Cybersecurity market analysis with AI impact
Sidekick AI chat created on TrendSpider.
Compare the largest cybersecurity companies by revenue in an extensive executive report. Weave in recent transcripts, notable Segments & KPIs, news, etc. Narrate the rise of AI agentics across the Internet and how it's reshaping the cyber sector.
I'll build this properly — parallel deep-dives on each major cyber name, plus a market-wide transcript search on the agentic-AI theme so the narrative is grounded in what management teams are actually saying.
Initial batch partially failed. Retrying the missing tickers.
Let me process the data
Let me process the data
Let me process the data
Let me process the data
Let me process the data
Let me process the data
Let me process the data
Let me process the data
Let me process the data
Let me process the data
2 tools called
TTM sums: FTNT = 2047.9+1849.6+1905.0+1724.9 = 7527.4M (from segment detail). CRWD = 1471+1400+1310+1200 = 5381M. GEN = 1336+1283+1240+1220 = 5079M. ZS = 898.2+850.5+815.8+788.1 = 3352.6M. OKTA = 805+765+761+742 = 3073M. CHKP = 673.6+668.4+744.9+677.5 = 2764.4M. NET = 696.1+639.8+614.5+562.0 = 2512.4M. RBRK = 427.3+387.1+377.7+350.2 = 1542.3M. SAIL = 308.8+280.1+294.6+281.9 = 1165.4M. S = 292.0+276.7+271.2+258.9 = 1098.8M. TENB = 268.5+262.1+260.5+252.4 = 1043.5M. NTSK = 220.5+201.6+196.3+184.2 = 802.6M. QLYS = 182.2+175.6+175.3+169.9 = 703.0M. VRNS = 180.0+173.1+173.4+161.6 = 688.1M. NTSK YoY = 220.5/170.8 − 1 = +29.1%.
YTD computed from weekly closes: 29 Dec 2025 → 24 Sep 2026. OKTA 83.64 → 206.64 = +147.1%. PANW 179.37 → 389.92 = +117.4%.
Product YoY = 773.0/508.9 − 1 = +51.9%. Geography Q2'26: EMEA $871.7M > Americas $808.2M > APAC $368.0M.
NET $100K+ customers: 4,698/3,712 − 1 = +26.6%. Gross margin Q2'24 77.80% → Q2'26 71.76% = −604bps.
Current P/S = real-time market cap ÷ TTM revenue. CRWD 268.8/5.381 = 49.95x. NET 125.5/2.512 = 49.96x. PANW 321.7/11.50 = 27.97x. FTNT 131.1/7.527 = 17.42x. RBRK 24.0/1.542 = 15.56x. OKTA 35.9/3.073 = 11.68x. SAIL 12.2/1.165 = 10.47x. ZS 35.0/3.353 = 10.44x. NTSK 7.6/0.803 = 9.47x. QLYS 6.4/0.703 = 9.10x. VRNS 5.6/0.688 = 8.14x. S 8.4/1.099 = 7.64x. CHKP 14.4/2.764 = 5.21x. TENB 4.0/1.044 = 3.83x. GEN 15.7/5.079 = 3.09x. Price vs SMA: OKTA 206.64/157.24 − 1 = +31.4%; 206.64/108.63 − 1 = +90.2%. CHKP 136.29/147.47 − 1 = −7.6%. GEN 23.07/28.55 − 1 = −19.2%.
Cybersecurity Sector — Executive Report
As of 24 Sep 2026, 16:51 New York time. All figures from TrendSpider data pulls; every computed number shows its arithmetic.
A note on method before anything else: I attempted to run parallel deep-research sub-agents across 12 tickers and the sub-agent layer failed repeatedly (only OKTA returned). I rebuilt the report from direct data pulls — fundamentals, revenue breakdowns, earnings, transcripts, news, prices — so the coverage is broad and the numbers are solid, but per-name insider/13F detail and full analyst-rating counts are thinner than they would otherwise be. Where I lack data I say so rather than fill it in.
