This group makes the heavy electrical "plumbing" that connects power to a building and routes it safely inside. Two products dominate: power transformers (steel-and-copper boxes that step voltage up for long-distance transmission and back down for use) and switchgear (the breakers, panels and busways that protect circuits and split power from the utility feed down to individual machines). Every gigawatt-scale data center, every new power plant, and every grid upgrade needs both. The names here — Eaton (ETN), Hubbell (HUBB), GE Vernova (GEV), ABB, plus the contractor that installs much of it, Quanta Services (PWR) — sell this gear, or build it into projects, and earn cash as orders convert to shipments.
On the evidence below, demand is running ahead of supply today: lead times have stretched to multiple years and near-term capacity is described as effectively sold out. What limits supply is physical — only a few factories, a scarce specialty steel, and new plants that take 3+ years to build — not money or willingness to order. The market currently assigns more dollars of market value per dollar of revenue for several of these names than it did a few years ago (see "The price of exposure" for the arithmetic). This sheet lays out the demand evidence, the supply evidence, and what you pay today, and leaves every judgment to you.
Transformers change electricity from one voltage to another. Large power transformers (LPTs) are custom-engineered, weigh hundreds of tons, and are built to order over roughly 12-24 months of factory time. Switchgear is the assembly of circuit breakers, disconnects, and metering that controls, protects, and isolates power as it flows from the grid into a facility and down to each load. Both are sold as units — priced per transformer (often on the order of $1-10M+ each for large units est.) or per switchgear lineup/section — though contractors also sell them bundled inside a project price.
The money mechanics are straightforward. A customer places an order; the maker books it into backlog (orders received but not yet shipped — future revenue that is contracted). The maker spends on materials and labor (its COGS, cost of goods sold — the direct cost of building the unit), ships the unit, and recognizes revenue. The cash an owner ultimately keeps is roughly: revenue, minus what it costs to build, minus overhead, minus the capex (capital expenditure — spending on new factories and machines) needed to expand. When demand outruns supply, makers can sometimes raise prices faster than costs rise, which lifts margin (the slice of each sales dollar left as profit). Contractors like Quanta earn instead on labor and project management — they install and connect the gear rather than fabricate the steel.
Source: 500-stocks Industrial & Construction scan, sections 01 (Power Transformers) and 02 (Switchgear & Power Distribution).
Today's demand comes from three overlapping waves: (1) data-center build-out, (2) grid replacement and expansion, and (3) general electrification (EVs, factory reshoring, renewables). All three need the same transformers and switchgear, and they are arriving at once.
The data-center wave is the new, fast-moving driver. Per the scan, every gigawatt-scale campus needs multiple large transformers, and a data center is described as "the most switchgear-dense building on the planet" — AI clusters at 60-100+ kW per rack (kW per rack = the electrical load packed into one server cabinet) require more distribution gear per square foot than older facilities. The scan states that hyperscaler (the largest cloud operators, e.g. the companies running planet-scale data centers) build-out plans imply 2-3x the historical annual transformer demand through 2029, and that grid interconnection queues (the waiting lists to connect new loads to the grid) are now dominated by AI-related loads. forecast
Reasoning from AGI arriving: as compute demand keeps compounding, the binding real-world constraint shifts from chips toward power delivery. More compute means more megawatts, and every megawatt drags transformers and switchgear behind it in a near-fixed ratio — you cannot energize a GPU without stepping the voltage down and routing it through breakers. Physical AI (robotics, automated factories) layers additional electrified load on top of the same grid. So the demand curve here is not a single product cycle; it tracks the entire electricity footprint of the build-out. These are forecasts, not contracts: actual orders depend on how fast hyperscalers and utilities break ground.
The buyers are: utilities (replacing an aging grid and adding interconnections), hyperscalers and colocation operators (companies that rent out data-center space), power developers (gas, nuclear, renewables), and industrial customers. Utility and data-center demand is the durable core; both are reported to be spending at record levels.
✓ VERIFIED — the following figures were confirmed from primary sources after initial publication:
Supply is the central constraint in this group, and the scan describes it as "severely constrained" for transformers. The limits are physical, not financial:
Expansion is underway but lands later: the scan says Eaton, Schneider, and ABB have announced major switchgear capacity expansions that "won't fully come online until 2027-2028." forecast So the supply curve is rising but with a multi-year delay.
