Caterpillar Falls 4% While Deere Edges Higher: Is the Data Center Power Trade Unwinding?

Caterpillar fell 4% to $783.15 on Monday and is down 8% over the past month, underperforming the 2% decline in industrials as investors reassessed AI data-center power demand. The selling followed calls from Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman to slow frontier-model capability advances, prompting an unwind in AI-infrastructure-linked valuations. Caterpillar’s fundamentals remain intact for now: Q2 2026 power-generation sales rose 72%, backlog was $72 billion, and management said it is taking orders into 2029-30; Deere rose 1% and Paccar fell just 0.6%, consistent with Caterpillar-specific AI exposure driving the move.

Analysis

CAT’s relative weakness is more important than the absolute move: the market is beginning to separate its data-center power multiple from its cyclical machinery earnings base. Because high-margin power systems are likely carrying a disproportionate share of incremental earnings expectations, even unchanged aggregate backlog can mask a negative mix revision if hyperscaler projects are deferred, resized, or shift toward grid-connected capacity. The next 1-3 months hinge on whether peers across the power chain—Cummins (CMI), Generac (GNRC), Vertiv (VRT), Eaton (ETN), and Caterpillar—confirm delivery schedules and pricing rather than merely cite demand interest.

DE’s resilience reinforces a potentially useful rotation: agricultural-equipment earnings are driven by farm income, replacement cycles, and precision-ag adoption rather than AI capex duration. That said, DE is not a clean defensive substitute at its elevated year-to-date performance; weaker commodity prices or a dealer-inventory rebuild reversal would quickly pressure its own earnings multiple. PCAR is a weaker short hedge against CAT because commercial-truck demand has independent freight-cycle and North American Class 8 order risks.

The contrarian read is that a policy-oriented AI safety debate is unlikely to alter near-term electrical-load commitments already embedded in utility interconnection queues and construction contracts. But CAT’s valuation may not require cancellations to derate: a shift from orders extending several years out to a more normal lead-time profile would lower the scarcity premium and expose its conventional construction/resource-cycle sensitivity. Falsification for the bearish relative thesis is a next-quarter acceleration in power revenue, stable-to-higher power margins, and explicit confirmation that key customer delivery dates remain intact; conversely, any backlog conversion slippage or commentary on financing/grid delays would validate further downside.

Key Decisions for Investors

• Initiate a 1-3 month relative-value position: short CAT / long DE in equal dollar amounts after CAT’s initial gap-down stabilizes. Target 8-12% relative outperformance from multiple compression in CAT versus resilient farm-equipment estimates; stop out if CAT recovers above its pre-selloff level on confirmed power-order acceleration.
• Use CAT’s next earnings update as the decisive catalyst. Maintain only a small tactical short ahead of it; add if management identifies delivery deferrals, weaker power-system pricing, or backlog-conversion delays. Do not add solely on AI-sentiment headlines without corroborating order data.
• Monitor CMI, VRT, ETN, and GNRC for a broader power-infrastructure read-through. If two or more report unchanged backlog but lower shipment timing or margin commentary, expand the CAT short and consider a basket short versus XLI; if they reaffirm shipment cadence, cover the CAT short.
• Avoid using PCAR as the primary long leg. Its freight-cycle exposure can create idiosyncratic downside independent of data-center capex; DE offers cleaner diversification, though position sizing should account for agricultural commodity and dealer-inventory risk.