Two structural facts that reshape the peer set this year:
- CYBR no longer exists as a standalone. Palo Alto closed the CyberArk acquisition in early fiscal Q3 2026 and rebranded it Idera. There is no news or price feed for CYBR. It is now PANW's third reported platform ($1.26B FY26 revenue).
- Two new public names matter now: NTSK (Netskope) and SAIL (SailPoint), both of which are among the loudest agentic-AI stories in the sector.
1. Scale ranking — the largest cybersecurity companies by revenue
Ranked by trailing-twelve-month revenue. TTM = sum of the last four reported quarters.
| Company | TTM revenue | Latest qtr rev / YoY | Mkt cap |
|---|---|---|---|
| PANW Palo Alto Networks | $11.50B (FY26) | $3.41B (31 Jul 26) / +34.5% | $321.7B |
| FTNT Fortinet | $7.53B | $2.05B (30 Jun 26) / +25.6% | $131.1B |
| CRWD CrowdStrike | $5.38B | $1.471B (31 Jul 26) / +25.8% | $268.8B |
| GEN Gen Digital | $5.08B | $1.336B (03 Jul 26) / +6.3% | $15.7B |
| ZS Zscaler | $3.35B | $898.2M (31 Jul 26) / +24.9% | $35.0B |
| OKTA Okta | $3.07B | $805.0M (31 Jul 26) / +10.6% | $35.9B |
| CHKP Check Point | $2.76B | $673.6M (30 Jun 26) / +1.3% | $14.4B |
| NET Cloudflare | $2.51B | $696.1M (30 Jun 26) / +35.9% | $125.5B |
| RBRK Rubrik | $1.54B | $427.3M (31 Jul 26) / +37.9% | $24.0B |
| SAIL SailPoint | $1.17B | $308.8M (31 Jul 26) / +16.8% | $12.2B |
| S SentinelOne | $1.10B | $292.0M (31 Jul 26) / +20.6% | $8.4B |
| TENB Tenable | $1.04B | $268.5M (30 Jun 26) / +8.6% | $4.0B |
| NTSK Netskope | $803M | $220.5M (31 Jul 26) / +29.1% | $7.6B |
| QLYS Qualys | $703M | $182.2M (30 Jun 26) / +11.0% | $6.4B |
| VRNS Varonis | $688M | $180.0M (30 Jun 26) / +18.3% | $5.6B |
The headline structural change: PANW has pulled decisively away. At $11.5B FY26 revenue growing 34% it is now roughly 1.5x Fortinet and 2.1x CrowdStrike, having bought its way from a network-security vendor into a four-front platform (network, Cortex/observability, identity, AI security). Growth at that scale is acquisition-fed — CyberArk and Chronosphere closed within FY26 — so treat the 34% as reported-not-organic. CFO Deepak Galecha framed organic as "bookings growth accelerated for the second consecutive quarter on a pro forma basis." He did not quantify the pro-forma rate. That omission is doing a lot of work.
2. The narrative: agentic AI is now the sector's operating premise, not its marketing garnish
I ran a semantic sweep across every indexed earnings-call transcript looking for cybersecurity-specific agentic AI discussion. The density of hits is the finding: at least 15 security-adjacent companies are now building their entire equity story around AI agents, and they are converging on the same three-part thesis. Here is the arc, in management's own words.
Act I — "Agents broke out of the sandbox." The industry now has shared reference incidents, and every company cites the same ones.
PANW CEO Nikesh Arora, 01 Sep 2026: "This summer served as a wake-up call, as rogue agents compromised environments at several frontier AI labs. In one notable instance, an agent escaped its sandbox and exploited system vulnerabilities because its access had never been properly restricted. At its core, this represents a fundamental identity crisis for the enterprise."
ZS CEO Jay Chaudhry, 03 Sep 2026: "the recent Hugging Face incident where a swarm of agents went to extreme lengths to break out of a sandbox training environment, get onto the corporate network, and move laterally to conduct a sophisticated attack."
S CEO Tomer Weingarten, 27 Aug 2026: "Agents don't just generate text; they execute code, call APIs, handle credentials, and interact with infrastructure—often finding execution paths their developers never intended."
