Zscaler CEO: AI can’t replace mission-critical traffic inspection requiring 100% reliability

At the annual RSA cybersecurity conference, Zscaler (NASDAQ:ZS | ZS Price Prediction) Chairman and CEO Jay Chaudhry sat down to address a question rattling investors: can AI replace what Zscaler does?

The concern wasn’t hypothetical. A release from Anthropic sent cyber stocks lower as investors interpreted it as suggesting AI could simplify or replace certain cybersecurity functions. Chaudhry’s response was direct:

“When you’re talking about mission critical applications, sitting in line, inspecting traffic to make sure bad guys are not trying to reach your application, steal your data, you need a global distributed infrastructure that Zscaler has and 100% reliability. That’s where AI doesn’t really happen.”

His argument: AI can handle tasks like detecting code vulnerabilities, but cannot replicate the global distributed infrastructure and 100% reliability requirement that inline traffic inspection demands. Think of it like air traffic control: you can automate a lot of aviation, but you can’t have the system go down 2% of the time when planes are in the air.

He also flipped the AI threat narrative entirely. Chaudhry has previously discussed the concept of 50 to 100 AI agents per employee — a massive proliferation of non-human entities that each need to be screened and secured. Every one of those agents is a potential attack surface. Where users are today’s weakest link, AI agents become tomorrow’s. That’s not a threat to Zscaler’s business; it’s an expansion of its addressable market.

The numbers back up the demand signal. Enterprise AI usage surged 91% year-over-year across more than 3,400 applications in Q2 FY2026, with data transfers to AI and ML applications up 93%. Zscaler’s Q2 revenue hit $815.75M, up 25.91% year-over-year, with ARR reaching $3.36B growing 25% year-over-year. The company raised full-year FY2026 ARR guidance to $3.730B–$3.745B.

Where We’d Push Back

Chaudhry’s argument is coherent, but it deserves scrutiny. CEOs have an obvious incentive to frame existential threats as tailwinds. The “100% reliability” moat sounds durable today, but AI capabilities are compounding fast. What’s genuinely impossible for an AI-native security architecture to handle in 2026 may look very different by 2028 or 2029.

The stock’s -32.51% year-to-date decline from a 52-week high of $336.99 tells you the market isn’t fully buying the reassurance either. Analysts remain largely constructive, with 39 buy ratings versus 9 holds and zero sells — but the gap between the $234.79 analyst target and the current price near $151.80 reflects real uncertainty.

Chaudhry’s core thesis, that AI creates more attack surface requiring more inspection, not less, is the right framework for thinking about Zscaler’s long-term positioning. But investors should hold that alongside an honest question: how long does “AI can’t do this” remain true in a field where AI capabilities are rewriting the rules every 12 months.