NextFin News - Caterpillar Inc. shares are on track for a more than 30% drawdown from their record high, a reversal that lays bare the uncomfortable truth investors are only now pricing: the industrial giant was never a pure artificial-intelligence growth story, and the market paid for it as if it were. The stock slid as much as 4% to a session low of $742.01 on Wednesday and closed at $746.32, down 3.4% on the day, leaving it roughly 30% below the late-June peak of $1,064.90. The selloff is not a single-stock story. It is the point where three separate repricings collide — oil prices rolling over on slashed demand forecasts, Treasury yields pinned near their highest levels since 2002, and the AI-infrastructure trade losing its momentum — and Caterpillar, more than almost any other industrial name, is exposed to all three.
The central question this selloff forces is whether Caterpillar's AI-driven re-rating was a cyclical overshoot that will mean-revert, or whether the data-center power boom represents a structural floor that will catch the stock well above its old multiples. The evidence points to a hybrid answer: the multiple expansion was cyclical and is unwinding fast, but the underlying power-generation demand is structural enough to keep a full collapse off the table. That distinction matters because it tells you where the 30% drawdown is likely to stop — and why the consensus price target of $974.50, implying 30.6% upside from here, may be betting on a recovery the bond market is not willing to underwrite.
The Setup: A Stock That Became an AI Proxy
To understand the magnitude of the reversal, start with how far Caterpillar traveled. Through the first half of 2026 the stock gained 86%, making it one of the best performers in the S&P 500, and it pushed past $1,060 in late June on the back of a single, powerful narrative: AI data centers need enormous amounts of electricity, the grid cannot deliver it fast enough, and Caterpillar sells the diesel and gas-fired generators that fill the gap. Investors stopped treating the company as a cyclical builder of mining shovels and backhoes and started valuing it as infrastructure exposure to the AI capital-expenditure supercycle.
The fundamentals initially justified the enthusiasm. In the second quarter of fiscal 2026, Caterpillar reported record sales and revenues of $20.54 billion, up 24% from a year earlier, with adjusted earnings per share of $8.17 beating the $6.20 consensus estimate by nearly 32%. Operating profit rose 50% to $4.29 billion, lifting operating margin to 20.9% from 17.3%, and the company generated $4.40 billion in enterprise operating cash flow. Construction Industries led with $8.35 billion in sales, up 35%, while North America alone jumped 50% to $5.07 billion. Power & Energy contributed $8.24 billion, buoyed by data-center demand for large reciprocating engines and turbines. Management's outlook was equally confident: CEO Joe Creed pointed to strong order rates and a growing backlog, and the company has said it expects its power-generation business to triple by 2030.
But the stock price outran even those results. At $1,064.90, Caterpillar traded at a price-to-sales ratio that a chart shared by investor Michael Burry showed was the highest in at least three decades, and the shares carried a trailing price-to-earnings multiple above 33 times. That is a valuation that assumes not just a good year, but a permanent reclassification of the business. When a cyclical industrial trades at a technology multiple, it does not need bad news to fall — it only needs the narrative to stop working.
The first crack appeared on June 30, when Burry, the Scion Asset Management founder famous for his subprime-mortgage bet, disclosed a short position in Caterpillar at $1,060.98, alongside bearish wagers on Nvidia, Applied Materials, Tesla and the iShares Semiconductor ETF.
"Caterpillar jumped out at me. I have never shorted Caterpillar. It has always done great for me on the long side in the past."
In August he added to the position around $844, writing that "data center plans are already aging." Burry's thesis was not that AI demand would disappear; it was that the market had capitalized decades of data-center buildout into a single year's earnings, and that the heavy equipment would finish getting paid for long before the data centers finished paying for themselves.
Why the Selloff Accelerated: The Bond Market Does the Math
The narrative broke in early October, and the transmission channel was the Treasury market. The yield on the 10-year note climbed to 5.34%, its highest level since 2002, before settling at 5.24% on October 9. That matters for Caterpillar in two distinct ways, and understanding both is the key to reading the selloff.