NTSK management: "An agent does not need to be prompted to be a bad actor. Instead, a stated task, enough autonomy, and an environment that can be circumvented can be enough to do damage."
Act II — "AI found the vulnerabilities faster than you can patch them." This is the demand unlock, and it has a name in the industry now: the "Mythos moment" (Anthropic's cyber-capable model).
Arora: "legacy flaws and persistent misconfigurations that once took months for a human to uncover are now exploited in minutes... there is no longer anywhere to hide for our customers." And the commercial translation: "I've strived for eight years to go and get CEOs' interest in cybersecurity. I couldn't. But Dario did a phenomenal job by having Mythos. Because every CEO now wants to talk about, what does this mean to us?"
Chaudhry, quantifying it: "Over 90% of organizations had AI applications, models, and other servers exposed to the Internet, and more than one third had known exploitable vulnerabilities... You cannot win a patching race against AI."
Arora also put a number on the TAM implied: "There's approximately 1 trillion of global cybersecurity debt that must be modernized." That $1T figure is a CEO's rhetorical framing, not a bottoms-up estimate. Treat it accordingly.
Act III — the land grab, and this is where the companies actually diverge. Four distinct claims to the agentic control plane:
| Claim staked | Companies | Core argument |
|---|---|---|
| Identity is the control plane | OKTA, SAIL, PANW (Idera), CRWD | Every agent needs a scoped, revocable, auditable identity tied to a human owner |
| Runtime is the control plane | S, CRWD, PANW | Governance says what an agent should do; runtime sees what it did |
| Inline network/proxy is the control plane | ZS, NTSK, NET | Only the inline exchange can enforce policy on agent traffic in real time |
| Data + recovery is the control plane | VRNS, RBRK | Agents attack data; you need classification, guardrails, and the ability to undo |
The most pointed exchange in the whole dataset is ZS taking direct aim at the identity camp: "We are often asked why our solutions are needed alongside Identity for AI security. While Identity solutions answer who is requesting access, our inline exchange determines what that user or agent should be allowed to do and enforces that policy in real time. Put simply, Identity is only the starting point for securing AI."
S makes the mirror-image attack on the governance camp: "Governance defines what an AI agent is permitted to do, but runtime is where the actions are executed... monitoring is not the same as security." (SAIL uses almost the identical line in reverse: "Simply watching an agent go rogue is just a dashboard of active breaches.")
Read this honestly: four vendors cannot all own the same control plane. The market is currently paying all of them as if they will. That is the single most important thing in this report.
3. The monetization question — and where the numbers actually exist
This is the part most of the sector is still hand-waving. Here is who has produced a hard AI number and who has not.
| Company | Disclosed AI metric | Assessment |
|---|---|---|
| PANW | Prisma AIRS crossed $100M ARR in 4 quarters of GA, 800+ customers — "the fastest-scaling product in the history of Palo Alto Networks" | Real, scaled, fastest hard proof point in the sector |
| SAIL | AI-driven ARR "crossed $70M", >30% of Q2 net new ARR; $100M FY target | Real, but definition is loose (bundles human + non-human) |
| CRWD | Charlotte usage +6x YoY, ARR more than tripled; AIDR +5x QoQ | Growth rates without a base = unverifiable |
| S | AI security ARR (Prompt + Purple) tripled YoY; "next nine-figure ARR category" | "Next" nine-figure means it is not nine-figure yet |
| ZS | Security-for-AI bookings +50% sequentially; AI pipeline +75% QoQ | Bookings and pipeline, not ARR. Deliberately soft |
| TENB | Hexa: >80% of users submit a prompt, >90% of recommended actions accepted | Engagement metrics, zero revenue disclosure |
| RBRK | "over 15 and growing" paying Agent Cloud customers; explicitly no ARR contribution guided this year | Refreshingly honest that it is immaterial |
| OKTA | None — CEO: "it's still too small to show up in the numbers right now"; CFO: "for FY27 we don't think it's going to be material" | The most honest disclosure in the group |
The asymmetry here is the trade. OKTA is up ~147% YTD on a thesis its own CFO says will not appear in revenue until FY28. PANW is the only company with a scaled, dollar-denominated AI product, and it is up ~117%. The market is not discriminating between proven and promised monetization. It will eventually.
