The Caterpillar Correction Shouldn't Last Much Longer

Caterpillar's AI-linked power and construction businesses are accelerating, with power & energy revenue up 17% year over year and construction revenue up 35%. Its $72 billion backlog nearly doubled from a year earlier and increased $9.4 billion, or 15%, sequentially, supporting multi-year revenue visibility. Following a share-price correction, the article argues Caterpillar's 1.4x PEG ratio is more attractive versus its historical level above 2x, as hyperscalers are projected to spend $1.3 trillion in 2027.

Analysis

CAT’s AI linkage is more valuable through power-generation equipment than through earthmoving: onsite/bridging power is a high-urgency purchase with pricing power, whereas construction equipment demand is more exposed to contractor utilization and dealer inventory cycles. The key underwriting question is therefore not aggregate backlog, but the mix, cancellation terms, and delivery schedule of power orders; a backlog concentrated in multi-year data-center projects can support earnings durability, while a dealer-replenishment component would be materially less defensible.

Competitive positioning favors CAT versus CMI in large reciprocating-generation packages, but ETN and VRT may capture a greater share of electrical distribution, cooling, and power-management spend per data-center megawatt. A second-order beneficiary is PWR, since data-center projects ultimately require grid interconnection and transmission work; conversely, constrained utility interconnects can delay construction starts and defer CAT equipment deliveries even if hyperscaler capex remains intact.

Over the next 1-3 months, the stock needs evidence that Power Generation orders and segment margins are accelerating rather than merely benefiting from a favorable comparison. Over 6-18 months, the risk is that temporary generation demand normalizes as grid capacity comes online, while a broader construction slowdown reasserts CAT’s cyclical exposure. The consensus may be over-attributing the whole company’s earnings power to AI: CAT remains sensitive to mining, energy, dealer inventories, and global construction, making a premium multiple vulnerable if non-data-center end markets soften.

Key Decisions for Investors

• Establish a modest long CAT / short DE pair over the next 1-2 months: CAT has a more credible data-center power and North American infrastructure offset, while DE remains more exposed to a slower agricultural-equipment replacement cycle. Target 10-15% relative upside; exit if CAT power-generation order growth decelerates materially for two consecutive quarters or dealer inventories rise.
• Prefer long ETN or VRT over adding outright CAT if the objective is concentrated AI-infrastructure exposure. Electrical equipment and thermal-management content scales more directly with commissioned megawatts; CAT is the lower-beta, more cyclical expression. Reassess after hyperscaler capex guidance and utility interconnection updates.
• Do not chase CAT on the article’s aggregate-backlog claim without confirming backlog composition, expected conversion timing, and cancellation provisions at the next earnings release. A miss in Power & Energy margin, weaker order book-to-bill, or reduced 2027 capex guidance from META would falsify the near-term AI re-rating thesis.
• Use any post-earnings pullback tied to weak legacy construction commentary to build CAT only if power-generation backlog remains expanding and pricing offsets input costs. Risk/reward is unfavorable if the stock retains an AI premium while segment mix fails to improve.