GE Vernova
The maker of the big gas turbines and grid equipment that power plants and data centres are queuing years to buy.
The scores
Growth3/4 ↑
Growth excluding acquisitions of 13% in nine months and 18% in Q3, orders up 96%; under three years as a company, and Prolec added revenue.
Business quality3/4 →
One of three makers of large gas turbines, raising prices into a shortage; untested in a downturn, and Doosan, fuel cells and engines win buyers.
Financial position4/4 ↑
Net cash of $10.27B, borrowings of $2.85B due from 2031, an undrawn revolver, and a plan funded without outside money.
Execution record3/4 ↑
Beat every 2025 guide and raised 2026's in January, April and July; but the record is two and a half years, and Wind misses.
Per share capture4/4 →
Share count falling, buybacks far exceeding issuance, Prolec paid in cash, no convertibles; the buybacks are funded by deposits, not earnings.
Potential3/4 ↑
Most of the 2028 plan sits in $176.28B of contracted backlog; the reservations behind 2029-2031 can lapse, so beyond 2028 it is expected, not committed.
Risk profile3/4 →
No customer concentration, but one dependency: the gas order cycle, set by others' capital plans, plus open offshore wind claims.
Valuation2/4 →
Buyers pay about $19.90 of company value per dollar of the 2028 outlook's profit before depreciation: the plan, plus years beyond it.
Evidence quality3/4 →
Central figures are filed and the market is measured by the EIA, IEA and PJM; pricing and reservation terms rest on management's word.
If it works 0.7× · if it fails you keep about 75%.
The Verdict
We rate GE Vernova neutral. It builds the gas turbines and grid equipment that power plants and data centres are queuing years to buy, and it is mostly sold out through 2030, so the business is strong. But its cash this year came mostly from customers paying ahead, and its reported profit is mostly one-off gains. The price is full, not cheap: about right only if management's 2028 plan arrives in full. As a bet on its small nuclear reactor and its new solid-state transformer it is priced for success: even a good start leaves the shares worth less than today, and if it fails shareholders keep about three-quarters.
The long caseCountries want steady power without carbon, and a small reactor built from standard parts could be ordered many times once Ontario's first one works.
The evidence under this - 8 claims
- 2026-10-28editedQ3 2026: 122 GW under contract, backlog 64 GW, reserved 58 GW
- 2026-10-28editedThe call put a fourth-quarter gigawatt figure on the pre-registered 130 GW mark.
- The first half of 2026 also carried $0.9B more tax paid, partly on the Prolec GE and Proficy gains, and a voluntary $0.5B pension contribution, so the underlying figure is less negative than minus $1.0B.
- A deposit is not a loan: as long as orders keep growing, new deposits outrun the ones being worked off, the way an insurer's float does, and management says the 30 GW expansion is funded by customer down payments (C18).
- Adjusted EBITDA of $2.15B in the half shows the business is profitable before working capital; cash lags because tax, capex and pension timing sit between the two.
- The 2028 outlook was set before first-half 2026 orders priced more than 20% above Q4 2025 (C17), so it may understate 2028 profit; each plan so far has been raised (C19), and the 2026 guide has now been raised four times.
- Cash of $16.76B nets against value here although $45.75B of customer deposits sits against it; counting deposits as owed would raise company value, not lower it.
- Services on a fleet of HA turbines that doubles as units under contract are commissioned carry profit past 2028 that a 2028 multiple does not see (C37).
- 2026-10-28editedRe-struck at the 2026-10-27 close on the Q3 2026 balance sheet and trailing profit: 19.9 times 2028, from 21.9
- The first unit costs about twice the average of the four, so a first-of-a-kind unit spends more a year and later units less.
- GE Vernova Hitachi supplies the reactor design, fuel and key parts, not the civil works, so its share could be well under a quarter.
- No SST order, price or volume exists; the first unit is an R&D-shared prototype for one hyperscaler (C70).
- Management dates deliveries from a 2027 order to 'later in the decade', so revenue by 2028 is the optimistic end.
The company
When a country or a data centre needs electricity day and night, the usual answer today is a gas power plant, and its heart is a gas turbine: a machine the size of a house that burns gas to spin a generator. GE Vernova is one of the few companies building the biggest ones.
It also sells the transformers and switchgear that connect plants to the grid, services about 7,000 of its turbines already running, and builds wind turbines. Buyers now wait three years or more.
The evidence under this - 4 claims
How it makes money
It sells equipment, then earns for decades servicing it. In Q3 2026 the power business kept 21.4 cents of profit before depreciation from each dollar of sales and the grid business 19.1 cents, while wind earned 2.6 cents after losing money for most of the year.
Buyers pay deposits years before delivery: customer deposits reached $45.7B in September, from $26.0B in January. That cash arrives now; the work to earn it comes later.
The evidence under this - 3 claims
The release credits Power's margin to favourable price and higher volume partly offset by inflation; Electrification's margin includes the acquired Prolec business, and its organic margin was 17.9%, 1.2 points lower than reported.
- 2026-10-28editedThe call said the Q3 Power margin is the base for 2027, which the margin claim had not assumed.
- 2026-10-28editedRe-sourced on the Q3 2026 10-Q: deposits $45.75B, cash $16.76B
- 2026-10-28editedRe-sourced on the Q3 2026 10-Q: remaining performance obligation $196.72B
Where the money comes from, stream by stream
| Stream | Share of sales (FY2025) | How it earns it |
|---|---|---|
| Power services | 34.3% | Long-term service agreements, parts and outages on about 7,000 installed gas turbines plus steam, nuclear and hydro fleets. |
| Power equipment | 16.9% | New heavy-duty and aeroderivative gas turbines, nuclear and hydro equipment, paid for partly in advance through deposits and slot reservations. |
| Electrification | 24.8% | Transformers, switchgear, substations, HVDC links, power conversion and grid software, mostly equipment sold to utilities and data centres. |
| Wind | 23.9% | Onshore and offshore wind turbines and their service; loss-making, with offshore winding down. |
Who runs it
Scott Strazik has run the business since 2021 and ran its gas turbine arm before that; a new finance chief arrives in January 2027. Insiders own well under 1%. Most of the bonus and half of the long-term shares pay on free cash flow and profit. Free cash flow counts customers' down payments, so last year the cash part of the bonus paid its maximum largely on deposits.