4. The pricing-model problem nobody has solved
If agents multiply while human seats flatten, per-seat security pricing captures almost none of the value created. Two CEOs were asked this directly, and both answers are weak.
OKTA CEO Todd McKinnon: "our pricing model is per user. So if you want to buy Okta for AI Agents, it's an uplift to your per-user charge." Then, conceding the objection: "the first thing everyone says is, well, that's crazy. Seats are going away and you got to charge per agent. That all may be true, but the way customers are using agents now and the way they want to buy is per user." Okta's own 10-K names it as a risk: competitors may "adopt preferable pricing models for evolving technologies, such as AI agents."
CRWD CEO George Kurtz, asked about fewer knowledge-worker seats: "industry stats is that each knowledge worker will have 90 AI agents. So even if the mix moves around, we have a massive opportunity to protect AI agents." Note what that is: an assertion that volume offsets the seat decline, not a pricing mechanism.
The vendors who have partially escaped the trap:
- ZS is furthest along: "our non-seat-based, metered-usage solutions delivered approximately 30% of new and upsell ACV. ARR tied to these offerings grew more than 100% year over year." AI security products are "essentially consumption based, linked to queries—essentially tied to token consumption."
- CRWD Falcon Flex / S Flex / ZS Z Flex are pool-of-credit models that sidestep per-seat mechanics without being true consumption pricing. Flex is now $1.69B of CRWD's ARR (+120% YoY, 1,600+ customers), >10% of SentinelOne ARR, and $1.7B FY26 TCV at Zscaler.
- OKTA is scaffolding for it but not enforcing: "we actually have a cap of agent SSO that we're actually not going to enforce right away, but we're putting the framework and the scaffolding in there to have a consumption-based pricing eventually."
Verdict: the seat-compression risk is real and mostly unpriced. Zscaler is the best-positioned on pricing architecture and, notably, the cheapest large-cap in the group on sales. That is not a coincidence the market has yet worked out.
5. Segments and KPIs — where the growth actually sits
PANW — new three-platform disclosure (FY26, first time given)
| Platform | FY26 revenue | FY26 growth | FY27 guided |
|---|---|---|---|
| Network & AI Security | $8.35B | +17% | low double digits |
| Cortex (SecOps + observability) | $1.92B | +25% | ~+30% |
| Idera (ex-CyberArk identity) | $1.26B | +21% | ~$1.5B, high-teens to 20% pro forma |
| KPI | Q4'26 | Q2'26 | Q4'25 |
|---|---|---|---|
| NGS ARR | $9.1B (+63%) | $6.3B | $5.6B |
| RPO | $21.2B (+34%) | $16.0B | $15.8B |
| Net new platformizations | ~220 (record) | — | — |
| XSIAM ARR | >$700M (+70%) | — | — |
| Observability ARR | >$500M (2x since Q2 close) | — | — |
The one number to be suspicious of: FY27 NGS ARR guidance is +22–23%, versus +63% just reported. Management disclosed why — "our fiscal year '26 net new NGS ARR included a nine-figure benefit from a large LLM customer migrating to Chronosphere from an incumbent provider." A nine-figure single-customer ARR event is not a run rate. The 63% headline is substantially a one-off, and the guide admits it. Also note the gross-margin bleed: 74.8% in Q4'26, down 100bps YoY, with management guiding cloud hosting costs to grow faster than revenue in FY27 plus memory/storage inflation on hardware. That is what produced "Palo Alto Stock Suffers Worst 2-Day Drop in 30 Months" (02 Sep 2026) on an otherwise clean beat-and-raise.
