Stocks making the biggest moves midday: Zscaler, CrowdStrike, Marvell Technology, Rum Group & more

AI-related equities diverged sharply after industry leaders warned that AI capabilities may be advancing too quickly: Nvidia fell nearly 3%, Intel and AMD each lost 4%, Marvell declined almost 6%, and HPE dropped nearly 9%. In contrast, cybersecurity stocks surged on heightened AI-safety concerns, with Palo Alto Networks up 13%, Zscaler and CrowdStrike each up 15%, and cybersecurity ETFs gaining 6%-10%. Rum Group rallied 18% following a reported six-year, $13.7 billion Anthropic computing-power agreement, while Baldwin Insurance agreed to a $7.7 billion all-cash take-private deal at $32.50 per share.

Analysis

The sharper implication is a rotation within AI rather than a broad reduction in AI spending: firms whose economics depend on rapid accelerator deployment and leveraged hardware turns face both demand-duration and inventory-risk repricing. NVDA, AMD, MRVL, SMCI and HPE are not equally exposed; HPE/SMCI have lower-margin, working-capital-intensive integration models and should de-rate first if customers defer cluster acceptance, while NVDA retains the strongest software and supply-chain control. Over the next 1-3 months, hyperscaler capex commentary and server-rack delivery lead times—not AI-safety rhetoric—will determine whether this is a transient positioning washout or an earnings-risk event.

The cybersecurity move is directionally understandable but likely conflates AI-governance anxiety with near-term security budget creation. PANW has the most credible platform-consolidation and free-cash-flow case, whereas CRWD and ZS require sustained net-new growth to support richer duration multiples; a broad 10-15% one-day move leaves limited immediate upside absent raised billings guidance. The more durable 6-18 month beneficiary of tighter AI controls may be identity, data-security and governance vendors, but the article provides no evidence that enterprise procurement budgets have yet shifted.

Treat the reported RUM contract as an unverified financing-and-execution question, not simply a revenue multiple catalyst. A multi-year compute commitment can require substantial up-front power, GPU and debt capacity, creating customer-concentration and funding risk if contract economics, termination rights and capex obligations are not disclosed. ELMT similarly has strategic optionality, but the key issue is whether government capital is grant-like, dilutive, or tied to return-constraining production commitments; after a vertical move, that distinction dominates valuation.

MARA's relative underperformance highlights that the market is assigning little value to a generic data-center conversion narrative without contracted power monetization. Bitcoin price remains the dominant driver of its equity beta, while conversion capex, power interconnection timing and competition from purpose-built GPU lessors can impair its optionality over 6-18 months.

Key Decisions for Investors

• Initiate a 1-3 month pair: long PANW / short HPE in equal dollar amounts after the opening volatility settles. PANW offers recurring security platform exposure while HPE is more exposed to AI-server order deferrals and margin pressure; reassess if PANW billings guidance weakens or HPE reports improving AI backlog conversion and gross margin.
• Do not chase CRWD or ZS after the gap higher. Set an alert to accumulate PANW on a 7-10% pullback, contingent on unchanged next-quarter RPO/billings guidance; the thesis is invalidated by evidence that AI-security concern is not translating into enterprise budget reallocations.
• Maintain an underweight/short bias in MARA versus a bitcoin proxy such as IBIT over the next quarter. This isolates data-center execution risk from digital-asset beta; cover if MARA discloses a funded, contracted high-performance-computing deployment with economics superior to mining returns.
• Avoid initiating RUM exposure until contract terms, counterparty credit support, incremental power capacity, GPU procurement and financing are independently disclosed. A verified take-or-pay structure with funded capex would warrant reassessment; equity or high-cost debt funding would make the post-move risk/reward unfavorable.
• For energy exposure, prefer EOG over APA on a 1-3 month horizon only while WTI holds above its pre-escalation range; use a 5-7% oil-price reversal as a stop discipline. Geopolitical premium can unwind faster than upstream earnings estimates adjust, making this a tactical hedge rather than a structural allocation.