The evidence under this - 4 claims
Free cash flow is the measure that customer deposits carry: $4.1B of 2025 operating cash came from working capital (C09) and cash before deposits was about $0.9B (C11), so the largest bonus weight paid its maximum on down payments rather than on earned cash.
Why it wins, and who could beat it
Why customers come back
Building a large gas turbine takes decades of know-how in metal parts that survive temperatures above their melting point. Three companies, GE Vernova, Siemens Energy and Mitsubishi Power, supply turbines for about two-thirds of the gas plants being built worldwide.
When demand outruns what three factories can make, they set the price: GE Vernova says new orders are priced more than 20% above late 2025's. Even Baker Hughes resells its turbines rather than building its own.
The evidence under this - 4 claims
- Q2's higher price per kilowatt is partly mix: more aeroderivatives, which cost more per kilowatt than heavy-duty units, and more combined-cycle scope.
- No price per kilowatt is disclosed in any filing, so the comparison cannot be checked.
The threat
The shortage is sending buyers elsewhere. Doosan of Korea now has twelve big turbines to deliver in the US, sold on delivery time. Data centre builders have taken Mitsubishi and Siemens machines, Oracle swapped planned turbines for Bloom fuel cells, and a Texas site chose 42 Wärtsilä gas engines.
The evidence under this - 4 claims
The rivals: BE, CCJ, ETN, and 5 more that are private, in-house or listed abroad.
Who else does this, rival by rival
| Ticker | Rival | Where it competes | How far along | Priced against it |
|---|---|---|---|---|
| Doosan EnerbilityPriced against it: Listed abroad | Power | On sale | Listed abroad | |
| Siemens Energy and Mitsubishi PowerPriced against it: Listed abroad | Power | On sale | Listed abroad | |
| BE | Bloom Energy solid-oxide fuel cellsPriced against it: Different product | Power | On sale | Different product |
| Wärtsilä 50SG gas enginesPriced against it: Listed abroad | Power | On sale | Listed abroad | |
| Hitachi Energy and Siemens Energy grid businessesPriced against it: Listed abroad | Electrification | On sale | Listed abroad | |
| Conventional transformers from Hitachi Energy and Siemens Energy | Small modular reactors and solid-state transformers | On sale | ||
| Vestas Wind SystemsPriced against it: Listed abroad | Wind | On sale | Listed abroad | |
| CCJ | Westinghouse, Framatome and Rolls-Royce, the reactor makers GE Vernova's 10-K namesPriced against it: One segment only | Small modular reactors and solid-state transformers | Announced | One segment only |
| ETN | EatonPriced against it: One segment only | - | On sale | One segment only |
How big it could get
How big this could get
The International Energy Agency counts 130 GW of new gas power plants ordered worldwide in 2025, the most in 25 years. GE Vernova took 29.8 GW of turbine orders that year, roughly 23% of that. In the US alone, planned gas plants add up to about 65 GW through 2030. Nobody outside the makers says how long a boom this size lasts.
The evidence under this - 4 claims
- The IEA counts whole plants including steam turbines, so a like-for-like turbine share would be higher than 23%.
- GE Vernova also signed slot reservations in 2025 that it keeps out of orders, so its share of commitments made that year is larger than its share of orders.
Who has signed
Contracted work reached $196.72B at 30 September 2026. Of the 122 GW of turbines under contract, 64 GW are firm orders and 58 GW are reservations, which the company itself says may never become orders. NRG reserved 3.6 GW; Hawaiian Electric and Israel's OPC have signed firm purchases.
The evidence under this - 5 claims
- 2026-10-28editedRe-sourced on the Q3 2026 10-Q: remaining performance obligation $196.72B
- 2026-10-28editedQ3 2026: 122 GW under contract, backlog 64 GW, reserved 58 GW
- 2026-10-28editedThe call put a fourth-quarter gigawatt figure on the pre-registered 130 GW mark.
Where the growth is coming from, business by business
Where the growth is coming from
| Business | Share of revenue | Growth: last quarter | Last half-year | Last year |
|---|---|---|---|---|
| Product | 66% | +41.8% | +21.9% | +18.2% |
| Service | 34% | +5.2% | +5.7% | +7.1% |
| Whole company | 100% | +26.8% | +15.0% | +13.4% |
The filed split through 2026-09-30. These rates are not carried forward: the reading below says why.
How we read it: Equipment grew 42% in Q3 (26% organic) against 5% for services, but Prolec added $0.9B and deliveries follow backlog booked earlier. The score believes the plan's 9% a year to 2028, not the 16% the mix implies.
The results
The numbers
Sales grew 17% over the twelve months to Sep 2026, faster than the 9.0% in the year to Dec 2025.
Operating margin is 6.4%, up from −2.8% in the year to Dec 2023.
Running costs take 15¢ of each sales dollar, down from 21¢ in the year to Dec 2022.
Each dollar put into the business earns 25% a year after tax, up from 3.4% in the year to Dec 2024.
It turned 39% of sales into free cash, $17.1B over the last twelve months.
Yearly sales growth has ranged from +5.1% to +12% over the last 3 years.