Market-share structure. Supply is concentrated. A small set of global incumbents — ABB, Eaton, Schneider Electric (not US-listed), Hitachi Energy, GE Vernova, Siemens Energy (not US-listed), plus Hubbell, Powell (POWL) and others in specific niches — control most capacity. est. That concentration is one reason the shortage persists: there is no large pool of marginal suppliers able to flood the market quickly.
Source: 500-stocks scan sections 01, 02, and 05 (supply constraints, lead times, capacity-timing, labor).
Putting the two sides together: on the available evidence the product is short — demand exceeds what can be made — and has been for several years. The evidence shows up as physical queues and prices rather than opinion:
| Signal | Reading | What it means |
|---|---|---|
| Transformer lead time | ~12 mo → 3-4 yrs | Build rate far below order rate |
| Switchgear lead time | 40-70 weeks | Order book runs ~1+ year ahead |
| Capacity status | effectively sold out (near-term) | Near-term output is spoken for |
| Pricing trend est. | reported rising | Sellers have pricing power |
| New supply online | 2027-2028 forecast | Relief is delayed, not immediate |
When could it flip to oversupply? Three things would have to line up: the 2027-2028 capacity expansions arrive on schedule, the GOES steel bottleneck eases, and demand growth slows (a pause in data-center or grid spending). None of these is certain, and they cut in opposite directions — capacity is rising, but if AGI-driven compute demand keeps compounding, demand may rise just as fast. The factual statement is that the shortage is tight on the evidence through at least the mid-2020s, and the timing of any flip to balance or surplus is a forecast dependent on factory ramps and the AI capex cycle, not a settled fact.
These names range from near-pure electrical-equipment makers to broad conglomerates and a contractor. The "Exposure" column describes how concentrated each company's revenue is on grid equipment / electrical infrastructure, to show how direct the link to this product is. Revenue figures are rounded, from general knowledge of recent filings, and not live-verified.
| Company | What it makes | Exposure to this product | Rough size est. | Position |
|---|---|---|---|---|
| Eaton (ETN) | Switchgear, breakers, distribution, electrical systems | High — electrical is the majority of revenue; large data-center order book | ~$24-26B rev est. | Large switchgear supplier; broad electrical franchise |
| Hubbell (HUBB) | Utility & electrical distribution gear, grid components | High — utility/electrical focused; closest to a grid pure-play in this list | ~$5-6B rev est. | Utility-grid niche; smaller, focused |
| GE Vernova (GEV) | Grid transformers/switchgear, plus gas & wind power equipment | Medium — "Electrification/Grid" is one of three segments | ~$34-36B rev est. | Transformer/grid maker; also a power-generation business |
| ABB | Switchgear, transformers, electrification + robotics/automation | Medium — Electrification is a large segment, not the whole | ~$31-33B rev est. | Global incumbent across transformers & switchgear |
| Quanta (PWR) | Builds/installs grid & power infrastructure (contractor, not maker) | Indirect — installs the gear; revenue scales with build spend | ~$23-25B rev est. | Large electrical-infrastructure contractor; record backlog reported |
Source: 500-stocks scan company lists (sections 01, 02, 05); revenue scale and segment mix from general knowledge of recent filings, rounded and not live-verified.
In plain money terms, when you buy a share you are buying a claim on future owner cash. A simple gauge is market value per dollar of this year's revenue (price-to-sales): how many dollars of company value the market assigns for each $1 the company sells in a year. Another is price-to-earnings (P/E) — dollars of value per $1 of annual profit. The rough, not-live-verified picture for this group, all order-of-magnitude only: est.
Money-in / money-out shape. The equipment makers (ETN, HUBB, GEV, ABB) are moderately capital-intensive: they must spend capex to expand factories, but once a plant runs, each extra unit sold during a shortage throws off strong cash. They are the ones positioned to capture a scarcity-driven price directly. The contractor (PWR) is more capital-light (it ties up little money in plant and equipment) — it does not own scarce factories, so it cannot price like a sole supplier, but it also does not carry the heavy plant cost; it converts labor and backlog into cash with less invested capital. So the group splits into "owns the scarce capacity" (the makers) versus "rides the volume" (the contractor). This is neutral arithmetic — whether any of these prices is worth paying is the reader's call.
Source: valuation multiples and margin characterizations from general knowledge (cutoff ~early 2026), approximate and not live-verified; revenue scale from scan-listed companies' recent filings.
Where a company-level deep-dive would be most informative, grouped by type of exposure:
This is a factual map of where to look, not a recommendation.
What was used:
Hard vs approximate: