GE Vernova
Capital-equipment OEM plus high-margin long-tail services: sells gas/steam/wind turbines and grid hardware (HVDC, transformers, switchgear, software) against multi-year reservation-and-order backlog, then earns recurring revenue on a large installed-base service fleet (parts, upgrades, long-term service agreements).
The thesis on this name
State of AI Compute
The only name owning BOTH grid-power tolls (gas-turbine generation + Electrification gear) plus the genuinely durable 15-25yr turbine SERVICES annuity — but verify inverted the lean: the durable moat is the services annuity (which the deep-dive declines to act on), the gear margin-expansion is a peak-cycle extrapolation, and at ~37x+ fwd it's a watch-for-en…
State of Data-Center Power
The only name owning BOTH the gas-turbine generation chokepoint AND grid Electrification gear, layered over a genuinely durable 15-25yr turbine-services annuity. Gas-turbine backlog hit 100GW in Q1 2026 (up from 83GW at end-2025), CEO guiding reservations 'sold out through 2030' by YE2026, total backlog ~$163B (fact). Western heavy-frame turbine output is low-tens-of-GW/yr against a ~2,300GW US interconnect queue — an order-of-magnitude supply gap that holds pricing. The verify-inverted read: the durable moat is the services annuity, not the cyclical generation order book — size for that. Rich at ~37-71x fwd depending on source, so accumulate on pullbacks, not at the chase price.
State of Data-Center Power
The gas-turbine chokepoint: ~100GW backlog sold out to 2030, datacenters now ~20% of it — own the binding bottleneck of the power stack.
State of Data-Center Power
~100GW backlog sold to 2030, datacenters ~20% of it; the single most-binding supply constraint in the stack, owned at a still-reasonable ~30x.
State of Nuclear Energy
The diversified arms that monetize the nuclear build-out and data-center power bid without single-reactor or pre-revenue risk — reference-weighted as durable-compounder optionality and best owned via NUKZ. GEV is the picks-and-shovels capital-goods winner of the 100->400GW build; TLN and VST are the merchant-power names capturing the data-center PPA bid (TLN cleanest, VST cheaper but levered).
Earnings, margins, COGS & capex
GEV is inflecting from a low-margin GE spinoff into a structurally short-supply power-equipment franchise: FY25 revenue $38.1B (+9%) with adj EBITDA margin 8.4%, and Q1 FY26 revenue $9.3B (+16%) with adj EBITDA margin 9.6% (+390bps). Power (gas) is the cash engine (FY25 ~14.7% EBITDA margin), Electrification is the fastest grower (+61% in Q1 FY26 on datacenter demand), and Wind is the drag (~-$400M FY25 EBITDA). The 2028 framework targets $52B revenue at a 20% adj EBITDA margin and ~$200B backlog, with pricing now rising faster than inflation on a sold-out gas slot book.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~79¢ is cost of goods and ~15¢ operating expense, leaving ~6¢ of operating profit (~51¢ net).
Revenue trend
Margins
up
up
up
up
up
up
COGS structure
COGS splits into cost of equipment (the larger bucket — heavy frames, blades, generators) and cost of services. Drivers: specialized castings/forgings and hot-section airfoils (limited global suppliers — Howmet, Precision Castparts), steel/nickel-alloy and rare-earth input costs, skilled labor at US plants (Greenville SC, Schenectady NY), and Wind contract loss accruals/warranty. Margin lever is mix-shift to services + pricing on the sold-out gas book outrunning input inflation.
Capex
~$0.4B in Q1 FY26; ~$6B committed 2025-2028 (~$9-10B incl. R&D). Funds gas-turbine capacity expansion (target 20GW annualized output by Q3 FY26, ~24GW by 2028), US factory builds (~$600M / 1,500 jobs across Greenville, Schenectady), and Electrification grid-hardware capacity — i.e. capex is aimed squarely at relieving the supply bottleneck that is its pricing power.
Latest earnings
Beat. Adj EPS ~$2.01 vs ~$1.67-1.90 consensus (~+20%); revenue $9.34B vs ~$9.26B consensus. Stock rose ~14% on the print (fact, though sources vary on exact adj-EPS figure).
