GE Vernova
Arms-Dealer of the AI Power Supercycle

Executive Summary

GE Vernova is the public market's purest large-cap bet on the AI power supercycle: the surge in electricity demand, after two flat decades, as data centers come online. It sells the picks and shovels: gas turbines (the baseload powering data centers), grid & electrification equipment (transformers, switchgear, transmission), and emerging nuclear / SMRs. The franchise is genuinely scarce, large-frame gas-turbine slots are sold out into 2028, prices are up ~300% in three years, and the $163B backlog gives multi-year revenue visibility. FY2025 revenue was $38.1B² (+9%) at an 8.4% adjusted EBITDA margin, and margins are inflecting hard: FY2026 guidance (raised at Q1) is for $44.5–45.5B revenue, 12–14% adj EBITDA margin, and $6.5–7.5B free cash flow. The catch is valuation, at ~$1,088 the stock trades at ~49× forward (2026E) EBITDA and ~57× forward earnings, +103% over the past year and near its $1,181.95 all-time high. Our forward EV/EBITDA model implies a base fair value of ~$1,080: essentially at the tape (the Street is more bullish: 38 analysts at "Buy", ~$1,212 average target), with a supercycle-persists bull case to ~$1,425 and a multiple-normalization bear to ~$700.

The Four Pillars of the Debate

Investment Thesis

The GEV debate pits a genuinely best-in-class power franchise, a gas-turbine quasi-monopoly, a $163B backlog, an FCF inflection, against a valuation that already prices a multi-year supercycle at a premium multiple. GE Vernova has to keep converting orders into mid-teens-margin revenue and sustain a ~30×+ forward EBITDA multiple, or that multiple normalizes toward its (already premium) peer band. We frame three scenarios with a forward EV/EBITDA model: pick 2028E adjusted EBITDA, an exit EV/EBITDA multiple, and a discount rate.

Targets are the output of a simplified forward EV/EBITDA model with subjective assumptions, illustrative, not analyst price targets. See sections 08 and 10 to adjust. Probability weights are 30/45/25 (bull/base/bear), backlog visibility anchors the base, while the bull tail (supercycle persists) is fatter than the bear (AI-capex digestion / multiple normalization). The base sits roughly at the current price by design: a high-quality compounder, fully valued here.

Business Overview

GE Vernova was spun off from General Electric on April 2, 2024 as a pure-play energy company, and operates three reportable segments. Power (gas, nuclear and hydro turbines plus services) is the profit engine and the heart of the AI-power thesis: its HA-class gas turbines supply the firm, dispatchable baseload data centers need, and demand has gone vertical. Electrification (grid equipment, transformers, switchgear, transmission, plus grid software and power conversion) is the fastest-growing segment and the most direct beneficiary of connecting all that new load to the grid. Wind (onshore + offshore turbines) is the laggard, dragged by offshore losses. FY2025 mix was roughly Power ~52% / Electrification ~24% / Wind ~24% of $38.1B revenue.

The supercycle is most visible in the order book. Gas-turbine backlog plus slot reservations reached 100 GW in Q1 2026 (from 83 GW at year-end), with large-frame slots sold out into 2028 and pricing up ~300% over three years; ~20 GW is explicitly tied to data centers. Electrification booked $2.4B of data-center equipment orders in a single quarter: more than all of FY2025. Total orders rose 71% organically, lifting backlog to $163B (management targets $200B by 2027). On top of that, GE Vernova is building the GE Hitachi BWRX-300 small modular reactor, the first commercial SMR under construction in North America (Darlington, Ontario; construction began May 2025, targeted online ~2029).

The Three Segments: Power · Electrification · Wind

GE Vernova's differentiator is owning both generation (gas + nuclear) and the grid gear (Electrification) that connects it, the two bottlenecks the AI build-out runs into. The risk is concentration: the order surge is heavily levered to a single end-market (data centers) and to gas, where slots are sold out but cancellable, and offshore Wind remains a drag on the consolidated print.

The AI Power Supercycle

For two decades US electricity demand was essentially flat. That era is over. AI data centers, electrification of transport and industry, and onshoring are driving the first sustained demand inflection in a generation, and the generation fleet and grid are not ready. Data-center load alone is projected to roughly triple this decade. The binding constraints are firm generation (you cannot run a 24/7 data center on intermittent renewables alone) and grid interconnection (transformers and transmission carry multi-year lead times). GE Vernova sells into both bottlenecks.