The share count fell 2.2% in 3.8 years.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-10-28. Fiscal years end in December.
| Dec 2022 | Dec 2023 | Dec 2024 | Dec 2025 | 12 mo to Sep 2026 | |
|---|---|---|---|---|---|
| Size and growth | |||||
| Sales | $29.7B | $33.2B | $34.9B | $38.1B | $44.0B |
| Sales growth | - | +12% | +5.1% | +9.0% | +17% |
| Profit | |||||
| Gross margin | 12% | 14% | 17% | 20% | 22% |
| Operating margin | −9.7% | −2.8% | 1.3% | 3.6% | 6.4% |
| Net margin | −9.2% | −1.3% | 4.4% | 13% | 24% |
| What growth costs, per sales dollar | |||||
| Research and development | 3.3% | 2.7% | 2.8% | 3.1% | 3.1% |
| Selling, general and admin | 18% | 15% | 13% | 13% | 12% |
| Pay in shares | - | - | 0.4% | 0.7% | - |
| Capital spending | 1.7% | 2.2% | 2.5% | 3.4% | 4.8% |
| Cash | |||||
| Cash from operations | −$114M | $1.19B | $2.58B | $4.99B | $19.2B |
| Free cash flow | −$627M | $442M | $1.70B | $3.71B | $17.1B |
| Balance sheet | |||||
| Debt | - | $145M | $60.0M | $352M | $2.83B |
| Returns and shares | |||||
| Return on capital | - | - | 3.4% | 13% | 25% |
| Shares (diluted) | 274M | 274M | 278M | 276M | 268M |
What growth costs
Growth is making GE Vernova more profitable: office and sales costs grow far slower than sales and research spending stays small, so higher prices reach the margin. The real cost of growth is new factories, and customers' deposits are paying for them.
The evidence under this - 3 claims
- Capital spending per sales dollar more than doubled as turbine and transformer capacity is added, and the expansion is paid for with customer deposits (C18, C11), so part of the cost of growth sits on the cash flow statement rather than the income statement.
- Stock pay rose from $155M in 2024 to $257M in 2025, still under one cent per sales dollar.
- The 10-K warns that capacity expansions may outpace realised demand, leaving idle capacity and write-downs (C08).
- Shipments were 3.3 GW in Q2 2026 on the 10-Q unit table, so 20 GW annualised needs output to rise about half again from Q3.
- 2026-10-28editedThe Q3 release raised the year-end target from at least 125 GW to at least 130 GW
- The first half of 2026 also carried $0.9B more tax paid, partly on the Prolec GE and Proficy gains, and a voluntary $0.5B pension contribution, so the underlying figure is less negative than minus $1.0B.
- A deposit is not a loan: as long as orders keep growing, new deposits outrun the ones being worked off, the way an insurer's float does, and management says the 30 GW expansion is funded by customer down payments (C18).
- Adjusted EBITDA of $2.15B in the half shows the business is profitable before working capital; cash lags because tax, capex and pension timing sit between the two.
The last report (22 Jul)
Quarter ended June 2026, filed figures.
Q3 2026 beat the bar the case set: orders $29.3B, up 96% before acquisitions, revenue up 27%, and the power business kept 21.4 cents of each sales dollar as profit. Turbines under contract rose to 122 GW, with 13 GW of reservations turning into firm orders.
Free cash flow was $5.4B, mostly from customers paying ahead, and the yearly cash forecast went to $16B-$17B. It also agreed to buy Trench Group, a grid-parts maker, for about $9.0B, half in debt, closing in late 2027.
What it did to the thesisIt passed all three tests we set: firm turbine orders, deposits and the power margin. But the strength was already in the plan the price pays for, and the new purchase adds debt for a business whose profit is not disclosed, so the view stays Neutral.
The quarter in detail
Sales growth from the previous quarter, filed figures.
Earnings per share $2.47 against $3.17 expected; $2.47 as filed, with no adjustments.
The evidence under this - 6 claims
Free cash flow rose $4.7B on the year mostly from working capital: contract liabilities rose $5.6B in the quarter, and operating cash flow excluding working capital is a small part of the $6.06B (C11).
The release credits Power's margin to favourable price and higher volume partly offset by inflation; Electrification's margin includes the acquired Prolec business, and its organic margin was 17.9%, 1.2 points lower than reported.
- 2026-10-28editedThe call said the Q3 Power margin is the base for 2027, which the margin claim had not assumed.
- 2026-10-28editedQ3 2026: 122 GW under contract, backlog 64 GW, reserved 58 GW
- 2026-10-28editedThe call put a fourth-quarter gigawatt figure on the pre-registered 130 GW mark.
- 2026-10-28editedRe-sourced on the Q3 2026 10-Q: deposits $45.75B, cash $16.76B
- 2026-10-28editedFourth guide raise of the year, on 28 October 2026
What management did with the money
It bought back 4.3M shares for $3.6B in the first half of 2026, at an average $854, and paid $5.275B in cash for Prolec. The buybacks were paid for largely with customer deposits, money it still has to earn by delivering turbines.
Insiders in 2026Q2: 21 purchases and 15 sales, from their Form 4 filings.
The evidence under this - 3 claims
- 2026-10-28editedRe-sourced on the Q3 2026 release and 10-Q: 264.68M shares, authorization raised to $20B
- The first half of 2026 also carried $0.9B more tax paid, partly on the Prolec GE and Proficy gains, and a voluntary $0.5B pension contribution, so the underlying figure is less negative than minus $1.0B.
- A deposit is not a loan: as long as orders keep growing, new deposits outrun the ones being worked off, the way an insurer's float does, and management says the 30 GW expansion is funded by customer down payments (C18).
- Adjusted EBITDA of $2.15B in the half shows the business is profitable before working capital; cash lags because tax, capex and pension timing sit between the two.
The price, and where we disagree
In short. Full, not cheap. A real shortage of power equipment and a good business, but the price already pays for management's 2028 plan arriving in full.
