5 big analyst AI moves: Palo Alto and Reddit get fresh Buy calls, AMD upgraded

Analysts turned constructive across several AI-linked stocks: Evercore added Arista ahead of May 5 earnings on expectations for a revenue beat versus $2.61B consensus and a Q2 guide above $2.78B. Berenberg initiated Palo Alto at Buy with a $215 target, arguing the AI-driven selloff is overdone, while DA Davidson upgraded AMD to Buy and lifted its target to $375 on a structural rise in CPU demand. KeyBanc also raised Intel’s target to $110 after strong Q1 results, and D.A. Davidson started Reddit at Buy with a $200 target, citing AI data monetization and underperformance as an entry point.

Analysis

The common thread is not “AI demand” in the abstract, but a re-rating of the picks-and-shovels layer where workloads migrate from training-heavy GPU spend to always-on inference, orchestration, and security. That favors vendors with switching costs and installed-base leverage: networking can monetize the bandwidth step-up, CPUs reclaim wallet share in agentic workloads, and cybersecurity expands its TAM as identities, agents, and machine-to-machine traffic proliferate. The market is still underappreciating how quickly these second-order spends compound once enterprises move from pilots to production.

Intel’s surprise is the most important signal because it changes the capex mix, not just one name’s earnings power. If CPU demand is inflecting alongside GPUs, the winners are likely the suppliers that can price both compute and interconnect content into the same server build, while the losers are single-product exposure names that depended on a GPU-only narrative. That also creates a delayed benefit for component vendors, back-end packaging, and enterprise infrastructure software that typically lags by 1-2 quarters after the initial compute surge.

Cybersecurity is the highest-quality compounding story here, but the bull case is increasingly about monetization efficiency rather than headline growth. AI-driven attack surfaces support spend, yet the real upside comes from platform consolidation: when customers standardize, renewal economics and attach rates inflect faster than budget growth. The risk is not demand collapse; it is a digestion period if CIOs pause after large platform migrations, which can compress near-term billings even while long-term ARR improves.

The contrarian read is that the market may be over-paying for duration in the most obvious AI beneficiaries while underpricing the laggards that gain from workflow breadth. If agentic AI truly narrows the GPU-to-CPU ratio and increases security complexity, then the bigger winners over the next 6-12 months may be names that look less “AI pure-play” on the surface but have more diversified monetization paths. That argues for trading the theme through relative-value expressions rather than outright beta.

Key Decisions for Investors

• Long ANET into earnings; use the event as a 1-3 week catalyst trade. Risk/reward favors upside if management confirms AI order visibility and raises forward guidance; size against a stop on any signal that 800G/1.6T conversion is slower than expected.
• Long AMD vs short a weaker AI semiconductor basket over the next 3-6 months. The thesis is that CPU share gains from agentic workloads will expand faster than consensus models; risk is that guidance proves more conservative and the market keeps rewarding only the most established hyperscale beneficiaries.
• Buy PANW on pullbacks and hold 6-12 months. This is a quality-plus-compounding trade: platform consolidation and AI-driven security demand should support multiple expansion, with downside limited by sticky recurring revenue; the main risk is a short-term billings air pocket after large deployments.
• Long INTC as a medium-term turnaround/AI infrastructure trade, but express it as a smaller position or via call spreads. The path-dependent upside is meaningful if 18A and foundry momentum continue; the risk is that the market has already discounted too much good news after the sharp rerating.
• Consider a pair trade long RDDT / short a lower-quality ad-tech or social basket for 6-12 months. Reddit’s optionality lies in AI data licensing and under-monetization, while the risk is that user growth volatility or renegotiation timing pushes the rerating out; use downside-defined structures if entering before earnings.