If you follow AI cybersecurity, you'd notice something odd about Zscaler: its earnings were strong—revenue up 25%, ARR up 25%, and operating profit margins hit a record high—yet the stock was sold off.
This isn't just a routine earnings fluctuation; Wall Street is judging a bigger question: Is cybersecurity in the AI era a true necessity or just another overhyped software story?
If you only look at the stock price, you'd think Zscaler is in trouble.
But if you look deeper, you'll see that what's really in trouble might not be Zscaler, but the entire enterprise security rulebook being rewritten by AI.
In the past, companies defended against hackers—humans.
Now, companies must defend against AI that can automatically write code, find vulnerabilities, fake identities, and even autonomously enter systems to execute tasks.
What's scarier is that in the future, companies won't just have employee accounts; they'll have AI Agent accounts. These agents will access customer data, call internal systems, read databases, send emails, and even trigger business processes.
So the question arises.
Who decides what these AI Agents can and cannot see?
Who prevents them from exceeding permissions?
Who stops employees from tossing confidential data into external models?
And who can guard the last door when AI helps hackers scan vulnerabilities at machine speed?
That's why this Zscaler video is worth watching.
After watching, you'll understand three things.
First, why Zscaler's earnings were strong yet the stock fell—what is the market worried about?
Second, why zero trust, AI Red Teaming, and Agentic SecOps might become the hardest new theme in AI software stocks.
Third, is Zscaler a slowing software stock or a core AI security player mistakenly punished by short-term sentiment?
If you still think AI security is just a niche within cybersecurity, you might be missing the most dangerous and profitable dark lane in the next AI software rally.
Because the stronger AI gets, the less companies dare to go unprotected.
And whoever can guard the entry point of AI into enterprise systems may become the next true security toll gate. Before the video starts, please like, subscribe, and let's dive into today's focus.
First, the earnings themselves.
Zscaler's Q3 earnings looked very good on the surface.
Revenue up 25% year-over-year.
ARR up 25% year-over-year.
Non-GAAP operating margin reached 23%, a historical high.
What does this mean?
It means Zscaler is not just a software company burning money on stories. It has entered a more mature stage: revenue still growing, margins improving, customer demand not disappearing, and the zero-trust theme not fading.
But the problem is that the US stock market is never a gentle teacher.
It won't reward you just because you scored 90.
If the market expected you to score 95, or even expected you to continue scoring 98 next time, then handing in a 90 today might still get you beaten.
That's the cruelty of high-growth stocks.
The mistake isn't necessarily in the present, but in the future not being exciting enough.
Zscaler got sold off not because its performance collapsed, but because the market has doubts about the future.
Q4 guidance wasn't impressive enough, sales team adjustments worried investors about execution, and software stocks overall are in a very sensitive phase. As long as future growth shows even a hint of uncertainty, money runs first.
That's the software stock market now.
You can't just be good; you must be better.
You can't just grow; you must prove growth can continue.
You can't just say AI will bring opportunities; you must prove that AI will turn into orders, revenue, profit, and free cash flow.
So this decline in Zscaler is essentially the market asking one question.
When will your AI security story turn into real money?
But from another perspective, this divergence is interesting.
If a company's earnings were already bad and the stock fell, there's not much to discuss.
But if a company's operational data is still decent, its long-term logic is strengthening, yet the stock is sold off due to short-term expectations, then market divergence may exist.
And real investment opportunities often hide in divergence.
Next, we need to see whether Zscaler's long-term logic has changed.
The answer is: not only has it not changed, but it may have become more important.
Why?
Because the AI era is reshaping cybersecurity.
In the past, enterprise security was like defending a city.
Inside the company was the city, outside was the outside. Firewalls were like city walls, VPNs like city gates. As long as you defended the boundary, you felt safe.
But that world has changed.
Cloud apps, SaaS systems, remote work, API interfaces, third-party plugins, external model calls—they have riddled the old city walls with holes.
Enterprise data is no longer only on company servers; employees are no longer only in offices; apps no longer only run on internal networks.
The boundary is gone.
And the emergence of AI Agents makes this even more complex.
An AI Agent is not an ordinary tool.
It can automatically log into systems, call customer data, generate reports, modify code, send emails, create tickets, and even trigger business processes.
Sounds great, right?
But from a security perspective, it's also frightening.
Because an AI Agent that can autonomously execute tasks is essentially a digital employee that never sleeps.
It needs permissions.
It needs to access data.
It needs to call applications.
It needs to connect different systems.
If permission management is inadequate, data boundaries are unclear, the model is misled by prompt injection attacks, or the AI Agent is exploited by hackers, it can turn from an efficiency tool into a security vulnerability within the enterprise.
