# data center revenue — X 热门讨论 (2026-09-13 23:40 UTC)
## @HunterAllen4 (THE GAP FATHER) · 09-13 11:27 · ♥32 ↻3 💬13 $VRT
You smell that rip salad? 🥗
The AI chip trade is crowded, obvious and heavily owned.
The less obvious opportunity is what sits beneath it: the power, generation, cooling, transmission and construction bottlenecks required to keep AI data centers running.
Everyone knows $NVDA. More people are starting to understand $VRT, $ETN, $GEV and $PWR. The bigger opportunity is understanding what happens underneath them as AI infrastructure scales.
U.S. electricity demand is projected to hit record levels of 4,270 TWh in 2026 and 4,349 TWh in 2027, while U.S. data-center capacity is expected to more than triple from 2024 levels by 2030 to roughly 125 GW. This is becoming an electricity infrastructure cycle, not just a semiconductor cycle.
The demand is already showing up across the stack. Eaton’s Electrical Americas data-center orders were up roughly 240% YoY, with backlog up 44%. Vertiv is sitting directly in the middle of this buildout across critical power, UPS, power distribution and thermal management. FY2025 revenue reached $10.23B, up 27.7%, while backlog hit $15B, more than doubling YoY.
Q2 2026 accelerated the story: $VRT generated $3.27B revenue, up 24%, adjusted operating profit of $738M, up 51%, adjusted EPS of $1.52, up 60%, and $925M adjusted FCF, up more than 200%.
Management raised 2026 guidance to roughly $14B revenue, $6.65–$6.75 adjusted EPS and $2.5B FCF. This is what a real AI infrastructure compounder looks like when the demand starts flowing through the income statement.
And $VRT just made the stack even more interesting. Its planned $1.45B acquisition of UtilityInnovation Group, with up to another $1.15B earnout, moves Vertiv upstream into microgrids, onsite generation orchestration, behind-the-meter power architecture and grid interconnect.
In other words, Vertiv is no longer just helping cool and power the rack — it wants to help solve how the site gets power in the first place.
So yea we’re gonna rip gang.
Then look at generation. $GEV now has a $176B backlog, Q2 orders of $24.2B, and gas-equipment backlog plus slot reservations of 116 GW, targeting at least 125 GW by year-end. It is also working toward roughly 30 GW of annual gas-turbine production by 2030. Data-center electrification orders alone topped $5B in the first half of 2026.
That creates the opportunity most people aren’t connecting: time-to-power. When traditional generation and grid timelines don’t match the speed of AI construction, developers need alternatives — behind-the-meter generation, modular gas, reciprocating engines, fuel cells and steam-cycle systems. FastPower-type projects are already being built around 1.2 GW and $2.4B of generation capacity.
Then look downstream. $PWR reported $9.56B Q2 revenue, $33.6B RPO and $53.4B backlog, up 21.5% from year-end 2025. The money is spreading from compute into generation, electrical equipment, transmission, substations and construction.
That’s what I’m researching: the cross-stack bottlenecks. $GEV shows you the generation shortage.
$VRT shows you what happens between the power source and the rack. $ETN shows you the electrical path. $PWR shows you the grid buildout. But underneath those giants are smaller companies solving specific problems that become more valuable as the system gets constrained.
That’s where I’m hunting.
Not the crowded chip trade.
The overlooked infrastructure underneath it — before the market connects the dots.
Subscribe to see the hidden layer I just posted. the smaller companies, overlooked bottlenecks and infrastructure plays most investors haven’t connected yet. https://x.com/HunterAllen4/status/2099097655859159465