Gas turbines are the near-term winner, dispatchable baseload, supplied by an effective three-way oligopoly (GE Vernova, Siemens Energy, Mitsubishi Power), so when demand surged, pricing and lead times surged with it (GEV slots sold out into 2028, prices +~300% in three years). Nuclear / SMRs are the longer-dated answer. On the grid side the same scarcity benefits Electrification's transformers and switchgear. The whole debate reduces to durability: how long does the data-center capex cycle run, and how much is already in the backlog and the stock?

The Competitive Set

GE Vernova's edge is breadth and scarcity: it is one of three large-frame gas-turbine makers and a top grid-equipment supplier, so it captures both bottlenecks of the build-out. The threat is not a single competitor but the cycle itself, premium peers (Vertiv, Quanta) trade at similar multiples, so the cohort re-rates together if AI capex disappoints.

Where GEV Sits in the Cycle

The Opportunity: Power Demand & Backlog

The prize is the multi-decade re-wiring of the power system. Forecasters put global power-generation + grid capex in the trillions of dollars through 2035, and the near-term accelerant is data centers: AI load is projected to grow from roughly ~35 GW today toward ~110 GW by 2030 and higher, each gigawatt needing firm generation, transformers, switchgear and transmission. GE Vernova sells the equipment for all of it, across both generation and the grid.

GE Vernova's slice is unusually well-positioned. As one of three large-frame gas-turbine makers and a top-tier grid supplier, it captures both the generation and interconnection bottlenecks. The $163B backlog already locks in years of revenue, and management's $200B-by-2027 target implies the order surge continues. Unlike a TAM-share debate, GEV's near-term revenue is largely already contracted, so the bull/bear split is less about demand and more about the multiple: the bull case is that the backlog understates demand (reservations convert and grow); the bear case is that a chunk is data-center-concentrated and cancellable if AI capex slows.

Financial Health

GE Vernova is a profitable, cash-generative business with margins inflecting hard. FY2025 revenue was $38.1B (+9% organic), adjusted EBITDA $3.2B (8.4% margin), and free cash flow $3.7B. Q1 2026 continued the trend: revenue $9.3B (+16%), adjusted EBITDA nearly doubling to $0.9B (9.6% margin, +390bps). One caveat on GAAP: headline net income is distorted by one-time items in both directions, a ~$2.9B deferred-tax valuation-allowance release in FY2025, and ~$4.3B of M&A gains (Prolec remeasurement + Proficy sale) in Q1 2026, so read adjusted EBITDA and FCF, not GAAP net income or the trailing P/E.

The trajectory is the story. Management raised FY2026 guidance at Q1 to revenue $44.5–45.5B, adjusted EBITDA margin 12–14%, and free cash flow $6.5–7.5B (up from $5.0–5.5B). The margin expansion comes from price (gas-turbine and transformer pricing up sharply), mix (Power and Electrification growing while loss-making Wind shrinks), and operating leverage on the $163B backlog. Free cash flow roughly doubles year-on-year, and it is seasonally front-loaded by customer advance payments on the record order book (Q1 2026 alone generated $4.8B of FCF), so the full-year guide is the cleaner run-rate.

The balance sheet is solid: ~$10.2B cash at Q1 2026 against modest, investment-grade debt (GE Vernova issued $2.6B of notes in the quarter and keeps gross leverage below 1× adjusted EBITDA), so it remains roughly net-cash positive even after funding the Prolec acquisition and buybacks. That supports a doubled dividend ($2.00/yr) and a $10B buyback authorization while still investing in capacity. The capital structure is a strength, not a risk, the open question is valuation, not solvency.

Free Cash Flow: Inflecting, Funds Dividend + Buyback

Orders & Backlog

Orders are where the supercycle shows up first. Q1 2026 orders were $18.3B, up 71% organically, with growth across all three segments. Total backlog reached $163B (up $13B sequentially, including ~$5B from the Prolec GE acquisition), and management targets $200B by 2027. The marquee data point: Electrification booked $2.4B of data-center equipment orders in a single quarter: more than all of FY2025. On the Power side, gas-turbine backlog plus slot reservations reached 100 GW (from 83 GW at year-end), with large-frame slots sold out into 2028 and management guiding to ≥110 GW by year-end; ~20 GW is explicitly tied to data centers. GE Vernova signed 21 GW of new gas-turbine agreements in Q1 alone, spanning the US, Vietnam, Mexico, Brazil and Canada, global demand, not a single-market phenomenon.

Unlike a pre-revenue order book, GE Vernova's backlog is largely contracted equipment-and-services revenue with multi-year delivery, genuine visibility, not just a pipeline. The watch items are (1) conversion: slot reservations are firmer than an LOI but can still be cancelled if a data-center project slips; and (2) concentration: a large share of the incremental order surge is data-center-driven, so the backlog's durability tracks hyperscaler capex. Backlog margin is also improving, equipment margin in backlog grew ~$8B in 2025 (~6 points of accretion), which is what powers the forward EBITDA-margin guide.

