- Caterpillar reported second-quarter earnings per share of $8.17, dramatically beating the $6.17 analyst consensus — a 73% year-over-year increase from $4.72 in the prior year period — with revenue of $20.5 billion against $19 billion expected, as the company’s power and energy segment (which manufactures generators, engines, and gas turbines for industrial facilities and large-scale computing centers) has grown to become its largest business by sales, overtaking the iconic yellow construction machinery division that most people associate with the brand; the company posted a record backlog of $72 billion, up 92% from a year ago — a figure that represents genuine contracted future demand rather than speculative pipeline and that directly validates the thesis that AI data center build-out is a multi-year capital expenditure cycle with durable demand rather than a one-quarter phenomenon; CAT shares surged more than 11% in pre-market trading on the results.
- The composition of Caterpillar’s backlog tells the story of the AI infrastructure investment cycle in physical terms: data centers require not just chips and servers but an entire physical infrastructure stack — power generation (generators, gas turbines), cooling systems, uninterruptible power supplies, and backup capacity — and Caterpillar’s power unit is the dominant supplier of the generation layer of that stack; the 92% year-over-year backlog growth means that customers are committing to Caterpillar equipment purchases 12-24 months in advance, reflecting both the lead time required for large generators and turbines and the certainty with which hyperscalers and cloud providers are committing to their data center construction timelines; Siebert Financial’s CIO Mark Malek framed it clearly: “Those backlogs are pointing to the same drivers — data center build-outs. Chips, servers, now industrial equipment” — the AI trade has moved up the industrial supply chain to encompass the physical power infrastructure that makes data centers possible.
- The construction industries segment also delivered strong growth — 35% revenue increase as dealers restocked equipment — providing additional evidence that the broader industrial cycle remains robust beyond AI specifically; this diversification is important for the bull case on Caterpillar because it means the company is not purely a leveraged AI play but a broad industrial machinery company with an AI-driven tailwind in its largest and fastest-growing segment; however, the bear case — articulated by Baird (which downgraded CAT to hold last week citing growing state and local restrictions on data center construction) and by short-seller Michael Burry (who has publicly disclosed a short position) — centers on the risk that regulatory constraints on data center siting and permitting could slow the build-out timeline faster than the backlog suggests.
- The Caterpillar results have read-through implications for the broader AI infrastructure investment thesis: they confirm that the physical power and cooling infrastructure layer of the AI build-out is real, contracted, and growing at extraordinary rates — not speculative; Vertiv Holdings (data center cooling and power management) rose 8.89% and GE Vernova (power generation and grid equipment) rose 1.66% on the Caterpillar results, as investors extrapolated the validation of AI power demand to comparable companies; the results also contextualize the AI economy transformation thesis — when Caterpillar’s largest business segment is now supplying power equipment for data centers rather than construction sites, the AI investment cycle has achieved a scale that is reorganizing the physical capital stock of the U.S. economy, not just its software layer.
What Happened?
Caterpillar reported Q2 EPS of $8.17 vs. $6.17 expected and revenue of $20.5 billion vs. $19 billion expected, with a record $72 billion backlog — up 92% year-over-year — driven primarily by its power and energy unit, which now manufactures more revenue for the company than its iconic construction machinery segment. The power unit makes generators, engines, and turbines for data centers. Construction equipment also grew 35% as dealers restocked. CAT surged 11%+ premarket; Vertiv rose 8.89% and GE Vernova 1.66% on the read-through.
Why It Matters?
A $72 billion contracted backlog up 92% is not a demand signal — it’s contracted future revenue that validates the AI data center build-out as a multi-year, fully committed capital expenditure cycle. The fact that CAT’s power unit has overtaken construction as its largest segment means AI infrastructure is now physically reorganizing one of America’s most iconic industrial companies. For investors, this is the clearest real-economy confirmation that the AI capex cycle is translating from announcements into contracted orders for physical equipment.
What’s Next?
Watch whether Baird’s concern about data center regulatory restrictions materializes — state and local permitting constraints on new data center construction are the primary risk to the backlog conversion timeline; watch Vertiv and GE Vernova’s own earnings for confirmation of the power infrastructure demand signal across the ecosystem; watch Michael Burry’s short thesis for any public updates, since a contrarian call this specific deserves tracking; and watch the backlog’s conversion pace — a record backlog is only valuable if supply chain and manufacturing capacity can fulfill it, and any bottleneck in steel, copper, or specialized components could delay revenue recognition.
Source: Bloomberg