That's why zero trust becomes a necessity again in the AI era.
The core logic of zero trust is simple.
Do not trust anyone, any device, any application, or any connection by default.
Every access must be verified.
Every data flow must be inspected.
Every user, device, application, and AI Agent must operate with the least privilege.
This sounds troublesome, but it's exactly what large enterprises are willing to pay for.
Because enterprises fear not trouble, but loss of control.
In the past, hackers had to find doors; now AI can help them scan every window of the whole building.
In the past, enterprises worried about employee account theft; now they also worry about AI Agents being induced to overstep.
In the past, data leaks might be due to employee mistakes; now data could leak from an automated agent transmitting information in milliseconds.
So if your enterprise is still using old VPNs and traditional firewall thinking, it's like using a wooden door to block a tank in the AI era.
That's where Zscaler's value lies.
It's not simply selling a security product, but integrating enterprise access control, application connection, data protection, zero-trust architecture, and AI security capabilities.
That's also why its collaborations with OpenAI and Anthropic are most noteworthy.
Zscaler's partnership with OpenAI isn't just about saying we've integrated AI.
It emphasizes using advanced model capabilities for security development, vulnerability detection, AI Red Teaming, and Agentic SecOps.
In plain language, it's using AI to help security teams find code vulnerabilities faster, identify risks quicker, test system weaknesses more efficiently, and handle security incidents faster.
This is important.
Because the future of security may enter a new stage.
Not human vs. human.
But human + AI vs. human + AI.
In the past, cybersecurity was like two chess players playing.
Now it might become both sides playing with supercomputers.
Attackers can use AI to find vulnerabilities, write malicious code, forge emails, impersonate identities, and quickly test attack paths.
If defenders still rely on manual slow inspection, patching, and response, they'll struggle increasingly.
So security companies must also use AI.
Zscaler's involvement in Anthropic's Project Glasswing has similar significance.
If models like Claude can find software vulnerabilities faster and understand code risks better, that's good for defenders.
But conversely, it means vulnerability discovery speed will increase dramatically.
Security issues in the software world may be amplified by AI.
Previously, a vulnerability might take security researchers days or weeks to discover.
In the future, AI may dramatically speed up vulnerability discovery.
For enterprises, this is both an opportunity and a fear.
Opportunity: they can fix problems faster.
Fear: attackers can also exploit problems faster.
So the long-term demand for AI security is not hyped by marketing, but naturally driven by technological change.
The stronger AI is, the stronger the attacks.
The stronger the attacks, the more defensive budgets cannot be cut.
That's the hardest point in Zscaler's long-term logic.
So where is Zscaler's opportunity?
I think it can be broken into four layers.
First layer: replacing traditional VPNs and firewalls.
This is Zscaler's original core logic.
As enterprises become more cloud-based, employees more distributed, and applications more numerous, traditional boundary security becomes increasingly strained. The replacement of old architecture with zero trust continues; this path is not over.
Second layer: protecting enterprise use of generative AI.
Employees may now use ChatGPT, Claude, Copilot, and various internal AI tools.
What do enterprises fear most?
They fear employees dumping customer lists into these tools.
Or financial data.
Or source code.
Or trade secrets being absorbed by external models.
So enterprises need to know who is using AI, which AI they use, what data is transmitted, and whether there is a risk of leakage.
That's the demand for AI access security and data protection.
Third layer: protecting AI Agents.
This is the bigger future story.
When AI Agents do not just chat but start accessing systems, calling APIs, reading data, and executing processes, enterprises must assign them identities, permissions, boundaries, and audit trails.
Whoever can help enterprises control AI Agent behavior may capture new security budgets.
Fourth layer: AI Red Teaming and automated security operations.
Enterprises must not only defend against external attacks but also proactively test their AI systems for vulnerabilities.
Can the model be attacked via prompt injection?
Can AI Agents overstep permissions?
Is internal data incorrectly called?
Are there hidden flaws in application code?
Can the security team use AI to fix problems faster?
If Zscaler can combine these capabilities, it won't just be a traditional cybersecurity company, but a potential enterprise AI security gateway.
That's its most imaginative aspect.
In the future, companies won't ask whether to use AI.
That question is already settled.
The real question is: how to use AI securely.
Whoever can answer that question may capture the next budget round.
But we must also clearly state the risks.
Zscaler is not without risks, and the market's sell-off this time is not entirely unreasonable.
First risk: short-term guidance.
High-valuation software stocks fear downward growth expectations the most. Once the market senses revenue growth may step down, the stock is easily killed.
Second risk: sales team adjustments.
Cybersecurity software growth heavily depends on sales execution, large customer acquisition, cross-selling, and renewals. Any disruption in sales organization makes the market very sensitive.