First, the discount-rate channel. Every equity's value is the present value of its future cash flows, and the longer and further out those cash flows sit, the harder a higher discount rate hits them. A stock priced on AI infrastructure earnings expected five and ten years from now — the power-generation tripling by 2030, the multiyear backlog conversion — is effectively a long-duration asset. When the 10-year yield moves from the low-4% range to above 5.3%, the present value of those distant cash flows compresses mechanically. This is why Caterpillar fell harder than the broader market even on days when nothing changed in its own operations.
Second, the demand channel. Heavy equipment is almost never bought for cash. Bulldozers, mining trucks, locomotives and generators are financed, often through the manufacturers' own captive finance arms. Caterpillar runs one of the largest such financing operations in the industry, which means rising rates hit it twice: customers defer or downsize purchases because borrowing costs have jumped, and the company's own lending book earns a wider spread on riskier credit. On October 7, when long yields touched their two-decade high, Caterpillar dropped 6% to $811.70 while Deere fell 4% to $657.95 and PACCAR eased 2% to $106.66. The Industrial Select Sector SPDR ETF fell 2% against a 0.6% decline in the broad market — a clear signal that this was rate-driven, not company-specific.
The rate move also exposed a second-order problem that most investors missed on the way up: the same forces pushing yields higher are the ones that eventually slow the economy Caterpillar sells into. Yields rose on a combination of persistent inflation, heavy U.S. deficit issuance and the funding demands of the AI buildout itself. If the bond market is telling you that capital is becoming scarce and expensive, it is also telling you that the construction, mining and trucking customers who buy Caterpillar's machines will face tighter hurdles before committing to billion-dollar fleets. The AI trade and the rates trade are not separate stories. The AI capital boom is part of what is keeping yields elevated, and elevated yields are what will throttle the next leg of capital spending.
Oil's Signal: The Growth Slowdown Is Already Here
While rates attacked Caterpillar's valuation, oil sent a message about its end markets. Brent crude fell below $89 a barrel and WTI traded near $82.77 after both OPEC and the International Energy Agency cut their 2026 demand forecasts. OPEC reduced its expected 2026 demand growth to 580,000 barrels per day, down from the 780,000 barrels per day it projected in July. Oil is the most liquid macro indicator in the world, and when two institutions with opposing mandates — a producer cartel and a consumer watchdog — agree that demand is weakening, the market listens.
For Caterpillar this is more than a backdrop detail. Mining equipment demand tracks commodity prices with a lag, and lower oil prices eventually translate into reduced drilling activity, fewer mining projects and softer orders from energy-exposed customers. The company's resource industries segment had been one of the weaker links even during the AI rally, and a demand downgrade of this size raises the risk that the mining cycle turns negative just as the construction cycle is normalizing. The uncomfortable combination is this: Caterpillar is being asked to carry a growth-stock multiple at the exact moment the global growth indicators are rolling over.
There is also a valuation asymmetry that the selloff has only partially corrected. Even after the 30% drawdown, Caterpillar trades at roughly 33 times trailing earnings and about 27.5 times the consensus 2026 earnings estimate of $27.16. The 56 analysts covering the stock hold an average price target of $974.50, implying 30.6% upside, with the low end of the range at $749 — essentially the current price. That distribution tells you the Street is still modeling a recovery: the median analyst believes the AI power story remains intact and that the multiple will re-expand once rates stabilize. The problem is that the recovery depends on two things happening simultaneously — earnings growing into the multiple and rates backing off — and the oil data suggests the earnings leg is the one most at risk.
The Counter-Thesis: Why This Time Could Be Different
The strongest argument against the bear case is simple and data-backed: the data-center power buildout is real, it is urgent, and it is not going away. The U.S. electrical grid was not designed for the concentrated, round-the-clock load that AI campuses require, and permitting and transmission upgrades move on a timeline measured in years, not quarters. Hyperscalers have responded by contracting directly for on-site generation, which is precisely Caterpillar's wheelhouse. The company's partnership with Vertiv to combine power-distribution systems with Caterpillar's generation technology, and a backlog that reached a record $63 billion in the first quarter, are not the footprints of a hollow narrative. They are purchase orders.
Analysts who maintain bullish ratings have argued that a structural theme is emerging in demand for on-site power systems. The argument is that Caterpillar is not merely riding a capex cycle but capturing a permanent share shift: data centers will always need backup and primary generation, and the grid's inability to keep pace is a structural constraint, not a cyclical one. Under this view, the selloff is a valuation reset within a secular uptrend, and the power-generation business tripling by 2030 is a conservative estimate rather than a stretch goal.