FY26 raised to revenue $44.5-45.5B, adj EBITDA margin 12-14%, FCF $6.5-7.5B. 2028 framework: ~$52B revenue, 20% adj EBITDA margin, ~$200B backlog, ≥$22B cumulative FCF 2025-2028. Dividend doubled to $0.50/qtr; buyback authorization raised to $10B.
- Gas turbine backlog
- 100GW (from 83GW YE25), ~110GW target YE26
- Datacenter share of gas backlog
- ~20%
- Orders / organic growth
- $18.3B / +71% organic
- Total backlog
- $163B (incl. Prolec GE)
Growth drivers
- AI datacenter load — gas backlog at 100GW with datacenters ~20%; Electrification booked $2.4B of datacenter equipment orders in Q1 FY26 alone (>all of FY25)
- Pricing power on a sold-out gas slot book into 2030 — pricing rising faster than inflation
- Electrification segment compounding (+61% revenue Q1 FY26 — backlog targeted to double to ~$60B by 2028) on grid buildout + interconnection demand
- Installed-base services flywheel — high-margin parts/upgrades/LTSAs on a growing global fleet
- US reshoring + electrification capex supercycle (data centers, manufacturing, grid replacement)
- Margin self-help: Wind loss reduction, capacity ramp absorbing fixed cost, mix toward services
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-01-29. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’25
Bull & bear
The scarcity thesis was tested this quarter and it held with room to spare: 20 GW of new gas contracts signed, the book to 116 GW and heading to 125 GW-plus, agreements now running into 2031, and a brand-new 30 GW-for-2030 capacity plan that customers are pre-paying for. Cash conversion re-rated with it - FY26 FCF guidance nearly doubled to $11.5-12.5B - on a balance sheet holding ~$10.3B net cash. Margins are compounding at the segment level (Power 18.8%, Electrification 18.4%), and the stock is ~16% off its June peak while the operating case got better, not worse.
- The bear's own falsifier failed: 18 GW of new slot reservation agreements plus 2 GW of orders in Q2 against a marker of <5GW/qtr - gas backlog plus SRAs went 100 to 116 GW with the year-end target raised to at least 125 GW and agreements signed into 2031
- Capacity, the real constraint, is being lifted faster than planned and cheaply: 20 GW annualised run-rate already reached, 24 GW in 2028, new 30 GW in 2030 - lean, inside the existing factory footprint, 'all funded by customer down payments'
- Cash is the proof: $5.11B FCF in a single quarter (more than all of FY25) and FY26 guidance raised to $11.5-12.5B from $6.5-7.5B, with $13.1B cash against $2.85B of borrowings
- Margin expansion is broad, not a mix trick - Power segment EBITDA 18.8% (+240bps), Electrification 18.4% (+390bps reported, +700bps organic), GAAP gross margin 21.3% vs 20.3%, and H1 equipment orders priced >20% above 4Q FY25
- Electrification is now a genuine second engine: orders +66% organic at ~1.7 book-to-bill, equipment backlog $40.6B (+69% YoY), data center orders >$5B YTD - more than double all of 2025
- Capital return continues through the drawdown: $3.9B returned YTD, 4.3M shares repurchased at an average $854, dividend $0.50/qtr
The demand story got better and the earnings story got worse, which is exactly the risk when a stock carries a scarcity multiple. GAAP gross margin is 21.3% and operating margin 5.9%; EPS missed consensus by roughly 20%; adjusted EBITDA missed; and the adj EBITDA margin guide was the one number management did NOT raise even as revenue and FCF went up. Wind is deteriorating rather than narrowing, and the spectacular free cash flow is $6.4B of customer down payments - a liability against future deliveries, not earnings. An 8.7% one-day drawdown on a 2% EBITDA miss is the market telling you how little cushion the multiple carries.
