernization. Palo Alto CEO Nikesh Arora stated on the earnings call that the latest advancements in AI are pushing cybersecurity to the top of enterprise CIO priorities and will be a "long-lasting tailwind." He further stressed that approximately $1 trillion in global cybersecurity infrastructure is unprepared for AI threats, presenting a significant long-term growth opportunity for the industry.

Who exactly is Zscaler? How does its main business differ from cybersecurity giants CrowdStrike and Palo Alto Networks?

Zscaler is a high-purity play on zero trust and AI agent traffic growth, with higher flexibility but greater reliance on sales execution and converting new ARR; CrowdStrike leverages its vast endpoint telemetry and security operations automation to benefit more directly from AI-driven demand for attack detection; Palo Alto Networks owns the most comprehensive product line with the strongest cross-selling capability, making it well-positioned for the trend of enterprises consolidating vendors in a "platform integration" strategy. In short, Zscaler bets on "eliminating network trust," CrowdStrike bets on "detecting and stopping attacks," and Palo Alto bets on "consolidating the complete security system on a single platform."

Zscaler is a cloud-native zero-trust-focused cybersecurity company. Its "Zero Trust Exchange" platform sits in the enterprise traffic path, connecting users, workloads, branches, or AI agents to authorized applications without directly routing them into internal networks.

The company's primary business covers Secure Service Edge, Security Access Service Edge (SASE), Internet Access Security, Private Application Zero Trust Access, Data Loss Prevention, Cloud Workload and Branch Security, Digital Experience Monitoring, and is expanding into AI agent communications, model access control, and Agentic SecOps.

Zscaler is the "traffic and access control layer," using a cloud-based inline proxy to replace traditional VPNs, network perimeters, and some security appliances; CrowdStrike serves as the "endpoint telemetry and detection-response layer," based on the Falcon platform and endpoint agents, focusing on Endpoint Detection and Response (EDR), Extended Detection and Response (XDR), cloud workload, identity protection, threat intelligence, and Next-Gen SIEM, excelling in spotting and terminating attacks happening on devices and workloads; Palo Alto Networks is the broadest "full stack security platform" player, starting from Next-Generation Firewalls (NGFW) and now covering both Network and SASE, Prisma Cloud Security, Cortex Security Operations, and Identity Security, combining hardware, software subscriptions, and cloud platforms.

AI risks morph into cybersecurity urgency: Zscaler delivers a blowout “zero trust”-focused report

As the era of artificial intelligence inference drives up cybersecurity risks, enterprises are compelled to seek security tools urgently. This cloud security company's Q4 results and forward outlook topped Wall Street's consensus, which is why Zscaler’s stock price surged sharply after-hours on Thursday.

For Q4, the company reported revenue of $898.2 million, a 25% year-on-year gain and above expectations of $877 million; adjusted EPS of $1.19, versus the $1.09 expected. Zscaler reported a GAAP net loss of $3.4 million, or $0.02 loss per share; the prior-year net loss was $17.6 million, or $0.11 loss per share.

For all of fiscal 2026, revenue was $3.353 billion, up 25% year-on-year; excluding Red Canary, revenue was $3.209 billion, up 20%. Non-GAAP operating profit rose to $767.1 million, with the operating margin increasing from around 21.7% to 22.9%; non-GAAP EPS climbed from $3.28 to $4.21.

Full-year operating cash flow was $1.13 billion, higher than last year’s $972.5 million; free cash flow increased to $779.1 million, but the free cash flow margin fell from 27% to 23%. Q4 free cash flow dropped from $171.9 million in the prior-year period to $60.8 million, mainly due to capital expenditures and internal software investments rising from $78.7 million to $218.5 million; thus, profitability quality and capital spending remain variables that need valuation scrutiny.

On the earnings call, management disclosed net new ARR of $246 million, up 24% year-on-year; excluding Red Canary, it was $232 million, up 17%. Remaining Performance Obligation (RPO) grew 27% to $7.365 billion; total contract value for Z-Flex in Q4 exceeded $770 million, up over 60% sequentially and above $1.7 billion for the year.

Management revealed that 70% of AI security deals also included data security products, and Zero Trust Everywhere enterprise clients jumped from over 700 last quarter to over 950, with seatless products accounting for about 30% of new and upsell annual contract value.

For Q1, the company expects revenue between $935 million and $939 million, and adjusted EPS of $1.15-$1.16, both beating consensus forecasts of $927 million in revenue and $1.08 in adjusted EPS.

Meanwhile, the company will restructure about 3% of employees and record $30-$33 million in charges; management expects FY2027 revenue of $3.908-$3.938 billion, above consensus, ARR guidance at $4.396-$4.426 billion, with ARR growth slowing to about 17% from 2026, so whether the stock can be re-rated will depend on whether the AI security pipeline can be converted into net new ARR, offsetting sales team restructuring and higher Red Canary attrition.

Cybersecurity rallies but stock prices diverge; Zscaler bets on a repricing of cybersecurity in the era of AI agents

Regardless of whether closed-source or open-source models ultimately dominate, cybersecurity is one of the most "model-path-neutral" beneficiary layers: closed-source models must tackle third-party interface, data perimeter, and vendor concentration risks; open-source and open-weight models raise issues of model provenance, component dependencies, self-hosting environment, and patch fragmentation risks. In other words, the more frequent the inferences and the greater the autonomy of agents, the more identities, endpoints, APIs, cloud workloads, data, and runtime environments that enterprises must protect.

Palo Alto CEO Nikesh Arora emphasized on this week's earnings call the “around $1 trillion in global cybersecurity debt,” which essentially underpins a massive upgrade cycle for pre-AI architecture that must now be modernized.

The company’s CEO Jay Chaudhry highlighted that the popularization of the company’s zero trust cloud security architecture and its innovative technologies were the main drivers behind the better-than-expected quarterly performance.

In an interview with CNBC during the results conference, he expressed strong optimism about the company’s new generation zero trust product for AI agents. The product is already showing early momentum and is expected to accelerate quickly in fiscal years 2028 and 2029.

“This is a longer-term opportunity, but I think it’s a very attractive one with significant barriers to entry,” Chaudhry said.

Annual recurring revenue data grew 25% year-on-year to $3.77 billion, slightly ahead of the consensus expectation of approximately $3.75 billion.

As new, increasingly sophisticated cyber attack models and agent-driven threats emerge, forcing companies to adopt newer security tools, cybersecurity stocks have surged this year.

Despite competitors’ new highs and gains of over 80% this year, Zscaler shares have plummeted 20%. Last quarter, after two sales executives departed, management adopted a "cautious strategy" on guidance, leading to the worst single-day performance in the company's history.

But Chaudhry said the market is missing the key to Zscaler's differentiated strategy. "Our core capabilities are quite unique; as AI agents become more widespread, the market will increasingly realize Zscaler is an essential player," he said. Like other cybersecurity executives, Chaudhry sees AI security as one of the largest opportunities. Over the past year, the business booked a total of $100 million in orders, with quarter-on-quarter growth of over 50% this quarter.