Third risk: intense competition.
Zscaler is not alone.
Palo Alto Networks, CrowdStrike, Okta, Cloudflare, Microsoft, Wiz—all are vying for enterprise security budgets.
Every company is talking about AI security; every company wants to be a platform.
Customer money is limited; whoever can truly prove effectiveness will get more budget.
Fourth risk: the AI security story still needs revenue realization.
AI Red Teaming, Agentic SecOps, AI Access Security—these sound advanced, but the market ultimately looks at whether customers are paying, contracts are growing, renewals are improving, and revenue contributions are appearing.
If it's just concept heat without visible conversion in earnings, valuation expansion is hard to sustain.
Fifth risk: valuation and interest rate environment.
Software stocks most fear changes in interest rates and risk appetite.
Even if the company is good, if valuation is too expensive, a market environment shift will compress it.
So Zscaler now is not without a story, but the story is so big that the market starts asking: when will the story become money?
That's my core judgment.
Zscaler is not a stock you can blindly buy.
But it is definitely a core candidate worth putting on the AI security watchlist.
So how does it compare to other AI security stocks?
We can briefly break it down.
Zscaler's keywords: zero trust, SASE, cloud access, data security, AI access control.
CrowdStrike's keywords: endpoint security, threat detection, cloud security, AI security operations.
Okta's keywords: identity management, AI Agent identity, access permissions.
Palo Alto Networks' keywords: platform security, firewall, cloud security, security operations.
Cloudflare's keywords: network edge, application security, Zero Trust, developers and edge network.
So you can't lump all security stocks together.
AI security will benefit the entire track, but each company captures different budgets.
If enterprises worry about AI Agent overstepping, Okta's logic is more direct.
If they worry about endpoint threats and attack detection, CrowdStrike is more direct.
If they worry about cloud access, data flow, VPN replacement, and zero trust, Zscaler is more direct.
If they want a comprehensive platform security, Palo Alto has advantages.
That's why cybersecurity stocks will continue to diverge.
Not all security stocks are the same.
What really matters is which position they occupy in enterprise AI security.
So what should ordinary investors watch next?
I think there are several key indicators for Zscaler.
First, whether ARR growth can stabilize at a high level.
This indicates whether long-term contracts and subscription demand remain strong.
Second, whether the number of large customers and million-dollar customers continues to grow.
For security software, large customers—especially multinational enterprises and large institutions—are most valuable.
Third, whether non-GAAP operating margin and free cash flow continue to improve.
This shows the company is not just burning money for growth but moving toward a more mature business model.
Fourth, whether new products like AI Security, AI Red Teaming, and Agentic SecOps start to show clear revenue contributions.
This determines whether the AI story stays in press releases.
Fifth, whether sales execution issues can be fixed.
If the impact of sales team adjustments persists, short-term stock price may remain under pressure.
Sixth, whether guidance for the next fiscal year can regain market confidence.
Growth stocks ultimately compete not on the past but on the future.
The real opportunity isn't about how much the stock price fell or how good the earnings headline looks.
The real opportunity lies in this: if the market sold off due to short-term guidance, but the medium- to long-term AI security demand continues to rise, then this divergence may become an investment opportunity.
So, back to the original question.
Is Zscaler this time a risk signal or an opportunity window?
My answer: both.
Short-term, it indeed exposed issues with growth expectations, sales execution, and high market demands.
Long-term, it stands at the intersection of zero trust and AI security, a crucial position.
In the AI era, security is not a supporting role but a pass.
Enterprises must solve access control, data protection, identity management, model security, vulnerability detection, and zero-trust architecture to use AI.
If these problems are unsolved, enterprises won't dare to put AI into core systems.
So AI security is not optional; it's part of the infrastructure for AI deployment.
That's the biggest takeaway from Zscaler's earnings.
The market may punish its guidance in the short term, but the long-term theme of AI security has not disappeared.
In fact, the more AI spreads, the more this line is worth watching.
If AI Agents massively enter enterprises in the future, the real beneficiaries won't just be model companies or GPU companies.
Security companies will likely become the hardest link in the next round of AI budgets.
Final summary.
The focus of Zscaler's earnings is not the strong earnings nor the stock decline, but that it pushed to the forefront a line most underestimated in AI software stocks.
Zscaler's opportunity: zero trust is upgrading from a cloud security architecture to an enterprise access control foundation in the AI era.
Zscaler's risks: short-term guidance, sales execution, competitive pressure, valuation volatility—all require market reassessment.
So it's not a buy-and-forget target, but a core AI security observation subject worth continuous tracking.
If you feel this content helped you see Zscaler and the AI security line clearly, remember to like, save, and subscribe.
See you next time.
