Power Equipment Makers

The question

Are US-listed manufacturers of turbines, generators, fuel cells, transmission hardware, and precision flow-control equipment priced for a normal industrial capex cycle, or for the structurally longer build-out implied by data-center load growth, reshoring, and grid reconductoring commitments already in utility IRP filings?

Base rates

The reference class is industrial equipment cohorts during prior multi-year grid or infrastructure build cycles: the 1998-2002 merchant-gas turbine wave, the 2005-2008 transmission capex cycle, and the 2010-2014 shale-driven flow-control cycle. In each episode, the equipment cohort compounded revenue 12-20% annually through the mid-cycle phase and delivered 3-year forward returns in the 70th-80th percentile of industrial subsectors. The derating phase on the back end was severe: bookings peaked roughly 18 months before earnings, and the stocks peaked roughly 6 months before bookings.

The imputed probability embedded in current forward multiples for this cohort is that 2028 equipment revenue is within 10-15% of 2026 levels -- consistent with a short, sharp cycle. That is a coherent prior, but it assigns a low weight to the possibility that grid and data-center orders are funded on 8-to-12-year contract tails rather than annual capex budgets. The base rate for a capex cycle driven by regulated utility commitments and multi-year hyperscaler PPAs running longer than the sell-side model is higher than the base rate for a cycle driven by discretionary corporate capex.

Why the consensus view is wrong (or incomplete)

The sell-side is modeling power equipment as a cyclical book whose peak bookings imply peak earnings within eighteen months. That framework conflates two different buyer types. The first is the regulated utility, whose orders are backed by rate-case filings that already assume a ten-year reconductoring and transmission build. The second is the hyperscaler and its PPA counterparty, whose orders are backed by inference-compute capex that is, on the current disclosures, supply-constrained.

Neither of those buyers is a classical cyclical. Both are running into the same physical constraint: a North American electrical-equipment supply chain that has not had meaningful capacity added in twenty years. The constraint shows up as lead times of 50-100 weeks on large power transformers and 2-3 years on gas turbines. The margin profile of the cohort in that environment is structurally higher than the back-book average because the incremental unit is price-sensitive to the buyer and capacity-constrained to the seller. The consensus view is not wrong about the cycle existing; it is wrong about when it rolls.

Position construction

Three core positions anchor the book. GE at 20% owns the gas-turbine franchise and the aftermarket services stream that damps the derating risk. AME at 20% owns a diversified precision-instrumentation platform with 25%+ operating margins and a proven multi-cycle reinvestment record. BE at 15% is the highest-variance core: a solid-oxide fuel-cell manufacturer selling into behind-the-meter data-center power, sized to reflect genuine execution risk but weighted materially because the payoff distribution is wider than the rest of the book.

ITT (~7.8%), CR (~6.1%), and GTLS (~5.2%) are the flow-control and cryogenic specialists; their role is exposure to industrial-gas, LNG, and process-industry capex without the utility rate-case dependency. WTS (~5.3%) and GNRC (~6.4%) capture distributed water-heating and backup-power demand at the premise level. VMI (~4.6%) is the transmission-pole and irrigation franchise; MIR (~3.1%) is nuclear instrumentation and the cleanest listed proxy for the small-modular-reactor option. FLS (~3.8%) and NPO (~2.8%) round out the flow-control and sealing book at weights that reflect thinner moats.

The weighting deliberately concentrates in the top three names because the 1Y-vs-SPY hurdle already did the work of filtering the disappointments out of the cohort.

Asymmetric payoff

If utility-filed transmission capex rolls through 2028 at the pace of current IRP commitments and hyperscaler load growth holds, the weighted book compounds roughly 18-26% per year with modest multiple expansion. If capex flat-lines in 2027 and multiples compress one-and-a-half turns, the book returns roughly -12 to -20%. If grid-queue interconnection delays force utilities to accelerate behind-the-meter and on-site generation, the right tail for BE, GE, and MIR carries the book to 35-50% annually.

At a 55% weight on the base case, 25% on the bear, and 20% on the bull, expected value is roughly +14 to +20% annualized against an SPY base rate near +8%. The asymmetry is that the bear case is bounded by installed-base services revenue and the bull case is bounded only by equipment capacity.

Three things that would change our mind

1. Utility IRP filings begin materially deferring transmission spend to the back half of the decade, indicating that regulators are capping capex to protect ratepayer bills.
2. Gas-turbine lead times compress below 18 months, signaling that OEM capacity has caught demand and the pricing power in the cohort is fading.
3. Hyperscaler behind-the-meter generation orders roll off without replacement, indicating that grid interconnect has normalized and the behind-the-meter premium has collapsed.

What we are explicitly NOT betting on

We are not buying renewable-developer equities, residential-solar installers, or EV charging infrastructure. Those businesses depend on policy continuity and consumer adoption curves that sit outside this thesis. We are also not buying the utilities themselves; the regulated utility captures the volume but not the scarcity rent. The scarcity rent sits with the equipment maker whose order book is already full. That is the narrower and higher-conviction claim the portfolio is sized for.