Big Manufacturers Find New Demand in Equipping AI Data Centers
Companies such as Caterpillar and Cummins are pivoting to feed a booming market for once-prosaic power equipment
Caterpillar (CAT)
- The Move: Investing $725 million to expand an Indiana generator plant, converting a Kansas facility for turbine engines, and reviving discontinued 10-megawatt generators.
- The Payoff: Generators are now Caterpillar’s #1 profit driver, eclipsing its core construction business. Q2 revenue hit a record $20.5 billion, pushing stock up ~50% this year with data center order backlogs stretching to 2030.
Cummins (CMI)
- The Move: Dropping $450 million to boost generator output (following a $200 million push last year). Launching massive 4-megawatt, natural-gas-powered generators by 2028 designed to serve as primary grid power, not just emergency backup.
- The Payoff: Projects data-center sales will jump 80% to $9 billion by 2030.
Ford Motor (F)
- The Move: Committing $2 billion via its new Ford Energy unit to redirect excess EV batteries into large-scale energy storage for data centers.
- The Strategy: Offloads unsold battery inventory caused by the EV sales slump without abandoning battery manufacturing assets.
Eaton (ETN)
- The Move: Spent $13 billion on acquisitions since 2025 to scale up production of circuit breakers, transformers, and switchgear.
- The Payoff: Data center operations surged to 21% of total sales (up from 14% in 2023), driving Q2 sales up 21% overall.
The Overheating Risk: Boom or Bust?
While AI-driven demand lifted overall U.S. factory production to its highest level since 2022, heavy economic headwinds remain:
- Tariffs & Materials: A 50% tariff on imported metals has pushed domestic steel and aluminum prices to world-record highs. While Caterpillar passes these costs onto hyper-eager data center developers (generating ~$600M in pricing-driven profit), other sectors are buckling.
- Contraction Signals: Higher oil prices and interest rates forced 20% of the manufacturing GDP into contraction last month, hit hardest in chemical manufacturing.
- The Dot-Com Ghost: Experts warn of a "boom-and-bust cycle." If AI algorithms become more efficient or less power-hungry, developers could end up with billions in unused power capacity—much like the excess fiber-optic networks laid in the late 1990s.
For now, suppliers are hedging their bets: scaling capacity cautiously in small increments while riding the most lucrative power grab in modern industrial history.