This Boring Industrial Stock Is Helping Power AI Data Centers

Cummins raised its 2030 targets to $45 billion-$50 billion in revenue with EBITDA margin above 20%, while also lifting 2026 revenue guidance to 8%-11% growth. Power systems revenue rose 19% year over year to $2 billion in Q1, with a record 29.5% EBITDA margin, driven by surging backup power demand from data centers and AI-related infrastructure. The company is investing $450 million to expand high-horsepower engine and generator capacity by 20 gigawatts, offsetting weakness in the North American truck market.

Analysis

CMI is morphing from a cyclical truck OEM beta trade into a capacity-constrained infrastructure compounder. The key second-order effect is that AI data-center power demand is not just additive to revenue; it also re-rates the earnings quality of the whole mix because high-margin power generation and distribution now dilute the lower-margin engine cycle. That supports a higher floor on multiples, but it also means the market will begin valuing backlog conversion, lead times, and capex execution more like a quasi-electrical infrastructure name than a traditional industrial.

The main beneficiary set extends beyond CMI. Generator component suppliers, switchgear, cooling, and electrical distribution vendors should see follow-on demand as hyperscalers and colocation operators accelerate power redundancy builds; the constraint is increasingly around installed power rather than rack demand. A hidden loser is anyone exposed to delayed truck capex cycles without a similar offset: if freight weakens again, the earnings asymmetry at CMI now matters less than before, while peers still tied to on-highway demand may see multiple compression.

The risk is that the market is extrapolating peak margins too far into 2026-2028. Data-center power demand is strong, but the install cycle is lumpy, and incremental capacity additions can invite competition or pricing pressure once supply catches up; that is the key medium-term reversal mechanism. Over the next 3-6 months, the stock is more likely driven by backlog conversion and margin prints than by macro truck data, so any disappointment on capacity expansion or order timing could trigger a sharp de-rating from elevated expectations.

Contrarian view: the move is probably underappreciated on quality, but overpriced on duration. At ~22x forward earnings, the market is paying today for a multi-year AI infrastructure run-rate that may already be reflected in consensus, while the cyclical legacy business still creates headline risk. The better setup may be to own the earnings inflection through the next two quarters, but avoid paying up for the full terminal growth story unless management proves the 25-26% segment margin is sustainable beyond the current supply pinch.

Key Decisions for Investors

• Long CMI on a 3-6 month horizon into upcoming earnings, but size modestly: the setup is supported by backlog visibility and margin expansion, yet the current valuation leaves limited room for a clean beat without guidance upside.
• Buy CMI Jan-2026 $350/$400 call spreads if implied vol stays contained: captures continued AI power demand re-rating with defined downside if margin normalization arrives faster than expected.
• Pair trade: long CMI / short a truck-exposed industrial with limited data-center exposure over the next 2 quarters, expressing the view that power-generation demand offsets freight cyclicality better than the market credits.
• Watch for a pullback to post-earnings volatility: add on any selloff driven by concerns about 2026 guidance, since the medium-term catalyst remains backlog conversion rather than truck volumes.
• If valuation expands further without order-book acceleration, trim 20-30% of the position and rotate into names with more direct hyperscale capex torque.