The shares closed at $1,174.86 on 30 June 2026 and were at $895.08 on 27 October, 23.8% lower, with the third-quarter results still to come. Those results arrived on October 28 and beat every test, and the company raised its cash forecast and its buyback authorization to $20B.
The next answer is the fourth quarter in late January: whether turbine slots keep turning into orders toward the 130 GW the company now expects by year-end.
The evidence under this - 4 claims
- The 2028 outlook was set before first-half 2026 orders priced more than 20% above Q4 2025 (C17), so it may understate 2028 profit; each plan so far has been raised (C19), and the 2026 guide has now been raised four times.
- Cash of $16.76B nets against value here although $45.75B of customer deposits sits against it; counting deposits as owed would raise company value, not lower it.
- Services on a fleet of HA turbines that doubles as units under contract are commissioned carry profit past 2028 that a 2028 multiple does not see (C37).
- 2026-10-28editedRe-struck at the 2026-10-27 close on the Q3 2026 balance sheet and trailing profit: 19.9 times 2028, from 21.9
- 2026-10-28editedRe-sourced on the Q3 2026 release and 10-Q: 264.68M shares, authorization raised to $20B
- The question itself is the bear case, asked by Morgan Stanley on the record.
- The answer describes 2027 contracting, not orders, so a year in which reservations stop converting would still satisfy it.
- 2026-10-28editedQ3 2026: 122 GW under contract, backlog 64 GW, reserved 58 GW
- 2026-10-28editedThe call put a fourth-quarter gigawatt figure on the pre-registered 130 GW mark.
How we read the multiple
Counting debt and cash, a buyer pays about $42 for each dollar of the last twelve months' cash profit before interest, tax and depreciation.
Against management's plans:
- About $34 for each dollar of this year's forecast profit.
- About $20 for each dollar of the $11.2B promised for 2028.
- That needs $56B of sales at 20 cents of profit per dollar, against 15.4 cents in the September quarter.
- So the price assumes the plan arrives in full and growth continues after it.
Against its own history. Its own history is little guide: the company lost money for most of it, so every old ratio looks low.
The evidence under this - 3 claims
- The 2028 outlook was set before first-half 2026 orders priced more than 20% above Q4 2025 (C17), so it may understate 2028 profit; each plan so far has been raised (C19), and the 2026 guide has now been raised four times.
- Cash of $16.76B nets against value here although $45.75B of customer deposits sits against it; counting deposits as owed would raise company value, not lower it.
- Services on a fleet of HA turbines that doubles as units under contract are commissioned carry profit past 2028 that a 2028 multiple does not see (C37).
- 2026-10-28editedRe-struck at the 2026-10-27 close on the Q3 2026 balance sheet and trailing profit: 19.9 times 2028, from 21.9
- Wind, guided to lose about $400M in 2026, has to reach a 6% margin by 2028 for the group figure to work.
- The cumulative cash target has not been updated since January, although the 2026 guide alone has since risen by $6.5B, so the target no longer states the plan's cash shape.
Priced against its rivals
None of the listed rivals above sells the same thing to the same buyers as a large part of what it does. A price comparison needs at least two, so the price is judged against its own history instead.
Also checked and set aside: Baker Hughes Company (a substitute, not the same product).
Private or foreign rivals that cannot be priced: Siemens Energy AG, Mitsubishi Heavy Industries Ltd (Mitsubishi Power), Westinghouse Electric Company, Framatome, Rolls-Royce Holdings plc, Vestas Wind Systems, Siemens Gamesa, Nordex, Envision, Goldwind, Hitachi Energy, Siemens AG, Schneider Electric, Mitsubishi Electric, ABB.
The bet: what it pays if it works
The bet is on two new products: the BWRX-300, a small nuclear reactor designed with Hitachi, and a solid-state transformer, an electronic one for data centres. One reactor is being built in Ontario, the transformer is a prototype, and neither reports sales, so the rest is counted separately, at the price this thesis calls attractive.
If a few reactors are building and the first transformer order is delivering by 2028, the shares are worth less than today. If both fail, shareholders keep the rest, about three-quarters of today's value.
| Small modular reactors and solid-state transformers revenue in two to three years | BWRX-300s under construction needed | At the stated multiple |
|---|---|---|
| $350.00M | about 1 at 25% kept | 0.7× |
| $600.00M (the big win) | about 3 at 25% kept | 0.7× |
| $1.60B | about 8 at 25% kept | 0.8× |
To pay 2× (a Fair bet) it would need about $95.04B of Small modular reactors and solid-state transformers revenue (about 520 BWRX-300s under construction at 25% kept); to pay 5× (a Cheap bet), about $321.61B of Small modular reactors and solid-state transformers revenue (about 1.8K BWRX-300s under construction at 25% kept), at the same multiple.
Each row is Small modular reactors and solid-state transformers alone, plus the rest of the company (Gas power, grid equipment, wind and nuclear services (the rest of GE Vernova)) counted at about $179.20B.
The steps it has to take
- Done Darlington's first BWRX-300 licensed for construction
- Done First commercial small reactor contract in North America signed
- Done Two more US technology selections with early work agreements
- Not started NRC licence to construct TVA's Clinch River unit due 2026-12-31
- Not started US-Japan funding turned into a firm US reactor order due 2027-12-31
- In progress Darlington's first unit built due 2029-12-31
- In progress First SST prototype delivered to a hyperscaler due 2026-12-31
- Not started First solid-state transformer order due 2027-12-31
The detail behind the label
The evidence under this - 9 claims
- The budget is Ontario Power Generation's whole project; GE Vernova Hitachi's share of it is not disclosed by either side.
- The later three units are planned for the mid-2030s, so most of the C$20.9B is spent after the bet's horizon.