CRWD — the cleanest KPI set in the sector
| KPI | Q2'27 | Q4'26 | Q2'26 |
|---|---|---|---|
| ARR | $5.8B | $5.25B (+24%) | $4.7B |
| Net new ARR (Q4'26) | — | $331M (+47%, all-time record) | — |
| RPO | $10.7B | $9.0B | $7.2B |
| Net retention / gross retention | — | 115% / 97% | 111% / 97% |
| 8+ module adoption | 26% | 24% | 23% |
| Flex ARR | — | $1.69B (+120%) | — |
Sub-businesses at Q4'26: Next-Gen SIEM $585M ARR (+75%), Cloud >$800M (+35%), Next-Gen Identity $520M (+34%), and endpoint accelerating two consecutive quarters — which is the single most surprising datapoint in the sector, because endpoint was supposed to be the commoditized part. Kurtz attributes it to AI: "Our sensors detected more than 1800 distinct AI applications running on enterprise devices representing nearly 160 million unique application instances." GAAP operating margin is still −2.3% in Q2'27 versus 25% non-GAAP in Q4'26 — a ~27-point stock-comp gap. That gap is the largest in the peer group and deserves more scrutiny than it gets.
FTNT — the profitability outlier, and the most under-discussed inflection
| Segment | Q2'26 | Q2'25 | YoY |
|---|---|---|---|
| Product revenue | $773.0M | $508.9M | +51.9% |
| Security subscription | $729.7M | $644.4M | +13.2% |
| Technical support & other | $545.2M | $476.7M | +14.4% |
| Total billings | $2.4B | $1.8B | +33% |
| Total RPO | $7.7B | $6.6B | +17% |
Product revenue up 52% YoY is the firewall refresh cycle arriving with force, and it is why Fortinet prints a 33.7% GAAP operating margin and 29.6% GAAP net margin — the best real profitability in the sector alongside Check Point and Qualys. Fortinet also beat Q2 revenue by +8.65%, the largest revenue surprise of any name in the group. Caveat I must flag: the most recent transcript available for FTNT in this dataset is 06 May 2025 (Q1 FY2025), so I have no current management commentary on AI agents for Fortinet. That is a genuine data gap, not an absence of strategy.
ZS — best pricing architecture, decelerating guide
| KPI | Q4'26 | Q1'26 |
|---|---|---|
| ARR | $3.8B (+25%) | $3.2B |
| Net new ARR ex-Red Canary | $232M (+17%) | — |
| $1M+ ARR customers | 785 (+18%) | 698 |
| $100K+ ARR customers | 4,182 (+20%) | 3,754 |
| RPO | $7.4B (+27%) | $5.9B |
| Zero Trust Everywhere enterprises | >950 | ~350 (FY25 exit) |
Net new ARR growth re-accelerated 7% (FY25) → 10% (1H26) → 17% (Q4'26). Then FY27 ARR is guided to +16.6–17.4%, versus 25% just delivered. Two admitted drags: a sales-leadership transition landing in 1H27, and Red Canary contributing zero net new ARR. Also: a 3% workforce restructuring ($30–33M charge) and capex guided to "low teens as a percentage of revenue" from 8% — memory inflation hitting their own data centers. Zscaler is the one large-cap here trading below its own growth rate on sales (10.4x TTM sales on 25% growth).
S — improving, but the ARR math is the tell
Record Q2 net new ARR of $56M, up only +4% YoY, against revenue +21% and RPO +45%. Analyst Josh Tilton pushed CFO Sonali Parekh on exactly this and got no upgrade to the full-year net-new framing — only "we still do absolutely expect our net new ARR to grow year over year." Meanwhile $100K+ customers grew 13% versus 17% last quarter. Management explained it as deliberate upmarket mix. That explanation is plausible and also conveniently unfalsifiable. Real progress elsewhere: record 10% non-GAAP operating margin (+820bps), S&M down 900bps to 34% of revenue. But FY27 non-GAAP EPS was cut to $0.30–0.32 from $0.32–0.38, which is why the stock fell on 28 Aug despite the beat, and why Bernstein downgraded to Market Perform on 17 Sep.
NET — best top-line, worst margin optics
Revenue +35.9% YoY, NRR 120%, $100K+ customers 4,698 (+26.6%), RPO $2.7B. But GAAP operating margin is −29.6% in Q2'26 versus −13.1% a year earlier, and gross margin has fallen six straight quarters from 77.8% (Q2'24) to 71.8%. That is the cost of being the inference/edge layer. At 50x TTM sales it is priced for flawless execution.
SAIL — loudest agentic story, softest revenue
ARR $1.231B +25%; SaaS ARR $847M +36%; net new SaaS ARR $66M +34%; SaaS was 97% of net new ARR; AI-driven ARR >$70M, >30% of net new ARR; customers adopting AI solutions increased annual spend by over 60%. Discovery stat that sells the product: at one Fortune 500, within a week of POC they found "more than 10,000 unknown agents." And the research claim: "97% of AI agents now have access to sensitive enterprise data, while only 21% of organizations surveyed say they're highly confident in their ability to manage that risk."
But Q2 revenue of $308.8M missed consensus $310.2M (−0.47%) and FY27 guidance was merely affirmed, not raised. The 11 Sep headline says it precisely: "SailPoint Delivers Solid Q2, But AI Monetization Still Not Flowing to Revenues." ARR +25% against revenue +16.8% is the gap to watch — it resolves either into future revenue or into a definitional problem.
GEN and CHKP — the melting-ice-cube question
GEN: revenue +6.3%, 81M total paid customers, but the stock is −11.5% YTD and fell from $29.01 to $23.07 in the week of 21 Sep. On 24 Sep it "reportedly approached GoDaddy about a potential takeover." A consumer-security company at 3.1x sales bidding for a domain registrar is not an agentic-AI story; it is a capital-allocation question. CHKP: +1.3% revenue growth, operating margin compressing from 36.1% (Q4'24) to 27.5%, stock −24.7% YTD and below its 200-day SMA of $147.47 — the only name in the group in a confirmed downtrend. These two are the control group proving the AI bid is selective, not indiscriminate.
6. Earnings scorecard — the most recent print for each
| Company | Date | EPS act/est (surprise) | Rev act/est (surprise) |
|---|---|---|---|
| PANW | 01 Sep 26 | 1.02 / 0.88 (+15.9%) | $3.41B / $3.35B (+1.75%) |
| ZS | 03 Sep 26 | 1.19 / 0.88 (+35.2%) | $898.2M / $877.5M (+2.36%) |
| CRWD | 26 Aug 26 | 0.31 / 0.24 (+29.2%) | $1.471B / $1.44B (+2.25%) |
| FTNT | 29 Jul 26 | 0.90 / 0.73 (+23.3%) | $2.05B / $1.9B (+8.65%) |
| NET | 06 Aug 26 | 0.29 / 0.21 (+38.1%) | $696.1M / $666.2M (+4.49%) |
| RBRK | 27 Aug 26 | 0.22 / −0.04 (+650%) | $427.3M / $396.3M (+7.81%) |
| S | 27 Aug 26 | 0.08 / −0.21 | $292.0M / $290.2M (+0.60%) |
| NTSK | 02 Sep 26 | −0.03 / −0.11 (+72.7%) | $220.5M / $214.2M (+2.95%) |
| TENB | 29 Jul 26 | 0.51 / 0.40 (+27.5%) | $268.5M / $264.9M (+1.36%) |
| VRNS | 28 Jul 26 | 0.04 / −0.05 | $180.0M / $176.9M (+1.75%) |
| QLYS | 04 Aug 26 | 1.98 / 1.68 (+17.9%) | $182.2M / $178.7M (+1.96%) |
| GEN | 06 Aug 26 | 0.71 / 0.68 (+4.4%) | $1.336B / $1.31B (+2.04%) |
| SAIL | 09 Sep 26 | 0.09 / 0.08 (+12.5%) | $308.8M / $310.3M (−0.47%) |
| CHKP | 30 Jul 26 | 2.55 / 2.35 (+8.5%) | $673.6M / $675.3M (−0.25%) |
Only two revenue misses in fourteen prints — SAIL and CHKP. Every single company beat on EPS. When an entire sector beats EPS by 15–35% simultaneously, that is a story about conservative guidance and cost discipline, not about a demand shock. Note also that every EPS figure for the loss-makers is non-GAAP; S reports +$0.08 non-GAAP EPS against a −31.99% GAAP net margin.
