The Broader Market
The world's largest maker of construction and mining equipment, plus power-generation systems used by data centers.
{'verdict': '3 signals sit in the elevated band: the valuation premium vs fundamentals, operating leverage and insider-selling intensity. This does not trip the convergence flag. Ranks 16th of 68 on composite fragility (F\xa055.6), below Intel and MU.', 'as_of': '2026-07-11', 'source': 'engine-restatement (T1)', 'snapshot': {'composite_f': 55.6, 'n_elevated': 3, 'convergence': 'moderate', 'rank': 16, 'elevated': ['the valuation premium vs fundamentals', 'operating leverage', 'insider-selling intensity']}}
You say CAT is trading like an AI company — isn't the datacenter revenue real?
Yes. Power Generation at $10.275B with explicit "primarily data center applications" disclosure is the most concrete AI-infrastructure revenue among Level 5 names — it is filed, hardware delivered, not deferred. The revenue is real. The question is whether 37-50x P/E is the right price for that revenue, when CAT historically trades 15-20x and when the operating margin is compressing (tariffs, manufacturing costs).
The $51.2B backlog is essentially 2 years of production — how fragile is that?
Backlogs in capital goods are typically firm purchase orders, not soft reservations. However, if AI capex pauses, customers delay or cancel orders, and backlogs shrink faster than they appear. The jump from ~$40B to $51.2B in Q4 2025 alone is suspicious in speed — it suggests a wave of orders placed ahead of expected tariff increases, which may pull forward demand from 2026-2027 into Q4 2025.
Power Generation is $10B. At 32.5% growth it doubles in ~3 years — is that priced in?
At 37x forward P/E, yes — consensus is modeling Power Generation continuing to compound at 20-30% annually. If that rate slows to 10-15% (still strong for a capital goods business), the earnings miss vs expectations triggers a P/E re-rate from ~37x to ~25x, and the stock falls ~30-40% even with solid underlying growth.
What's the real tariff exposure?
$2.6B projected tariff headwind in 2026 — CALL-STATED. If that hits margins in a year when Power Generation growth moderates, the double-compression (lower growth + lower margins) removes the "AI premium" justification simultaneously. This is the highest-probability near-term bear trigger.
Is there a bear scenario where CAT is actually a value trap?
Yes. If: (1) AI datacenter buildout peaks in 2026 after the initial wave; (2) hyperscalers shift from gas gensets to grid power as transmission capacity expands; (3) tariffs persist; (4) Construction and Resource remain weak. In that scenario, CAT earns $15-17 EPS at 15-18x multiple = $225-300/share vs $948 current. Backlog burns down over 2 years, and the stock reaches fair value when the datacenter narrative fades.