- The $40B is a government announcement, not an order; no customer or contract value is named for the two technology selections.
- An NRC licence to construct is a permission, and Clinch River's owner has not disclosed a firm order.
- The first unit costs about twice the average of the four, so a first-of-a-kind unit spends more a year and later units less.
- GE Vernova Hitachi supplies the reactor design, fuel and key parts, not the civil works, so its share could be well under a quarter.
- Wind, guided to lose about $400M in 2026, has to reach a 6% margin by 2028 for the group figure to work.
- The cumulative cash target has not been updated since January, although the 2026 guide alone has since risen by $6.5B, so the target no longer states the plan's cash shape.
- The 2028 outlook was set before first-half 2026 orders priced more than 20% above Q4 2025 (C17), so it may understate 2028 profit; each plan so far has been raised (C19), and the 2026 guide has now been raised four times.
- Cash of $16.76B nets against value here although $45.75B of customer deposits sits against it; counting deposits as owed would raise company value, not lower it.
- Services on a fleet of HA turbines that doubles as units under contract are commissioned carry profit past 2028 that a 2028 multiple does not see (C37).
- 2026-10-28editedRe-struck at the 2026-10-27 close on the Q3 2026 balance sheet and trailing profit: 19.9 times 2028, from 21.9
No order, price or revenue is disclosed; a prototype for one customer is not a product line yet.
- The same answer says the shift to orders at scale waits on how quickly AI data centres move to 800-volt DC power, which nobody at GE Vernova controls.
- Management says the MV-UPS could reach orders before the SST, so the two to three times is not the SST's alone.
- No SST order, price or volume exists; the first unit is an R&D-shared prototype for one hyperscaler (C70).
- Management dates deliveries from a 2027 order to 'later in the decade', so revenue by 2028 is the optimistic end.
Where we disagree with the market
The cleanest way to own the power shortage: sold out through 2030, prices rising, cash pouring in.
The demand is real and already priced; the cash is mostly deposits and the profit mostly one-off gains.
The deposits sit in the quarterly filing's cash-flow detail, and the gains in the release's footnotes; the headline cash and profit figures show neither.
The case against us: if deposits keep rising through 2027 while margins climb toward 20%, they are tracking a growing pile of orders, not a peak, and the market is right.
What analysts say, and the evidence
The evidence under this - 4 claims
- The first half of 2026 also carried $0.9B more tax paid, partly on the Prolec GE and Proficy gains, and a voluntary $0.5B pension contribution, so the underlying figure is less negative than minus $1.0B.
- A deposit is not a loan: as long as orders keep growing, new deposits outrun the ones being worked off, the way an insurer's float does, and management says the 30 GW expansion is funded by customer down payments (C18).
- Adjusted EBITDA of $2.15B in the half shows the business is profitable before working capital; cash lags because tax, capex and pension timing sit between the two.
- 2026-10-28editedRe-sourced on the Q3 2026 10-Q: deposits $45.75B, cash $16.76B
What could go wrong, and how we would know
The risks
- Reservations lapse. 58 GW of slots are reservations, not orders; a rival says such deposits simply lapse if the buyer walks away.
- Deposits unwind. If orders peak, the cash customers paid ahead turns into work to deliver with no new cash behind it.
- Buyers go elsewhere. Doosan, Mitsubishi, Siemens, fuel cells and gas engines are all winning buyers who cannot wait.
- Wind keeps losing. Wind lost $657M in the first half of 2026 and the 2028 plan needs it profitable.
- A plan three years out. The price rests on a 20% margin in 2028 that the company has never earned.
- A $9.0B bet on grid parts. The Trench Group purchase is half debt and closes in late 2027; the documents give no figure for what Trench earns.
The evidence under this - 8 claims
- 2026-10-28editedRe-sourced on the Q3 2026 release: first profitable quarter of the year, guide loss widened to $400M-$500M, Vineyard Wind resolved
- 2026-10-28editedThe call answered the wind-optionality question without changing the loss guide.
- The first half of 2026 also carried $0.9B more tax paid, partly on the Prolec GE and Proficy gains, and a voluntary $0.5B pension contribution, so the underlying figure is less negative than minus $1.0B.
- A deposit is not a loan: as long as orders keep growing, new deposits outrun the ones being worked off, the way an insurer's float does, and management says the 30 GW expansion is funded by customer down payments (C18).
- Adjusted EBITDA of $2.15B in the half shows the business is profitable before working capital; cash lags because tax, capex and pension timing sit between the two.
- Wind, guided to lose about $400M in 2026, has to reach a 6% margin by 2028 for the group figure to work.
- The cumulative cash target has not been updated since January, although the 2026 guide alone has since risen by $6.5B, so the target no longer states the plan's cash shape.
What we still do not know
- How big a deposit a reservation carries, and how many reserved gigawatts have ever lapsed.
- GE Vernova's exact share of big turbine orders against Siemens Energy and Mitsubishi.
- What a turbine costs per kilowatt, and how fast that price is rising.
- Who Doosan's US data centre customer is, and whether others are following.
The evidence under this - 5 claims
- The IEA counts whole plants including steam turbines, so a like-for-like turbine share would be higher than 23%.
- GE Vernova also signed slot reservations in 2025 that it keeps out of orders, so its share of commitments made that year is larger than its share of orders.
What would change our view
- Neutral→AttractiveThe plan proves out and the multiple comes in
It needs all of:
- ○Through 2026 more turbine gigawatts are firm orders than reservations (64 against 58 now), and 2027's margin before depreciation hits 15%.checked at the next results, due by 28 Jan
- ○Company value falls to about 16 times the profit before depreciation the 2028 plan promises, from about 20 times.about ≈$719.66 today, 24% below the last close
- Neutral→UnattractiveThe order book or the deposits turn
It needs any of:
- ○Turbine gigawatts under contract fall below 116 at any quarter-end in 2027.checked at the next results, due by 28 Jan
- ○Customer deposits fall two quarters running while revenue still grows.checked at the next results, due by 28 Jan
We would also be wrong if
- The power business earns a margin below 17% in 2027.
