Australia's Power Boom Pays the Cranes Well. The Real Earnings Are in the Crews.

Australia's grid operator AEMO finished its 2026 planning round with the number that frames the AI trade in Australian dollars and watts: data centres, which today use about 5 terawatt-hours of electricity a year — roughly 3% of the national grid's demand — are forecast to reach about 34 TWh by 2036, close to 13% of everything the network moves. The AI boom's Australian chapter is electricity, and the obvious way to own it was found months ago.

The market's pure play is GenusPlus (ASX: GNP), the contractor that builds and maintains the transmission lines and substations plugging new load into that grid. Its shares roughly doubled over the past year. Then, on August 25, it reported record FY2026 revenue of A$1.28 billion, up 71%, with normalised EBITDA of A$100.8 million — and the stock slipped the day it announced. The tell is in the margin line: net margin compressed to 3.8% from 4.7% while revenue grew 70%. The purest listed exposure to Australia's grid build-out is scaling volume faster than it is scaling profit. Being necessary gets you orders; being scarce decides who keeps the money.

The scarce thing in this build-out is not capital, and not steel. It is the qualified people who keep critical plant alive — and the ASX small-cap that owns a broad pool of them is Tasmea (ASX: TEA), a roll-up of more than two dozen specialist trade businesses that send electricians, fitters and shutdown crews to the plants that generate power, treat water, process ore and move gas. It only listed in April 2024.

This is the corner of the Australian market where "real earnings" is genuinely scarce, which is exactly why the headline exists. By index weight, roughly 60% of the S&P/ASX Small Ordinaries Index sits in companies with negative trailing earnings (VanEck, FactSet, 30 June 2026). The default way to own Australian small caps therefore buys the opposite of what the phrase promises. The names that do earn are usually the dull ones — the maintenance annuities that belong to no exciting segment and get no poster.

There is a second reason to prefer the annuity over the pure build-out bet: most of the visible pipeline is fiction. AEMO logged 44 GW of data-centre connection requests, and Oxford Economics estimates six of every seven of those megawatts is "phantom demand," leaving less than 8 GW of projects judged likely to proceed. You pay Tasmea to keep what is already built running. Its earnings do not require the 34, or even the 8, to arrive.

Tasmea's model is the tollbooth most investors walk past: it sells scarce crew-hours under long, pre-agreed contracts. It holds more than 125 master service agreements with asset owners, and fund-level analysis puts recurring maintenance at roughly 80% of its earnings. When the binding constraint is qualified labour — something a funded project cannot mint quickly — the customer's alternatives are limited. Tasmea's own answer to that constraint is telling: it bought WorkPac, a workforce-solutions firm, in December 2025, so it could control a share of the crew supply rather than rent it in a tightening market.

The FY26 result, out on August 28, shows who keeps the money. Revenue reached A$1.29 billion, up 136% — but that headline includes WorkPac's low-margin labour-hire gross; organic growth on its own was 18.1%. Pull WorkPac out, and the economics are what a bottleneck owner should earn: EBIT margin of 16.3%, up from 14.0%, and 18.5% in the second half alone, helped by segment mix, price escalations and higher crew utilisation. Underlying EBIT rose 54% to A$118.1 million and underlying NPAT rose 42% to A$73.7 million. Then the number that separates an annuity from a contractor: it converted 125% of that EBIT into operating cash flow, against net debt of just 0.4 times EBITDA and a 37.8% return on capital employed.

Do not file Tasmea under "AI stock." Its exposure runs across mining, power, water and defence — which is precisely why its earnings are real rather than borrowed from a forecast. Whichever boom wins, somebody must keep the fixed plant running. The trade-off is the mirror image: broad, durable exposure means no pure-play kicker, so it will never move the way a hyperscaler contractor does.

Compare the two candidates and the difference is economics, not image. GenusPlus grew revenue 71% and watched net margin fall to 3.8%; Tasmea grew EBIT 54% and watched margin widen. Yet the market prices them about the same — GenusPlus near 37 times trailing earnings, Tasmea near 36 times after a 118% rally in a year. Even SRG Global, its older, better-known cousin, earns roughly 80% of its income from recurring maintenance yet nets only around 4% on revenue. Tasmea is the purer read on the constraint: it carries the highest margin and the best cash conversion of the three.

This is where the hidden-winner test sharpens, because none of this is free and the scarcity has a clock written by management itself. Tasmea guides FY2027 underlying EBITA of A$205–210 million — a minimum 74% jump — but on an implied margin of about 14%, a deliberate step down from the second-half 18.5%. Peak margins are not assumed to repeat. The stock, up more than 100% and trading near its high, is "no longer undiscovered"; expectations are rising with the price. At a market value around A$2.5 billion against guided NPATA of A$130–133 million, it sells for roughly 19 times next year's guided profit. That is a real price for real growth, not a mispricing.

The clock runs on labour. The capacity cure for a crew bottleneck is more crews, and that arrives only as slowly as apprenticeships, migration policy, and rival consolidators who have noticed these margins. Labour availability is already listed by the company as a risk to its segments. So the confirmation metric is FY2027 EBITA landing near the A$205–210 million guide while cash conversion stays above 100%. And the signal that the hidden winner has become ordinary is the one every bottleneck eventually sends: when labour stops being the gate — when hire costs stop rising, crew utilisation slips, or maintenance margins start converging on the builders' — the scarcity rent is over, and so is the reason to pay 19 times for it.

The builder earned the rally. The crew-owner earns the cash.