ProCap Insights · April 15, 2026
Anthropic's Mythos just made 2 cybersecurity stocks far more attractive
Anthropic built an AI model so powerful at finding software vulnerabilities that it refused to release it publicly. Then it hand-picked two names as key partners to deploy it. The market sold both stocks on AI fear and got the direction completely wrong.
What to Know
- CRWD at $395 and PANW at $160 are the two names positioned to capture the Mythos-driven cybersecurity acceleration, with 25-29% upside to consensus price targets. PANW CEO Nikesh Arora backed the thesis with a $10M open-market buy at $147 on the day of the panic selloff.
- Mythos scored 83.1% on CyberGym, 25% above Anthropic's previous best model, and found zero-day vulnerabilities in every major operating system and browser autonomously. Anthropic deemed it too dangerous for public release, proving that cybersecurity spending must accelerate as AI models reach this capability tier.
- PANW reports Q3 earnings in May with revenue guidance implying 28-29% growth, CrowdStrike follows in June with Falcon Flex ARR growing 120% year-over-year, and the Glasswing 90-day report in July could serve as the next major catalyst for both names.
CrowdStrike and Palo Alto Own the AI Cybersecurity Premium, and the Market Has Not Priced It In

Source: Finnhub, company earnings reports, analyst consensus data as of April 14, 2026
The Theme
Anthropic built an AI model so powerful at finding software vulnerabilities that it refused to release it publicly. That single decision reframes the entire cybersecurity investment landscape for the next three to five years.
Claude Mythos Preview, the unreleased frontier model at the center of Project Glasswing, demonstrated autonomous vulnerability discovery at a scale no prior system has approached. It found a 27-year-old vulnerability in OpenBSD, widely regarded as the most secure operating system in existence. It discovered a 16-year-old flaw in FFmpeg, the near-ubiquitous video encoding library that underpins most of the internet's media infrastructure.4,5
On the CyberGym evaluation benchmark, Mythos scored 83.1%, compared to 66.6% for Anthropic's previous best.5
The timing is deliberate. CrowdStrike's 2026 Global Threat Report documented that AI-enabled cyberattacks surged 89% year-over-year, while the average eCrime breakout time collapsed to just 29 minutes.14 The fastest recorded breakout was 27 seconds.
The World Economic Forum's 2026 Global Cybersecurity Outlook found that 94% of executives now identify AI as the most consequential driver of cybersecurity change.13 This is not a theoretical future risk. It is the present reality.
The investment implication is straightforward. The same AI capabilities that scared investors away from cybersecurity stocks on March 27 are the capabilities that make cybersecurity spending an enterprise survival requirement. Global cybercrime costs are projected to exceed $10.5 trillion annually.14
The attack surface is expanding exponentially as AI agents, automated code generation, and machine-to-machine infrastructure proliferate across every enterprise. Every new AI deployment creates new vectors that require security. The companies with the platforms, the data, and the AI-native architectures to defend against Mythos-class threats are about to see a demand acceleration unlike anything since the cloud migration cycle.
The Consensus and Where It Breaks
The dominant narrative in early 2026 was that AI disrupts cybersecurity companies. The iShares Expanded Tech-Software ETF fell broadly as investors assumed AI tools would automate away the need for dedicated security platforms.18,19
On March 27 alone, CRWD dropped approximately 6% and PANW fell 6% in a single session after the Mythos leak. During the Mythos-leak selloff in late March, PANW traded as low as $143.50 intraday, extending a decline of more than 22% from its January open.18
CRWD pulled back below $370 in late March, trading as low as $362 intraday on March 27, before rebounding on the Wolfe Research upgrade and Project Glasswing announcement.22 As of April 14, both stocks remain well below their 2025 highs, with CRWD down approximately 16% YTD and PANW down approximately 13% for the year.1
The thesis driving the selloff was simple and wrong. If AI can find vulnerabilities, why pay CrowdStrike or Palo Alto to do it?
