Dan Ives says AI buildout only in 3rd inning. Here are his top 5 plays from here

Analyst Dan Ives called the AI investment cycle early-stage and cited an estimated $4 trillion spending wave over the next few years, naming Nvidia, Microsoft, Palantir, Apple and CrowdStrike as top equity plays. His price targets imply upside of 25% for Nvidia, 16% for Microsoft, 30% for Palantir and 20% for CrowdStrike; he also expects cybersecurity's share of IT budgets to rise from roughly 5% to 10% over the next three years. Ives sees potential upside of about 58% for Boost Run and 67% for Serve Robotics.

Analysis

The investable question is shifting from whether AI infrastructure is being built to who captures durable returns after deployment. Broader customer demand would reduce reliance on a handful of hyperscalers, but does not by itself establish attractive utilization or returns on capital; sovereign and neocloud projects may be more exposed to funding, power, and delivery constraints. For NVDA, the key check is whether customer breadth translates into repeat orders and sustained pricing, not merely announced build-outs. For MSFT, expanding RPO outside frontier-model customers may improve concentration risk, but RPO is not near-term revenue: conversion pace and cloud margins matter.

Agentic software creates a two-sided cybersecurity effect. More autonomous identities and machine-to-machine access expand attack surfaces, supporting security demand; meanwhile, frontier labs entering security could bundle capabilities or shift value toward platforms that consolidate tools. That makes CRWD’s upside dependent on measurable customer adoption and retention, not a sector-budget estimate. PLTR and AAPL are named without enough company-specific evidence here to underwrite a differentiated trade. BRUN’s reported conversion of contracted demand is an early signal, not proof of sustained scale. SERV remains a longer-duration execution story where deployment economics and local operating constraints require verification.

Near term, analyst targets may reinforce crowded AI positioning, but this article supplies no valuation or independent earnings evidence. Over 1–3 months, earnings, guidance, backlog conversion, and security-sector retention are the useful tests. Over 6–18 months, power availability, custom silicon, competition, and whether AI spending produces customer ROI could determine who keeps the economics. The contrarian risk is that spending broadens while returns accrue to buyers or infrastructure providers other than today’s favored names.

Key Decisions for Investors

• Do not trade the published price targets alone. Treat them as a sentiment catalyst; verify current valuation, earnings expectations, and positioning before adding exposure.
• Watch NVDA and MSFT earnings for order conversion, customer breadth, and evidence that infrastructure commitments are monetizing. A slowdown in orders or weaker cloud/AI monetization would falsify the broadening-demand thesis; power or delivery constraints could defer rather than cancel demand.
• For cybersecurity, keep CRWD on a confirmation list rather than buying the sector-budget claim: look for durable retention, module expansion, and guidance that supports incremental demand. Reassess if frontier-lab competition coincides with slowing customer growth or weaker retention.
• Treat BRUN and SERV as higher-uncertainty watch positions, not substitutes for established AI exposure. For BRUN, verify reported demand converts into recurring scale; for SERV, verify deployments and unit economics. The article provides insufficient operating detail to size either trade.