- Earnings quality is the weak seam: diluted EPS $2.47 against consensus of $3.04-3.17, adj EBITDA $1.25B vs ~$1.28B, GAAP operating margin only 5.9% - and the ~28x TTM P/E is itself flattered by the one-time Q1 Prolec gain sitting in trailing net income
- The margin guide did not move. Revenue raised, FCF nearly doubled, adj EBITDA margin held at 12-14% - with H1 running 10.5%, the full-year band now depends on an H2 step-up that Q2 did not demonstrate
- Wind is getting worse, not better: Q2 EBITDA -$275M vs -$165M, H1 -$657M vs -$312M, orders -40% organic against a FY26 guide of only ~-$400M in losses; US onshore demand is soft on permitting delays and tariff uncertainty with no called inflection
- The headline FCF is down-payment financing: $6.4B of the quarter's cash was working capital from order and slot-reservation prepayments. Real cash, but it converts into delivery obligations - and it means cash generation and profit generation have decoupled
- Being sold out into 2031 cuts both ways: it locks multi-year fixed-price exposure across an uncertain inflation, Section 232 tariff and thin-supply-chain window. Management already flagged that H1 Wind shipments carried fewer contractual tariff protections
- Valuation still assumes flawless execution - consensus average target $1,233 but with a $836 low, the shares are 16% off the 30 Jun peak, and Citi cut its target to $1,125 (Neutral) on the print even as Morgan Stanley raised to $1,350 (Overweight)
What it is worth
Blended P/E and EV/EBITDA on the 2028 framework, cross-checked against sell-side targets. ~32.5x trailing P/E and rich EV/EBITDA reflect a re-rating to a scarcity/secular-growth multiple; the debate is durability of the 20% (2028) margin + backlog growth vs. a contracted-out order book.
Sub-$900 — multiple compression toward a heavy-industrial ~20x if new gas slots fall <5GW/qtr, Wind losses persist, or AI-capex digests; +60% YTD run unwinds
~$1,190-1,350 (BNP Neutral target ~$1,190; Jefferies Buy ~$1,350) — fairly valued near current ~$1,090-1,130 with the multi-year framework largely in the price
~$1,400 (UBS)
capacity ramp + datacenter demand sustain 20%+ margin and backlog growth beyond 2028
Bull and bear agree on the business quality and disagree only on the price — the stock is priced for flawless execution of the 2028 ramp.
SWOT
Strengths
- Owns the scarcest input in the AI buildout — gas-turbine slots sold out into 2030 (100GW backlog) with pricing power as supply, not demand, is the constraint
- Net-cash, investment-grade balance sheet ($10.2B cash, no net debt) funding a $6B capacity ramp + $10B buyback without leverage
- Large installed-base services franchise — recurring high-margin parts/upgrades on a global fleet that grows as equipment ships
- One of only three credible large-frame gas-turbine OEMs globally (with Siemens Energy, Mitsubishi); ~70%+ combined capacity share
- Margin inflection underway — adj EBITDA margin from 8.4% (FY25) toward a 20% (2028) target
Weaknesses
- Wind segment is structurally loss-making (~-$400M FY25 EBITDA), hit by offshore contract losses, tariffs, and weak onshore demand
- Low absolute gross margin (~18-19%) and only mid-single-digit organic growth versus a ~32x P/E — operationally a heavy-industrial cadence
- Long-cycle equipment delivery exposes it to multi-year cost-inflation and execution/warranty risk on fixed-price contracts
- Reported earnings noisy — Q1 FY26 net income flattered by a $4.5B one-time Prolec gain; GAAP optics overstate run-rate profitability
- Heavy dependence on a thin specialized supply chain (forgings, castings, hot-section airfoils) it does not fully control
Opportunities
- Datacenter demand still early — hyperscalers are only ~10% of orders today but ~1/3 of the early-stage paid-reservation pipeline
- Capacity ramp (20→24GW) directly converts to revenue against a sold-out book — each new slot is pre-priced demand
- Electrification backlog doubling to ~$60B by 2028 on the grid-replacement + interconnection supercycle
- Nuclear (SMR/BWRX-300) and grid software as optionality on long-duration clean-firm power demand
- Pricing leverage persists as long as the industry is supply-constrained — multi-year reservation deposits de-risk the book
Threats
- Backlog is largely contracted-out — ~90% of gas capacity booked to end of decade, so growth becomes constrained by capacity adds, not demand; bears (BNP) see new slots falling <5GW/qtr by late 2026
- Valuation (~32x P/E, +60% YTD pre-downgrade) prices in the good news — multiple compression risk on any execution slip
- Datacenter power demand could be met by alternatives (grid interconnect, nuclear, behind-the-meter renewables+storage) or AI-capex digestion
- Supply-chain concentration (forgings/castings) and tariffs raise input cost and delivery risk
- Cyclical/policy exposure — gas-plant permitting, emissions regulation, and a potential AI-capex pause
Moats, dependencies & bottlenecks
Moats
5-7 years (backlog sold into 2030; pricing power lasts while supply-constrained) 100GW gas backlog with multi-year reservation deposits; the binding bottleneck of the power stack — the actual investment thesis.