Project Glasswing demolished that thesis in a single announcement. Anthropic, the company that built the most capable cyber AI model in existence, did not offer it as a standalone product. It recruited CrowdStrike and Palo Alto Networks as essential partners because finding vulnerabilities is only half the equation.4,6,7
The other half is remediation, patching, monitoring, and incident response across millions of endpoints in real time, and that is precisely what these two companies do.
Wedbush analysts noted in their March channel checks that cybersecurity budgets are still rising, with enterprise IT budgets up approximately 20% and separate AI budgets adding another 20%.15 About 80% of deals are now being revised upward, a dramatic reversal from the prior year when half of deals saw average discounts near 20%.
Morgan Stanley named both CRWD and PANW among its top five cybersecurity picks, with analyst Meta Marshall assigning price targets implying approximately 29% upside for CRWD (PT $510) and 39% upside for PANW (PT $223).23 The consensus mispricing is that investors treated AI as a headwind for cybersecurity when the data proves it is the most powerful tailwind the sector has seen in a decade.
The Names That Express It
CrowdStrike (CRWD) at $395.27
CrowdStrike is the AI-native endpoint security platform that Anthropic chose to help operationalize Mythos-class vulnerability discoveries across enterprise environments. The company's Falcon platform processes over 2 trillion security events per week, per CrowdStrike's corporate disclosures, giving it the largest real-time threat intelligence dataset in the industry.2
Q4 FY26 earnings confirmed the business is accelerating. Revenue hit $1.31 billion, up 23% year-over-year. Annual recurring revenue reached $5.25 billion with net new ARR of $331 million, up 47% year-over-year.2
Falcon Flex, the company's platform consumption model, grew ARR 120% year-over-year to $1.69 billion, proving that enterprises are consolidating security spend onto CrowdStrike's platform. FY27 revenue guidance of $5.87-5.93 billion came in above the $5.86 billion consensus.2
The stock trades at $395, roughly 30% below its 52-week high of $567.22 The consensus analyst price target sits at approximately $493-505, depending on the aggregator, implying 24-27% upside.20
Morgan Stanley's target is $510. Piper Sandler is at $520.17,23 Among the roughly 50 analysts covering the name, the overwhelming majority rate it a buy or overweight, with approximately 34-38 buy-equivalent ratings and 15 holds.20
Palo Alto Networks (PANW) at $159.60
Palo Alto Networks is the platformization play. CEO Nikesh Arora has spent the past two years building the most comprehensive integrated security platform in the industry, and Anthropic validated that strategy by naming PANW a Glasswing launch partner.4,6
The insider signal here is the loudest in the sector. On March 27, the same day cybersecurity stocks cratered on the Mythos leak, Arora bought 68,085 shares at $146.87 for approximately $10 million.8,9,10,11 It was his first open-market purchase since November 2019.
JPMorgan called it a "substantial vote of confidence."16 The stock has recovered to $160, but Arora's entry price still offers a reference point well below the $207 consensus target.
Q2 FY26 revenue came in at $2.594 billion, up 14.9% year-over-year. Next-Generation Security ARR reached $6.3 billion, growing 33%.3 Q3 guidance implies an acceleration to 28-29% revenue growth.
The platformization thesis is working. Enterprises are consolidating security vendors, and the CyberArk acquisition extends Palo Alto's platform into identity security, while Chronosphere adds AI observability.3
Of 56 analysts covering the stock, 44 rate it a buy or strong buy.21
PANW trades at roughly 42x forward earnings, a clear discount to CrowdStrike's approximately 82x forward earnings multiple, with a comparable growth trajectory and the added conviction signal from its CEO's pocket.1,3
Both Companies Have Beaten Earnings Estimates Consistently, Setting Up for Another Surprise Cycle

Source: Finnhub, company earnings reports. Beat percentages shown above actual EPS bars.
The Counter-Argument
The bull case on CRWD and PANW requires sustained enterprise cybersecurity budget growth and successful integration of Glasswing capabilities. Neither is guaranteed.