- Wind loses money again in the second half of 2026.
What is coming
- 1 January 2027 Claire McDonough becomes chief financial officer: Whether the guidance and the cash framing hold steady under a new finance chief in the middle of the capacity ramp.
Every dated event ahead
- 1 January 2027 Claire McDonough becomes chief financial officer: Whether the guidance and the cash framing hold steady under a new finance chief in the middle of the capacity ramp.
- 28 January 2027 Q4 2026 results and 2027 guidance: Whether gigawatts under contract reached 130, deposits held at or above $45.67B, fourth-quarter cash fell as guided, and 2027 keeps the 2028 plan on track.
- 31 December 2027 Trench Group regulatory approvals and close: Whether the $9.0B purchase clears review in the stated time and what Trench earns, which the documents read do not disclose.
Since we wrote this
- 28 OctEarningsOrders up 96%, deposits $45.7B, forecast raised; $9B grid dealThe quarter passed all three tests the case set, but the price still pays about 20 times the 2028 plan, so the view holds.
- 15 OctNewsVenture Global reserves GE Vernova capacity for eight power islands, with orders from 2027. View heldVenture Global signed a frame agreement with GE Vernova and Baker Hughes for up to 16 power islands.
- 15 OctBet moveBet re-struck: Priced for success to Safe, little upsideRe-priced at the 14 October close.
- 11 OctThesis reviewThe reactor line lifted out and priced as a betThe owner's rule of 2026-10-11: every thesis names the lines its story rides, and an early line owes a bet that cannot be declined.
- 11 OctThesis reviewThe bet now covers the solid-state transformer as well as the reactorA thesis carries one bet, so the owner asked that its early lines be folded into one business and priced together.
- 5 OctThesis reviewThe long case now sits under the verdictAfter the five-name pilot, the owner asked for the long case on every thesis.
- 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the proxy and filings.
- 27 SepThesis writtenNeutral, medium convictionA real turbine and grid shortage with rising prices, but first-half cash was customer deposits, reported profit mostly one-offs, and the price pays for the 2028 plan.
Sources 46
Sources - 46
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S01primarycompany filing2026-01-29the business description, the named competitors, the gas turbine installed base and service agreement counts, the risk factors on slot reservation agreements and capacity, headcount, the FY2025 statements, the annual working-capital table and gas turbine orders and sales in gigawatts for 2023 to 2025backsC01C02C03C08C09C24C34C56C58C59C63C66
- S02primarycompany filing2026-07-22the balance sheet, contract liabilities by segment, remaining performance obligation and its timing, the working-capital components of operating cash, borrowings, share count and repurchases, the Prolec GE remeasurement gain, the Vineyard Wind status and gas turbine units and gigawatts ordered and shippedbacksC04C06C09C10C14C20C21C22C23C26C34C59C63C64
- S03primarycompany filing2026-07-22Q2 2026 results by segment, the organic and adjusted-EBITDA reconciliations, the gas gigawatt bridge (signed, converted, shipped), the raised 2026 guidance and the capital returnedbacksC04C05C07C12C14C18C20C22C23C35
- S04primarytranscript2026-07-22Q2 2026 earnings call transcript (FMP, held 2026-07-22)management's figures with no filed home: order pricing against Q4 2025, 20, 24 and 30 GW capacity steps, the customer mix of gigawatts under contract, data centre orders, the HA fleet running against on contract. The questions are a separate origin: Morgan Stanley asked whether 2026 is the year of peak gas turbine orders, Goldman Sachs asked about industry capacity, Jefferies about slot reservation conversion
On the equipment side, first half 2026 orders were priced more than 20% above 4Q 2025 equipment orders, reflecting a conversion of higher-priced SRAs to backlog.
backsC17C18C20C31C32C37C68C70 - S05primarycompany filing2026-01-28FY2025 results, the $2.9B valuation allowance release, the 2026 guidance and the outlook by 2028 including Prolec GE, the Prolec GE purchase price, the credit rating upgrades and the capex and R&D commitmentsbacksC03C07C12C14C15C19C21C22
- S06primarycompany filing2026-04-22the first 2026 guidance raise and the Q1 gas gigawatt bridgebacksC07C12
- S07primarycompany filing2025-01-22the 2025 guidance, to test delivery against itbacksC19
- S08secondaryvendor data2026-09-27Research pack, frozen at content/GEV/research-pack.json (FMP market tier, FINRA short interest)the 2026-09-25 close, returns, the price-move decomposition, the analyst rating series, short interest and the price history around each earnings daybacksC26C27C28C56
- S09primarycompany filing2026-08-27Ken Parks retires 2 April 2027; Claire McDonough, Rivian's CFO, becomes CFO on 1 January 2027, with a $14.5M make-whole equity award and a $5.0M sign-on paymentbacksC30
- S10primarycompany filing2026-02-05a listed industrial that distributes GE Vernova heavy-duty turbines under licence and names the loss of GE Vernova as a supplier among its risk factorsbacksC25
- S11primarycompany filing2026-02-26Eaton's electrical business described at distribution level, for the cohort reviewread, but no claim rests on it
- S12secondarynews2025-01-27the objection stated at the time of the January 2025 sell-off: a cheaper AI model meaning less power demand
Energy and infrastructure companies - whose share prices had soared in the past year on the promise of powering a massive artificial intelligence buildout - have also seen their stock prices fall early Monday.