HA-class engineering, hot-section metallurgy, and certification are near-impossible to replicate; only 3 credible OEMs GEV/Siemens Energy/Mitsubishi ~70%+ of capacity; no new entrant can stand up large-frame capacity this decade.
20-30 years (turbine fleet service life) Every unit shipped locks in decades of high-margin parts/upgrades/LTSA revenue; switching OEM mid-life is impractical.
backlog ($30B→$60B) gives visibility but more competition than gas Real demand tailwind, but ABB/Hitachi/Siemens compete harder here than in large-frame gas.
Net-cash position funds capacity ramps and deposits that smaller rivals can't match, but not a standalone moat.
Dependencies
castings, hot-section airfoils) Rotor forgings and hot-section blades are the primary supply bottleneck — concentrated among Howmet, Precision Castparts and a few global forges; a delay cascades into turbine delivery.
Datacenters ~20% of gas backlog and the growth narrative; an AI-capex digestion or hyperscaler pause hits the highest-multiple demand. Today ~90% of orders are still utilities/IPPs, which cushions but doesn't eliminate.
Gas-plant viability depends on gas prices, interconnection queues, and emissions rules; tightening regulation or a carbon-policy shift dents new-build demand.
Long-cycle heavy industrial — the current order surge is the up-leg of a capex cycle that has historically reverted; backlog gives 4-5 years of visibility.
Input-cost inflation and tariffs (already hitting Wind) compress fixed-price contract margins; rare-earth access is a strategic exposure.
Advantages
- Sold-out gas backlog into 2030 (100GW) with pricing rising faster than inflation — pre-priced, deposit-secured demand
- One of only three large-frame gas-turbine OEMs; rivals are equally capacity-constrained, so share is sticky
- Net-cash, investment-grade balance sheet funding capacity + $10B buyback simultaneously
- Decades-long, high-margin installed-base services annuity on a growing global fleet
- Second growth engine in Electrification (+61% Q1 FY26, backlog doubling to ~$60B by 2028) diversifying beyond gas
- Direct, named hyperscaler/datacenter access — Chevron/Engine No.1 (7HA), Crusoe (29 LM2500XPRESS aeroderivatives), Microsoft via Joulent
Weaknesses
- Wind is structurally loss-making (~-$400M FY25 EBITDA) with offshore/tariff drag and no near-term fix
- Low absolute gross margin (~18-19%) and mid-single-digit organic growth against a premium ~32x multiple
- Earnings quality noise — Q1 FY26 net income inflated by a $4.5B one-time Prolec M&A gain
- Backlog is a double-edged sword — ~90% of gas capacity contracted to 2030 caps incremental order momentum (bears see <5GW/qtr new slots by late 2026)
- Thin, concentrated supply chain for the most critical parts (forgings/castings) it doesn't control
- Long-cycle fixed-price contract exposure to multi-year input-cost inflation and execution/warranty risk
Bottlenecks
- Gas-turbine manufacturing capacity — output ramping 20GW (Q3 FY26 target) toward 24GW (2028); the throughput ceiling that converts backlog to revenue
- Specialized supply chain — rotor forgings and hot-section airfoils/blades from a handful of global suppliers (Howmet, Precision Castparts) are the hard constraint
- Skilled labor + US factory capacity (Greenville, Schenectady) needed to staff the ramp; ~1,500 new jobs in flight
- Wind segment profitability — offshore contract losses and tariff exposure cap consolidated margin
- Capital intensity of expansion — ~$6B capex 2025-2028 must land on schedule to relieve the bottleneck without overbuilding into a cycle peak
Top signals & trends
Top signals
BNP's core bear marker — decelerating new bookings would confirm the 'backlog is a cap' thesis even as revenue keeps converting.