The valuation argument is the most obvious pushback. CrowdStrike trades at approximately 82x forward earnings based on the FY27 consensus EPS estimate of $4.81 per LSEG.1 Even after its recent recovery, PANW trades at approximately 42x.
These are growth multiples that assume near-perfect execution for multiple years.
If the broader software selloff deepens or if macro conditions force enterprises to cut IT budgets, both stocks have substantial downside before they approach value territory. The 2022 correction took CRWD from $300 to $100 when growth stocks repriced.
The AI commoditization risk is real, even if overstated. Anthropic's decision to restrict Mythos is a choice, not a permanent structural barrier. As more AI labs develop comparable cyber capabilities, the moat around finding vulnerabilities narrows.
If Google, OpenAI, or open-source alternatives achieve similar CyberGym scores within 12-18 months, the Glasswing partnership loses its exclusivity premium. The $100M in Anthropic usage credits is a subsidy that expires.4,7 After that, the $25/$125 per million token pricing adds to operating costs.
For Palo Alto specifically, the integration risk is material. The CyberArk acquisition was a $25 billion deal. Chronosphere added another $3.35 billion.
Absorbing two major acquisitions while maintaining 28-29% revenue growth and expanding margins is the kind of execution challenge that has tripped up even the best operators.3
PANW's operating margin guidance for FY26 already reflects integration costs.
CrowdStrike carries its own overhang. The July 2024 global outage resulted in customer concessions and remediation costs throughout FY26. While management has called this largely resolved, the reputational damage lingers in competitive deals, and any future reliability incident would be magnified by the heightened scrutiny.2
The regulatory dimension adds uncertainty. The April 7 Treasury meeting involving Secretary Bessent and Fed Chair Powell to discuss AI cyber threats could result in regulatory frameworks that constrain how commercial entities can deploy models like Mythos.24 If policymakers decide that frontier AI cyber capabilities must be government-controlled rather than commercially distributed, the Glasswing partnership model collapses.
Finally, the AI-drives-demand thesis cuts both ways. If AI-powered attacks surge as predicted, a successful major cyberattack on a Glasswing partner would undermine the entire narrative that these companies can keep pace with the threat. The very capabilities that make Mythos valuable for defense make it devastating in the wrong hands, and Anthropic's own CMS misconfiguration that leaked the model's existence is not exactly confidence-inspiring from an operational security standpoint.5
Key Data
| Metric | CrowdStrike (CRWD) | Palo Alto Networks (PANW) |
|---|---|---|
| Price (Apr 14) | $395.27 | $159.60 |
| Market Cap | ~$97B | ~$131B |
| Forward P/E | ~82x | ~42x |
| Revenue Growth (YoY) | 23% | 14.9% (accelerating to 28-29%) |
| ARR / NGS ARR | $5.25B (+24%) | $6.3B (+33%) |
| Latest EPS Beat | Q4 FY26: $1.12 (+1.8%) | Q2 FY26: $1.03 (+7.2%) |
| Consensus Price Target | ~$493-505 (24-27% upside) | $207 (29% upside) |
| Analyst Rating Mix | ~34 Buy / 15 Hold / 1 Sell | 44 Buy / 10 Hold / 2 Sell |
| YTD Performance | -16% | -13% |
| Mar 27 Single-Day Selloff | -6% | -6% (intraday low ~$144) |
| Glasswing Partner | Yes | Yes |
| Key Insider Signal | None recent | CEO bought $10M on Mar 27 |
| 52-Week High | $566.90 | $223.61 |
Sources listed in endnotes.1,2,3,20,21,22
The Glasswing Catalyst Window Runs Through H2 2026 with Five Named Events

Source: Company guidance, Anthropic announcements, EU regulatory calendar. Events after April 14 are forward-looking.