backsC56 - S13primarycompany filing2026-09-27three discretionary open-market sales totalling 12,452 shares, none under a Rule 10b5-1 plan, and no purchasesbacksC29
- S20primarygovernment data2026-03-12EIA Today in Energy - Fossil generation could rise with faster-than-expected growth in data center power demand (2026-03-12) US Energy Information Administrationthe rate of the driver, measured by the statistical agency
Between 2020 and 2025, U.S. electricity demand, as measured by net energy for load, grew about 1.7% annually compared with 0.1% annual growth between 2005 and 2019.
backsC40 - S21primarygovernment data2026-09-09EIA Short-Term Energy Outlook, September 2026 - Electricity US Energy Information Administrationthe newest forecast, which trimmed January's 3% for 2027 to nearly 2%
We expect electricity sales in the United States to total 4,135 billion kilowatthours (BkWh) in 2026, an increase of almost 2% over 2025. We expect an additional growth of nearly 2% in electricity sales in 2027
backsC40 - S22primarygovernment data2026-06IEA World Energy Investment 2026 International Energy Agencythe size of a year's global gas plant orders, measured outside the turbine makers
Orders for new natural gas-fired power plants surged to 130 GW in 2025, a 25-year high, with US data centre demand a major factor.
backsC41 - S23primarygovernment data2026-05-13IEA Oil Market Report, 13 May 2026 (public version) International Energy Agencyturbine lead times measured by an agency, not a seller
surging demand in the United States for data centre power generation has pushed lead times for gas compressors and turbines from around 18 months to three years or more
backsC42 - S24primaryregulator2026-05-06PJM Interconnection - Powering Reliability Through Market Design (2026-05-06) PJM Interconnectionwhat a missing turbine costs the buyer in time, stated by the largest US grid operator
Today, a new natural gas turbine plant - the reference technology for the capacity market - requires at least four years under optimistic assumptions from financial investment decision to commercial operation, due to supply chain constraints on major equipment (gas turbines, step-up transformers)
backsC42 - S25secondarycompany filing2026-06-10ERock, Inc. (Enchanted Rock) 424B4 prospectus ERock, Inc.a competitor's filing citing the IEA that GE Vernova, Siemens Energy and Mitsubishi Power supply turbines for roughly two-thirds of gas plants under construction, and naming GE Vernova among its own on-site power rivalsbacksC43
- S26primarytranscript2026-08-05Siemens Energy Q3 FY2026 press call transcript (FMP, ENR.DE, held 2026-08-05) Siemens Energy AGa rival's description of how turbine reservation agreements work, and its statement that its reservations are not counted in its order backlog
Let's say, 10% to 20% that is currently discussed about in terms of reservation agreements means that at the end of the day, we want to make sure that this looks like a down payment, but that down payment will lapse if you don't buy the turbine in other words.
backsC33 - S27primarycompany release2026-08-05Siemens Energy - Earnings Release Q3 FY2026 Siemens Energy AGthe largest rival's gas business booking record orders in the same period
Orders reached another record level of €17.9bn. Growth was driven by a new record order intake at Gas Services, as well as strong increases at Grid Technologies and Transformation of Industry.
backsC44 - S28primarycompany release2026-03-06Doosan Enerbility - contract with a U.S. company for seven gas turbines (2026-03-06) Doosan Enerbilitya fourth large-turbine maker winning US data centre orders, twelve units to deliver in the US in total, on delivery time
On March 6th, Doosan Enerbility announced that it had signed a contract with a U.S. client to supply the company with seven units of its 380MW gas turbine. Starting in May 2029, Doosan is to sequentially deliver gas turbines and generators, one of each per month, to the data center being built by the client.
backsC45 - S29primarycompany release2026-05-15Mitsubishi Power Americas - turbine allocation for Tallgrass' Cheyenne Power Hub (2026-05-15) Mitsubishi Power Americasa US data centre power buyer taking two Mitsubishi M501JAC turbines
The high-efficiency turbines are expected to provide approximately 1,150 megawatts (MW) of site-ready power
backsC46 - S30primarycompany filing2026-03-30Fermi Inc. Form 10-K FY2025 Fermi Inc.a data centre power developer buying three new Siemens Energy SGT6-5000F turbines, with its GE equipment refurbished older unitsbacksC46
- S31primarycompany filing2025-02-13a utility buying up to 1 GW of Bloom fuel cells as a bridge while filing for 2.3 GW of new gas generationbacksC47
- S32primaryother2026-07-02Oracle Corporation - comments to the New Mexico Environment Department on the Project Jupiter air permit (2026-07-02) Oracle Corporationa data centre tenant's own statement that fuel cells replaced planned gas turbines, reused from this repo's Bloom Energy thesis
The project will now use Bloom Energy fuel cells to power the AI data center campus, replacing the previously proposed gas turbines and diesel generators with a lower-impact fuel cell microgrid.
backsC47 - S33primarycompany release2026-04-23Wärtsilä - 790 MW data center order in Texas (2026-04-23) Wärtsilä Corporationan off-grid data centre plant built on gas engines rather than turbines
The 790 MW power plant will operate with 42 Wärtsilä 50SG engines running on natural gas.
backsC48 - S34primarycompany filing2026-02-24a customer's own disclosure of 3.6 GW of 7HA slot reservation agreements with GE Vernova, first projects by end-2029backsC49
- S35primarycompany filing2026-03-30Kenon Holdings Form 20-F FY2025 (OPC Energy, Hadera 2) Kenon Holdingsa binding February 2026 equipment agreement with GE Vernova for the roughly 850 MW Hadera 2 plant in IsraelbacksC50
- S36primarycompany filing2026-08-18Oglethorpe Power 8-K - Q2 2026 investor briefing (2026-08-18) Oglethorpe Powerthe Smarr combined-cycle plant: 1,425 MW on GE Vernova 7HA.03 turbines, in service mid-2029, budget $3.3BbacksC36
- S37primarycompany filing2026-05-08Hawaiian Electric Industries Form 10-Q Q1 2026 Hawaiian Electric Industriesthree contracts to acquire six GE Vernova gas turbine units, delivered in pairs in 2029 and 2030backsC50
- S38primarycompany release2025-07-2229 LM2500XPRESS aeroderivative units, about 1 GW, for a data centre developer; jointly written with GE Vernova, so only partly independent
The announced order of 19 units, booked in June 2025, builds on the first series of 10 units Crusoe ordered in December 2024, and combined, is expected to provide nearly 1GW of electricity.
