Caterpillar (CAT) dropped its third-quarter numbers at 5:30 a.m. CDT this morning, and the market’s first read is positive. The question isn’t whether CAT beat a consensus that sat around $7.04 to $7.12 per share on roughly $20.6 billion to $20.8 billion in revenue. Caterpillar has outpaced the Zacks Consensus Estimate in each of the trailing four quarters, with an average surprise of about 18%. Another beat was, by that standard, the base case. What the market needed was evidence that the margin held without the one-time help.
Why the Margin Test Is Different This Quarter
The second quarter was remarkable on the surface: operating profit margin reached 20.9% and adjusted operating profit margin hit 21.9%, compared with 17.3% and 17.6% a year earlier. But $392 million of expected IEEPA tariff recoveries helped. Q3 carried no such cushion. Management had pointed to roughly $600 million of tariff costs landing in the third quarter, making a clean 19% adjusted operating margin the real hurdle.
That number matters because it separates the underlying business from the noise. Caterpillar’s own framing has been that tariff costs can push results toward the bottom end of its target margin range, even if underlying operations are improving. A clean 19%-plus today confirms that framing. Anything softer reopens the debate.
Power and Energy Is the Core Thesis
The reason CAT belongs on any serious industrial shortlist right now isn’t construction equipment. It’s the Power & Energy segment, and specifically what’s happening inside it. Power Generation sales increased in large reciprocating engines and turbines, primarily in data center applications. Retail Power & Energy sales to users rose 33% in Q2 2026, with Power Generation up 72%. That pace of growth, sustained into Q3, is the data-center infrastructure trade in physical form.
Caterpillar’s Power & Energy Group President Jason Kaiser told Bank of America investors that a gas generator running continuously generates 40 times more lifetime services revenue than a diesel standby unit. That single figure reframes the data center story from a hardware order cycle into a decades-long recurring revenue engine. Rivals GEV, ETN, and VRT compete for pieces of this spending. Caterpillar’s advantage is that it already has the engines in the ground and the backlog to prove the demand is real.
The $72 Billion Backlog and What It Has to Prove
At the end of Q2, Caterpillar’s firm backlog reached approximately $72.1 billion, about $9.4 billion higher than the first quarter. The largest increase came in Power & Energy. Of the total, approximately $29.2 billion was not expected to be filled within twelve months. The remaining 59% is scheduled to ship within the year, meaning the revenue visibility is genuine, not theoretical.
The critical variable today is whether the backlog grew further. Caterpillar’s stock tends to move more on the backlog and management’s outlook than on the quarter’s revenue. A bigger number on Thursday would mean orders are still coming in faster than Caterpillar can ship them. That is the signal institutional investors are watching.
What Could Go Wrong
Full-year tariff impact is now estimated at around $2.2 billion in 2026. That headwind does not disappear because Power & Energy is strong. Tariff costs and higher spending on capacity, AI, and autonomy could keep profitability near the lower end of targets if pricing or volumes soften. Valuation is a second concern: CAT’s trailing P/E has been around 31x recently, materially higher than what investors got used to earlier in the cycle. And there is a behavioral pattern worth noting: beats have not consistently lifted the stock. Across five recent beats, the average one-week move was negative 1.3%.
The Bottom Line
Caterpillar is the most direct industrial way to own the data-center power build-out, and this morning’s report is the first clean test of whether those economics hold under real tariff pressure rather than one-time relief. Caterpillar has said it is restarting production of a 10-megawatt medium-speed gas reciprocating engine platform and plans to restore about 1.5 gigawatts of capacity, with first shipments targeted for Q4 2026 followed by an 18-month ramp. If the margin line is holding near 19% on its own and the backlog expanded, the recent pullback in the stock was a reset, not a signal. That verdict is now in the numbers.