Catalyst Map
May 2026, PANW Q3 FY26 Earnings. Management guided for $2.941-2.945 billion in revenue, implying 28-29% growth. A beat here would confirm the platformization thesis is accelerating and that the Mythos narrative has not dampened enterprise security budgets.3
June 2026, CRWD Q1 FY27 Earnings. CrowdStrike's first quarter with Glasswing partnership revenue potentially flowing through. Falcon Flex ARR trajectory and any early Glasswing-related deals in the pipeline commentary will be closely watched.2
July 2026, Project Glasswing 90-Day Report. Anthropic has committed to publishing progress reports on the initiative. The first major public update on how many vulnerabilities have been patched, how many organizations are actively using Mythos defensively, and whether the model's capabilities are expanding will either validate or undercut the defensive thesis.4,7
August 2026, EU AI Act Enforcement. The European Union's comprehensive AI regulation takes effect. Cybersecurity compliance requirements embedded in the Act will force European enterprises to upgrade security infrastructure, driving incremental demand for platform providers like CRWD and PANW with established EU operations.
H2 2026, Next Frontier Model Releases. OpenAI, Google DeepMind, and Meta are all expected to ship frontier models with enhanced coding capabilities. Each release that approaches Mythos-level vulnerability discovery will intensify the threat narrative and accelerate the cybersecurity spending cycle.
The Bottom Line
The market sold cybersecurity on the Mythos fear trade and got the direction completely wrong. CrowdStrike and Palo Alto Networks are not casualties of AI disruption but the named beneficiaries of the most powerful defensive AI initiative ever launched, trading at 25-29% discounts to consensus targets while their CEOs and Wall Street's sharpest analysts are pounding the table. The thesis holds unless AI models start patching, monitoring, and remediating at the enterprise level without human-operated platforms, and that is not what Anthropic built.
Sources
1. Finnhub API, real-time quotes and company fundamentals, accessed April 14, 2026
2. CrowdStrike Q4 FY26 earnings release, March 2026
3. Palo Alto Networks Q2 FY26 earnings release, February 2026
4. Anthropic, "Project Glasswing" announcement, April 7, 2026
5. VentureBeat, "Anthropic says its most powerful AI cyber model is too dangerous to release publicly," April 7, 2026
6. Fortune, "Anthropic is giving some firms early access to Claude Mythos," April 7, 2026
7. Linux Foundation, "Introducing Project Glasswing," April 7, 2026
8. CNBC, "Palo Alto stock pops as CEO Arora buys stock for first time in years," March 30, 2026
9. Markets Insider, "Palo Alto Networks CEO Nikesh Arora buys $10M in shares," March 30, 2026
10. 24/7 Wall St., "Palo Alto's CEO Said 'AI Must Fight AI' and Then Put $10 Million Behind It," March 31, 2026
11. Morningstar/MarketWatch, "Palo Alto Networks CEO sends a message through his $10 million stock purchase," March 28, 2026
12. Seeking Alpha, "Anthropic's Project Glasswing partners with CrowdStrike, Palo Alto," April 7, 2026
13. World Economic Forum, "Global Cybersecurity Outlook 2026," January 2026
14. Quandary Peak Research, "The New Shockwave: AI-Generated Cyber Attacks," January 2026
15. Wedbush channel checks on cybersecurity spending, March 2026
16. JPMorgan analyst note on PANW insider buy, March 31, 2026
17. Piper Sandler, CRWD upgrade to Overweight with $520 price target, March 2, 2026
18. Investing.com via Yahoo Finance, "Cybersecurity stocks plunge as Anthropic's Claude Mythos leak sparks AI fear," March 27, 2026
19. CNBC, "Cyber stocks fall on report Anthropic is testing a powerful new model," March 27, 2026
20. MarketBeat, CrowdStrike analyst consensus data, April 2026
21. MarketBeat, Palo Alto Networks stock data (market cap $133.18B, 816M shares outstanding), April 2026
22. MarketBeat, CrowdStrike price performance (52-week high $566.90, 52-week low $342.72), April 2026
23. TipRanks/Yahoo Finance, Morgan Stanley cybersecurity stock picks (Meta Marshall, CRWD PT $510, PANW PT $223), March 31, 2026
24. The Hill, "Treasury Sec. Bessent summons bank executives over Anthropic cyber risk" (meeting held Tuesday April 7), April 10, 2026