Why this $23-billion money manager is buying Caterpillar and selling Boston Scientific

AGF’s Mike Archibald remains bullish on U.S. equities, citing a strong first-quarter earnings season in which more than 80% of S&P 500 companies beat revenue and EPS expectations. He says the best opportunities are in AI-linked industrial and technology names, highlighting Caterpillar bought at an average US$661, GE Vernova at US$885, and Corning at US$95, while he exited Boston Scientific as health care growth slowed from about 22% annual EPS growth to roughly 12%.

Analysis

The market is increasingly mispricing the second-order beneficiaries of the AI capex cycle. The obvious software/semis winners are already crowded, but the cleaner earnings revisions now sit in the infrastructure layer: power generation, grid gear, thermal management, fiber, and heavy equipment. That matters because these businesses monetize a longer-duration buildout with less customer concentration risk than hyperscalers, so estimates can keep moving up even if AI monetization slows.

Caterpillar and GE Vernova are not just “AI proxies”; they are leverage to a structural bottleneck in power delivery. The key edge is that data-center demand is creating a multi-year queue for turbines, backup power, and grid upgrades, which supports pricing power before volume fully ramps. That makes these names less cyclical than their historical multiples imply, but also more vulnerable to execution slippage because backlog conversion and free-cash-flow inflection are what justify the rerating.

Corning is the most underappreciated of the three buys because its upside depends on a bandwidth migration, not a headline AI spend number. If the AI buildout shifts from training to inference at scale, fiber intensity per compute node should rise faster than consensus expects, and copper substitution becomes a multi-year secular tailwind. The risk is that procurement cycles remain lumpy and customers push out orders if cloud capex moderates for even one quarter.

The health care exit is more telling than the buys: it signals a preference for accelerating growth over quality-at-a-reasonable-price. That’s a late-cycle style tilt if earnings breadth narrows, but it can still work for months as long as the market rewards upward estimate revisions. The main contrarian risk is that these infrastructure names become consensus trades; if rates re-accelerate or AI capex is questioned, they can de-rate quickly despite solid fundamentals.

Key Decisions for Investors

• Long CAT vs short BSX for 3-6 months: express a growth-scarcity trade where industrial earnings revisions outpace slower medical device compounding; target 15-20% relative outperformance if AI/power capex stays intact.
• Initiate a basket long GEV/GLW on 1-3 month pullbacks; use 10-15% trailing stops because the thesis is backlog and order visibility, not near-term multiple expansion. Best risk/reward if entry follows any post-earnings digestion.
• Add AMAT on weakness as a second-order beneficiary of the same capex cycle; pair against a broader healthcare ETF short to isolate AI-infra earnings momentum from defensive rotation risk.
• Consider call spreads in GEV and CAT for 6-12 months rather than outright equity: the upside is driven by estimate raises, while the main risk is a brief de-rating if rates move higher before cash flow inflects.
• Avoid chasing GOOGL/AMZN here; the market already owns the AI demand story. Prefer the suppliers of energy, transmission, and connectivity, where consensus revisions still have room to run over the next two reporting cycles.