backsC51 - S39primarygovernment data2026-09-24EIA-860M Preliminary Monthly Electric Generator Inventory, August 2026 data - Planned sheet US Energy Information Administrationplanned US natural-gas nameplate capacity by planned operating year, summed here (our arithmetic, not an EIA sentence): 3.3 GW for the rest of 2026, 12.3 GW in 2027, 17.1 GW in 2028, 14.5 GW in 2029, 18.1 GW in 2030 - 65.3 GW in all
Inventory of Planned Generators as of August 2026
read, but no claim rests on it - S40primarygovernment data2026LBNL - Queued Up: 2026 Edition (OSTI 3376147) Lawrence Berkeley National Laboratorygas capacity seeking US grid connection, a ceiling on the US part of the market rather than a forecast
Active gas (253 GW) capacity in queues increased by 86% year-over-year, whereas solar (773 GW, -19%), storage (749 GW, -16%), and wind (220 GW, -19%) all decreased in 2025
backsC53 - S41primarygovernment data2026-06LBNL - United States Data Center Energy Usage Report: 2025 Update (OSTI 3374245) Lawrence Berkeley National Laboratorythe data centre load base, with a 2030 reference case of 649 TWh in a range of 521 to 843 TWh
The updated estimate of data center electricity use in 2024 is 192 terawatt-hours (TWh), accounting for 4.7% of total U.S. electricity consumption.
backsC54 - S42primarygovernment data2024-07-10DOE - Large Power Transformer Resilience, Report to Congress (July 2024) US Department of Energytransformer scarcity, the Electrification segment's driver, measured by the government
lead times for acquisition of an LPT have become exceptionally long, with 36-month lead times being commonly quoted and maximum lead times reaching as much as 60 months
backsC55 - S43primarycompany filing2026-04-03beneficial ownership as of March 3, 2026, the 5% holders, the executive officers' careers, the 2025 annual bonus metrics, goals and payouts, the 2025-2027 PSU design, the long-term award mix and the 2025 capital returnbacksC60C61C62C63
- S44primarytranscript2026-04-22Q1 2026 earnings call transcript (FMP, held 2026-04-22)management on the SMR pipeline: the US-Japan funding announcement, the Darlington basemat and the Clinch River licence timing
We are inspired and appreciative of the U.S. and Japanese governments' announcement of up to $40 billion for GE Vernova Hitachi to build SMRs in the U.S. ... we expect the NRC to issue the license to construct for Clinch River in Tennessee as soon as 2026.
backsC68 - S45primarytranscript2026-01-28Q4 and full-year 2025 earnings call transcript (FMP, held 2026-01-28)when management expects SMR revenue, what nuclear did to Power's margin, and the first solid-state transformer unit
We expect SMR to contribute meaningfully to the top line of our power business in the next decade. ... Nuclear was a drag on Power's '25 margins, and we expect '26 to be directionally similar.
backsC65C70C72C73 - S46primarygovernment data2026-10-11Government of Canada, Major Projects Office: Darlington New Nuclear Project Government of Canada (Privy Council Office, Major Projects Office)the budget of the first BWRX-300 and of all four Darlington units, and the April 2025 licence to construct
The first SMR will cost $7.7 billion including common infrastructure ... the total budget for all four units is expected to be $20.9 billion
backsC67C69 - S47primarygovernment data2026-10-02FRED series DEXCAUS, Canadian dollars to one US dollar (Federal Reserve H.10) Board of Governors of the Federal Reserve Systemthe exchange rate used to put the Canadian budget in dollars
2026-10-02: 1.4253
backsC69 - S48primarycompany filing2026-03-19the listed owner of Westinghouse, a reactor rival GE Vernova's 10-K names
We own a 49% interest in Westinghouse in a strategic partnership with Brookfield.
backsC71 - S49primarytranscript2026-07-22when the first solid-state transformer order could come, and management's scope per gigawatt with and without it
in our orders still in the first half of the year, the $5 billion, there isn't an SST order. [...] 2027 is as early as orders will come. [...] That $300 million directional of scope per gigawatt today, if you include the MV-UPS, if you include the SST and some of the other things we're working on, we certainly think entitlement just with the things we're already investing in is 2 to 3 times what our scope per gigawatt is today
backsC72C73 - S50primarycompany release2026-10-28third-quarter orders, revenue, segment EBITDA margins, gas gigawatts under contract, free cash flow, the raised 2026 guidance, the buyback authorization and the Trench Group agreementbacksC07C12C18C20C23C74C75C76
- S51primarycompany filing2026-10-28the balance sheet, contract liabilities by segment, remaining performance obligation and its timing, equipment and services revenue, borrowings, share count and repurchases, and the Trench Group purchase agreementbacksC06C10C20C21C74C75C76C77
- S52primarytranscript2026-10-28Q3 2026 earnings call transcript (FMP), held 2026-10-28data/cache/transcripts/GEV-2026Q3.jsonmanagement on Power margin as the base for 2027, the Trench Group rationale, fourth-quarter gigawatts guided to about 13 and a 130 total, turbine and electrification pricing, wind and financing; 11 analyst turns
there's no reason to think that the second half of this year isn't the right way to start to think about our projection into '27
backsC07C21C23C75