This counter-thesis is credible, and it is the reason a full collapse to pre-rally multiples is not the base case. But it has a specific vulnerability. The structural demand for data-center power does not automatically translate into structural earnings growth at today's price if the timing of that demand gets pushed out. Data-center developers fund their projects with debt, and debt is now priced at levels not seen in more than two decades. When financing costs rise, projects get delayed or phased, and a backlog that extends into 2027 becomes a backlog that converts to revenue more slowly than modeled. The structural story survives; the earnings-per-share trajectory for 2026 and 2027 does not necessarily survive intact.
There is also the question of what the market has already priced. The counter-thesis is now the consensus view — it is embedded in the $974.50 average target and in the 27.5 times forward multiple. A thesis that everyone holds is not a source of alpha, and it offers no margin of safety if the conversion of that backlog hits even modest delays. Burry's short, by contrast, is explicitly a bet on timing, not on the end of AI. That asymmetry — the bull needs everything to go right on schedule, the bear only needs slippage — is why the path of least resistance remains down until the conversion rate proves itself.
What to Watch: The Signals That Will Decide the Next Leg
Caterpillar's next earnings report, due later this month, will be the first real test of whether the structural demand story can outrun the cyclical headwinds. Investors should focus on three specific metrics rather than the headline number. First, the Power & Energy segment revenue growth rate: if it decelerates sharply from the second-quarter pace, the AI power narrative is losing steam. Second, the backlog conversion commentary — specifically whether the company is seeing project delays or cancellations in data-center orders. Third, pricing power in the face of the roughly $2.4 billion in tariff and raw-material cost headwinds the company flagged for 2026; margin compression there would confirm that demand is softening enough to limit pass-through.
The falsifying signal for the bear case is precise and observable: if Caterpillar reports Power & Energy revenue growth above 25% year over year for two consecutive quarters while maintaining operating margin above 20%, the structural-demand thesis wins and the 30% drawdown will look like an overshoot. Conversely, the bear case is confirmed if backlog growth stalls while the 10-year Treasury yield holds above 5.2% — that combination would mean demand is rolling over at the exact moment capital remains expensive, and the shares would have further to fall toward the low end of the analyst range near $749.
Outlook: A Cyclical Unwind Riding a Structural Wave
The cleanest way to frame the next twelve months is to separate the two forces rather than blend them. The multiple expansion was cyclical, and it is unwinding. A stock that traded above 33 times earnings on the promise of AI-infrastructure growth cannot hold that multiple when the 10-year yield sits above 5% and oil is signaling a global growth slowdown. That repricing has room to run in the near term, and the base case is for Caterpillar to trade in a range between the current level and the low-$800s while the market waits for evidence on backlog conversion.
But the underlying power-generation demand is structural, and that puts a floor under the stock that pure cyclicals do not have. The downside case — a break below $749 toward the mid-$600s — requires either a recession that crushes construction and mining demand or a decisive slowdown in data-center capex that invalidates the 2030 tripling target. The upside case — a move back toward the $900s — requires rates to retreat from current levels and Caterpillar to demonstrate that its backlog is converting to revenue and cash flow on schedule. Neither is the base case today.
Across time horizons, the picture diverges. In the short term, sentiment and positioning dominate, and with the stock down 30% from its high and the AI trade still unwinding, volatility will remain elevated. Over the medium term, the upcoming earnings report and the pace of backlog conversion will decide whether this is a buying opportunity or a value trap. Over the long term, the structural demand for data-center power is real, but the question is whether Caterpillar captures it at a valuation that rewards shareholders or whether the benefits get competed away as every industrial and power-equipment maker rushes into the same market.
The final judgment is this: Caterpillar's 30% drawdown is the market correcting a category error, not a company breakdown. The business is sound, the backlog is real, and the AI power theme is not a mirage. But a great company is not always a great stock, and paying a technology multiple for a cyclical industrial at the peak of its narrative was always going to end the same way. The bond market has done the math, oil has sent its signal, and the burden of proof has shifted to management to show that the future they sold investors is arriving on time.
Explore more exclusive insights at nextfin.ai.