7. Valuation and price action
Real-time prices, 24 Sep 2026 close, with current P/S computed against TTM revenue above.
| Company | Price | P/S (TTM) | Op margin (GAAP, latest) | vs 50d SMA | vs 200d SMA | YTD |
|---|---|---|---|---|---|---|
| CRWD | $259.67 | 50.0x | −2.3% | +22.7% | +73.7% | +129.0% |
| NET | $358.82 | 50.0x | −29.6% | +20.3% | +55.4% | +83.1% |
| PANW | $389.92 | 28.0x | +5.0% | +9.5% | +59.6% | +117.4% |
| FTNT | $178.67 | 17.4x | +33.7% | +10.8% | +54.4% | +129.4% |
| RBRK | $113.80 | 15.6x | −16.8% | +24.2% | +60.2% | +50.8% |
| OKTA | $206.64 | 11.7x | +13.3% | +31.4% | +90.2% | +147.1% |
| SAIL | $22.12 | 10.5x | −19.1% | +21.3% | +37.6% | +16.7% |
| ZS | $214.64 | 10.4x | −1.7% | +24.1% | +28.0% | −2.7% |
| NTSK | $18.57 | 9.5x | −40.7% | +26.7% | +43.8% | +9.9% |
| QLYS | $182.94 | 9.1x | +34.0% | +7.4% | +44.3% | +39.6% |
| VRNS | $47.95 | 8.1x | −22.6% | +7.7% | +42.8% | +49.7% |
| S | $23.98 | 7.6x | −31.1% | +14.6% | +46.0% | +63.8% |
| CHKP | $136.29 | 5.2x | +27.5% | +2.9% | −7.6% | −24.7% |
| TENB | $35.88 | 3.8x | +4.6% | +1.9% | +34.7% | +57.9% |
| GEN | $23.07 | 3.1x | +33.2% | −19.2% | −6.9% | −11.5% |
Thirteen of fifteen names are above both their 50-day and 200-day moving averages, most by 30–90%. That is a fully extended sector, not a selective one. The 18 Sep news cycle named it — "Cybersecurity Rally Hits a Valuation Reality Check" — and Bernstein used the week of 17 Sep to downgrade three names simultaneously (PANW to Market Perform PT $351, S to Market Perform PT $25, OKTA to Market Perform PT $174) while raising price targets on all of them. Downgrading on valuation while raising targets is an analyst telling you the fundamentals are fine and the price is not.
The most glaring dislocation in the table: ZS at 10.4x sales on +25% growth versus CRWD at 50x on +26% growth. Nearly identical growth, a 5x multiple gap. Part of that is justified — CrowdStrike's 115% NRR, 97% gross retention and $10.7B RPO are better than Zscaler's, and CRWD's net new ARR is accelerating harder. But a 5x gap is not 5x of business quality. Zscaler carries the sales-transition and Red Canary-churn discount, and it is also the only company that has actually restructured its pricing for a non-seat world.
The valuation question I would put to anyone bullish on the group: FTNT prints a 33.7% GAAP operating margin, grew product revenue 52%, and trades at 17.4x sales. CRWD prints a −2.3% GAAP operating margin and trades at 50x. The premium is being paid for narrative positioning on agentic AI, and Fortinet's absence from that narrative in this dataset may be an information gap rather than a business gap.
8. Recent catalysts — the last four weeks
The sector has become a macro instrument, which is new and important.
- 14 Sep 2026 — "OpenAI, Anthropic Call for AI Slowdown: Chips Sink, Cybersecurity Stocks Jump." S rallied 16% (second-biggest day ever), OKTA hit a 52-week high. Cybersecurity is now traded as the hedge against, and the beneficiary of, the AI trade simultaneously.
- 18 Sep 2026 — reports that hackers used Claude to access OpenAI's internal ChatGPT code sent the whole group lower. Same catalyst class, opposite sign.
- 23 Sep 2026 — cyber bucked a selloff with the 10-year at a 19-year high; CRWD hit a 52-week high of $260.64.