Capacity, not demand, is the constraint — every GW of added throughput is pre-priced revenue.
12-14% FY26 guide is the proof-point that the franchise is re-rating on profitability, not just revenue.
Rising direct hyperscaler conversion would extend the demand runway; a stall would validate AI-capex-digestion fears.
Narrowing losses unlock consolidated margin; widening (as in Q1 FY26 on offshore/tariffs) is the persistent drag to monitor.
Pricing currently outrunning inflation; the spread is the single biggest swing factor on the 2028 margin bridge.
Trends
The structural driver — datacenters ~20% of gas backlog and the fastest-growing order source; positions gas as clean-firm AI power for the next decade.
All three OEMs are sold out / capacity-constrained, handing pricing power to incumbents; a multi-year tailwind that's also the moat.
Drives the Electrification segment (backlog $30B→$60B by 2028); aging grid + interconnection demand is durable and less cyclical than gas.
US-made advantage and onshoring support pricing, but tariffs raise input costs and have already hit Wind margins.
Cancelled/delayed offshore projects and tariff exposure keep the Wind segment loss-making with no clear path to scale.
GEV has nuclear (BWRX-300) optionality, but SMRs and renewables-plus-storage are long-run substitutes for some gas demand.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Hot-section castings, airfoils, and forged components — a primary, capacity-constrained supplier of the most critical turbine parts.
(Berkshire Hathaway) Large structural castings and seamless rolled rings/forgings; concentrated global supply that gates turbine output.
Nickel-alloy and forging feedstock (e.g. ATI Inc, ticker ATI; Carpenter Technology, CRS) for hot-section parts.
Industrial gas / hydrogen + controls suppliers Hydrogen-fuel testing and control-system components for next-gen turbines; broad electronics supply base.
Transformer JV acquired/consolidated in Q1 FY26 (~$4.5B gain) — now in-house Electrification capacity, formerly a partner.
JV reserving seven 7HA turbines for co-located gas-plus-datacenter power, up to 4GW, first in-service ~end-2027.
Private AI-datacenter developer; 29-unit LM2500XPRESS aeroderivative deal (~1GW) for AI datacenters.
Hyperscaler datacenter power venture targeting reliable capacity for AI workloads.
NRG, Vistra, Southern Co, NextEra) Still ~90% of gas-turbine orders — the demand base beneath the datacenter headlines (Vistra VST, Southern SO, NextEra NEE).
~10% of orders today but ~1/3 of early-stage paid reservations — the marginal, fastest-growing buyer.
Closest large-frame gas-turbine peer (Germany-listed ENR.DE / OTC SMEGF); nearly doubled turbine sales 100→194 units 2024-25, ~60% of order GW tied to datacenters — equally capacity-constrained, so competes for the same slots, not for share.
Japan-listed (7011.T / OTC MHVYF); ~17% of Asia gas-turbine market, third of the large-frame oligopoly. Strong in Asia; also supply-constrained.
Hitachi-owned (Japan-listed 6501.T); leading HVDC/transformer competitor against the Electrification segment.
Swiss-based, US-listed ADR; competes in grid hardware, switchgear, and electrification — a peer to GEV's fastest-growing segment.
Denmark-listed (VWS.CO / OTC VWDRY); leading wind-turbine OEM — competes with GEV's (loss-making) Wind segment.
Alternative clean-firm power paths for datacenters (NuScale SMR ticker SMR; Bloom fuel cells BE) — long-run substitution risk, not direct OEM competition today.