Identifiable Demand & Catalysts

Backlog and orders are the visibility behind the thesis; the risks are conversion and concentration. The single best leading indicator is the data-center share of incremental gas and grid orders quarter to quarter, whether the supercycle is extending or digesting.

Valuation & Comps

GEV trades at roughly ~49× EV/2026E adjusted EBITDA (~57× forward earnings). On trailing FY2025 adjusted EBITDA of $3.2B the multiple is ~90×, but EBITDA is inflecting so fast (margin 8.4% → 12–14% → mid-teens) that the forward multiple compresses quickly, to ~49× on 2026E and the high-20s against 2028E EBITDA on today's EV. The question is not whether the headline looks rich (it does); it is whether the compression is fast enough to justify the price.

Our model asks the reverse question: to justify ~$1,088, what is the market paying as a 2028 exit multiple? Discounting back two years at 9%, the answer is ~34× 2028E EBITDA on ~$10B of EBITDA, a genuine premium. Our disciplined base uses exactly that (≈$10B 2028E EBITDA, ~34× exit, 9% WACC) and lands at ~$1,080: essentially at the tape. The bull case (≈$11.5B, ~38.5× exit, 8% WACC) reaches ~$1,425, matching the Street high; the bear (≈$9B, ~24.5× exit, 10% WACC) ~$700.

Against peers, GEV sits at the high end of an already-premium cohort: Vertiv ~38×, Quanta ~40×, Siemens Energy ~35×, Eaton ~28× forward EV/EBITDA. The premium is arguably warranted, GEV's large-frame gas franchise is a tighter oligopoly than most of these, but it leaves little margin for error. The sell side is more bullish than we are: 38 analysts rate GEV "Buy" with a ~$1,212 average target (high $1,424, low $836), and they have been actively raising targets through June 2026 (Baird to $1,400, Goldman to $1,328, Argus to $1,300). They underwrite a sustained premium; our disciplined base does not stretch quite as far.

Valuation Model: Forward EV/EBITDA

Pick a ~2028 adjusted EBITDA, an exit EV/EBITDA multiple and a discount rate; the model discounts the implied 2028 enterprise value back two years, adds net cash (~$5B), and divides by shares (~269M). Use the presets, then move the sliders. The base case (~$10B 2028E EBITDA, ~34× exit, 9% WACC) lands roughly at the current price, the model's way of saying GEV is fairly valued, with the bull/bear split driven mostly by the exit multiple.

EBITDA Ramp & Implied EV ($B)

Sensitivity: Implied Price ($)

Catalysts & Roadmap

Roadmap & Milestones

Scenario Targets

This calculator uses the same forward EV/EBITDA math over a 2-year horizon to 2028E. Move the inputs to see the implied price and a rough scenario-likelihood read. The swing factor is the exit multiple, at ~$10B of 2028E EBITDA, every 4× of exit multiple is worth ~$130/share. Note that the base inputs land right at the tape: GEV is a high-quality compounder, fully valued here.

Bull vs Bear

Technicals & Positioning

GEV has been one of the great post-spin performers, up ~103% over the past year and +70% year-to-date, near its $1,181.95 all-time high (April 2026). The 52-week range, $482 to $1,182, is a ~2.5× spread that reflects how fast the AI-power supercycle re-rated the stock from a sleepy ex-GE spin-off (April 2024) into a $290B+ market cap. At $1,088 the stock sits only ~8% below its high.

Positioning is very different from the speculative names in the library, GEV is a large-cap S&P 500 industrial held broadly by institutions, not a heavily-shorted retail favorite. It trades with the AI-power cohort (Vertiv, Eaton, Quanta, Constellation) and with broader capital-goods sentiment. The technical risk is momentum unwinding: a high-flying, richly-valued leader can give back a lot quickly on any growth or rate scare, even with strong fundamentals, which is exactly why we'd rather add on pullbacks than chase the high.

Risk / Reward Tool

Ask the Thesis AI-assisted checking…

Describe a scenario in natural language; the assistant returns a structured impact analysis against GE Vernova's three-segment model (Power / Electrification / Wind), the AI-power supercycle, the $163B backlog, capital returns, and a forward EV/EBITDA scenario framework. Powered by Claude via a Cloudflare Worker proxy (Anthropic key held server-side).

Note: The assistant reasons from the dashboard's data snapshot and thesis sections, it does not browse the web or access real-time fundamentals beyond what's in data.js. Treat its responses as scenario-modeling support, not as primary research.

Risks