- Frontier-lab partnerships are now table stakes. PANW is "the first certified commercial partner for MITOS 5" and launched Unit 42 Continuous Frontier AI Defense (22 Sep). CRWD brought OpenAI GPT-5.6 Cyber into Falcon and launched Falcon Guardian for AI-agent security (01–02 Sep). ZS launched Agentic SecOps with Anthropic and OpenAI agents (09 Sep). NET partnered with OpenAI Daybreak models (03 Sep). S expanded Wayfinder Frontier AI Services on Daybreak (03 Sep). TENB integrated Claude Mythos 5 into Tenable One (08 Sep). OKTA is "the first identity provider supporting Enterprise Managed Auth for MCP connectors" with Anthropic.
- Consolidation continues: PANW closed Console (01 Sep, AI-native agentic ops) plus Embrace; CRWD closed Signal AI (SGNL) and Seraphic; OKTA closed Promiso; SAIL acquired Intro Security; ZS owns Red Canary; GEN reportedly approached GoDaddy (24 Sep).
- TENB raised an upsized $725M convertible at 0.25% / $44.84 conversion (11 Sep) — financing into strength, and dilutive above $44.84.
Note on the 22 Sep "ShinyHunters claims to have breached the FBI" item: unconfirmed social-media reports, FBI has not commented. I am including it only because it moved sector sentiment, not as fact.
9. Where I come out
Best positioned on proof, not promise — PANW. It is the only company with a scaled dollar-denominated AI product ($100M+ Prisma AIRS ARR in four quarters), it owns the identity layer via Idera/CyberArk, and its platformized cohort runs >120% NRR. The honest caveats: 34% growth is acquisition-fed, the FY27 NGS ARR guide of +22–23% versus +63% reported tells you how much of FY26 was a single LLM migration, and gross margin is structurally bleeding from SaaS mix plus memory costs. At 28x sales it is not cheap, but it is the least faith-based of the mega-caps.
Best risk/reward if you believe the inline-enforcement thesis — ZS. 10.4x sales, +25% ARR, the only genuinely consumption-priced portfolio in the group, and the cheapest large-cap. The FY27 guide of +17% is the reason it is cheap, and the sales-leadership transition is a real execution risk in 1H27.
Best actual business economics — FTNT and QLYS. 33.7% and 34.0% GAAP operating margins respectively, at 17.4x and 9.1x sales. Both are under-represented in the agentic narrative, which is exactly why they are cheaper.
Most exposed to the valuation unwind — OKTA, CRWD, NET. OKTA is +147% YTD and +90% above its 200-day on a thesis its CFO says is immaterial to FY27. CRWD and NET are both at 50x sales with negative GAAP operating margins. None of that is a prediction — it is a statement about how much has to go right.
Structurally challenged — CHKP and GEN. +1.3% and +6.3% growth, margin compression, both below their 200-day SMAs, both negative YTD. They are the evidence that the AI bid is discriminating.
And the honest bear case for the whole sector, in the words of the people selling it: every management team claims their layer is the control plane for agents. Four different layers cannot all be the control plane. The revenue from securing agents is, with one exception, either undisclosed, expressed as a growth rate without a base, or explicitly called immaterial. Seat-based pricing — which is how most of these companies bill — is structurally misaligned with a world where agents outnumber employees 90-to-1 (CrowdStrike's own figure). The sector is being paid today for a resolution of all three of those problems that has not happened yet.
Data gaps I did not paper over: the deep-research layer failed, so per-company insider-trading and 13F detail is absent except for OKTA (net ~−$39.5M insider selling over six months, CFO $12.9M on 02 Sep, CEO $10.1M on 08 Jul, against $267K bought). The most recent FTNT transcript available is from May 2025, so I have no current Fortinet management commentary on agentic AI. CYBR has no independent data — it is PANW's Idera platform now. GEN's revenue breakdown does not populate its Consumer Security / Identity Protection lines. Full analyst buy/hold/sell counts were only obtainable for OKTA.
This is research, not trading advice. I can't make decisions for you, and nothing above is a recommendation to buy or sell any of these securities.