Nasdaq Reclaims 100-Day Average, 10% off Recent Low – What’s Moving?

The Nasdaq Composite pushed back above its 100-day simple moving average on Friday, up 0.6% on the day and more than 10% up from its March 30 close, even as cybersecurity and high-multiple software names suffered steep losses amid renewed fears that advanced AI models could disrupt traditional security business models.

By mid-morning trading, the technology-heavy index showed quite the mix when digging in.

Leading the advance were semiconductor and AI infrastructure names: ARM Holdings up 5.64%, Broadcom gained 4.99%, AMD 4.94% higher, Marvell up 4.82%, and ASML Holding rose 3.01%.

On the opposite end, cybersecurity stocks bore the brunt of selling pressure. Palo Alto Networks tumbled 7.02% to $155.26, its steepest single-day decline in months, while CrowdStrike fell 5.03% to $374.83. Zscaler dropped 5.11% to $115.98, Datadog lost 6.50% to $101.90, and Fortinet declined 4.00% to $77.43. High-valuation software names also struggled, with Atlassian down 4.78% to $56.14 and Intuit off 4.23% to $346.40.

The divergence reflects a sharp rotation within the technology sector, with markets rewarding tangible AI infrastructure plays while punishing companies perceived as vulnerable to AI-driven disruption or valuation compression.

What is driving the split in tech?

The catalyst for Friday’s cybersecurity weakness traces back to Anthropic’s announcement this week that Claude Mythos, a highly capable AI model that the company says has already identified thousands of zero-day vulnerabilities across every major operating system and web browser. Anthropic simultaneously launched Project Glasswing, a defensive cybersecurity initiative bringing together AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan, Microsoft, Nvidia, Palo Alto Networks, and the Linux Foundation.

The model is being withheld from public release and restricted to a small set of trusted partners specifically for defensive security work. According to technical reports, Mythos Preview has demonstrated the ability to discover critical software flaws at scale, including vulnerabilities that have remained unpatched for decades. Anthropic’s own internal red-teaming documentation flagged strategic manipulation and exploit-seeking behaviors in early versions of the model, though the company states these issues were substantially mitigated before the partner preview.

The disclosure has created a double-edged narrative for cybersecurity equities. On one hand, it validates the thesis that AI-native security tools represent the future of enterprise defense and could expand addressable markets. On the other, it raises concerns that ultra-capable AI agents could commoditize portions of vulnerability discovery and monitoring, potentially eroding pricing power for traditional point solutions.

Markets have responded by de-rating many cybersecurity names. Key stocks in the sector are down approximately 20% or more year-to-date, with some names like Zscaler and Atlassian suffering drawdowns exceeding 40% over six months. The weakness stands in stark contrast to the sector’s strong fundamentals, with enterprise security spending holding firm and attack surfaces expanding as AI adoption accelerates.

Meanwhile, semiconductor names have benefited from a “picks and shovels” rotation, with markets favouring companies that supply the physical infrastructure powering AI workloads. Over the past year, ARM Holdings has gained 67.63%, Broadcom is up 114.04%, AMD has surged 183.25%, and ASML has climbed 119.43%. These gains reflect tangible, volume-driven demand from hyperscalers and enterprise customers building out AI compute capacity.

Outlook

The technical picture for the Nasdaq shows an index regaining its footing even as internal breadth remains challenged. The move back above the 100-day simple moving average suggests that large-cap AI hardware and platform names are more than offsetting weakness in a long tail of software and cybersecurity stocks.

For cybersecurity equities, the path forward depends on whether markets come to view AI as a revenue tailwind or a structural disruptor. Bulls argue that enterprises will be forced to increase security budgets and embrace AI-augmented defenses as threats become more sophisticated. The inclusion of leading security vendors like CrowdStrike and Palo Alto Networks in Project Glasswing could position them to monetize this shift through new AI-first security offerings and incident-response tooling.

Bears counter that the same tools capable of securing systems can be weaponized, and that highly capable models being partly held back from public release highlight a threat landscape worsening faster than defensive tooling can be commercialized. There is also uncertainty about whether AI labs and hyperscalers might vertically integrate security services, competing more directly with pure-play vendors over time.

For software names beyond cybersecurity, the weakness reflects a broader repricing of high-multiple SaaS businesses. Markets are shifting from “growth at any price” to “profitable growth,” and concerns that AI-native tools from hyperscalers or AI labs could compress pricing power in observability, collaboration, and workflow automation are weighing on valuations. Until monetization models and competitive dynamics become clearer, the market appears inclined to discount rather than reward exposure to AI-driven uncertainty in these segments.