GE Vernova — The Sold-Out Decade

Its Turbines Are Sold Out Through 2030 — and at ~25× Its Own 2028 Target, So Is the Stock. What the Purest Winner of the AI Power Buildout Is Actually Worth, and Why the Multiple, Not the Business, Is the Risk

TON618 Capital Research · As of July 22, 2026 · Download PDF

A valuation, not a trade. Fair value is an estimate on an unusually wide band here, not a price target. Data: GE Vernova's FY2025 Form 10-K, its Q2-2026 results and raised guidance (reported July 22, 2026), and SEC/press disclosures (SEC EDGAR, CIK 0001996810); market data as of the July 22, 2026 close ($985.03, down 8.7% on the day of the Q2 print). Fact ("the filing shows…") is separated from opinion ("we estimate…"); figures are DISCLOSED, REPORTED, CONSENSUS, or ESTIMATED. This note stands on its own, but it is the single-name expression of our thematic note The Power Wall (July 19, 2026), which argues that the durable slice of the AI premium settles in power, not silicon; §6 carries the link. It is the supply-side valuation in our five-note AI-infrastructure series — the reader's guide maps all five and is the place to start. Independently fact-checked before release.

§0 Verdict

What this note is, and is not. It is a valuation of the equity built to answer the question our thematic note leaves open: the AI premium is migrating to the power layer — but is the migration already in the price of the layer's best company? The finding is not that GE Vernova is a bad business or a short — it is the highest-quality name in the entire AI-infrastructure complex and it carries only modest, investment-grade leverage. The finding is that its valuation, not its franchise, is the risk: at ~24.5× a 2028 target it embeds a secular outcome with no margin of safety, and the Street's $1,221 mean target ($836–$1,424, 38 analysts) prices beyond even our bull case. This is a name we would want to own — materially cheaper than here.

§1 The Story in Plain English

Read this if you read nothing else.

GE Vernova makes the machines that generate and move electricity — above all the large gas turbines that provide the firm, around-the-clock power AI data centers need, plus the grid equipment (transformers, switchgear) that ties them in. It was spun out of General Electric in April 2024, and in a little over two years its stock has gone from the mid-$100s to nearly $1,000 — up ~86% off its 52-week low even after falling 8.7% on the Q2 print — because the market realized that the AI buildout runs into a power wall, and GE Vernova sells the one thing that wall is made of.

The business behind that move is real, not a story. Its order book — the backlog of equipment already sold but not yet delivered — is about $176bn (as of Q2-2026), and its gas-turbine slots are sold out through 2030. That gives it something almost no other AI-linked company has: visibility. It largely knows what it will build and sell for the next several years. And it is getting more profitable fast — the slice of each revenue dollar that turns into cash profit has roughly tripled in two years and is guided to double again to 20% by 2028. It has more cash than debt. On the business, there is very little to dislike.

So why would it be worth less than its price? Because the price has already assumed the good news — and then some. Even after falling 8.7% on a strong Q2 that beat and raised guidance, at $985 the market is valuing GE Vernova at about 25 times the profit it hopes to earn in 2028 — a profit that requires everything to go right (margins doubling, no stumble in the notoriously lumpy power-equipment business, its money-losing Wind unit fixed). Put the future cash flows through a disciplined model, or apply a fair multiple to that 2028 target, and the value comes out around $672 — roughly a third below the price, and below even the most cautious Wall Street target. The whole gap is a single question: is the AI power boom a decade-long secular shift, or a cycle that will crest and normalize like every power-equipment cycle before it? At today's price you are paying full fare for the decade-long answer. Our thematic note argues the power scarcity is genuinely durable — but "durable" and "already fully paid for" are different things, and this is the second one.

§2 The Business

Two engines and a drag — and a backlog that makes the whole thing legible.

GE Vernova runs in three segments, and they are not equal:

• Power — the gas, nuclear and hydro turbine franchise, and the heart of the thesis. FY2025 revenue ~$19.8bn at a 14.7% EBITDA margin [DISCLOSED]; Q2-2026 ran hotter still — revenue $5.5bn (+14%) at an 18.8% segment EBITDA margin, on Power orders of $16.7bn (+134% organic). This is where the sold-out turbine book lives, and where the AI power demand lands.
• Electrification — grid equipment: transformers, switchgear, power conversion. FY2025 revenue ~$9.6bn, and the margin story of the company: EBITDA margin jumped from 9.0% to 14.9% in a year [DISCLOSED] as grid-equipment shortages (the same ones our thematic note documents) let it raise price — and Q2-2026 extended it to an 18.4% segment margin on revenue up 68% (with Prolec). This is the second durable-premium leg, and it is inflecting.
• Wind — onshore and, painfully, offshore. FY2025 revenue ~$9.1bn at a −6.6% EBITDA margin (−$598m) [DISCLOSED], the offshore business the drag; Q2-2026 losses persisted (−13.6% segment margin). It is the one segment working against the story, and the market is paying for the other two despite it.

The unifying asset is the backlog. At year-end 2025 it stood near $150bn — $64bn of equipment and $86bn of higher-margin services — grew to ~$163bn by Q1-2026 (including the Prolec GE grid-transformer acquisition), and reached $176bn by Q2-2026, up $13bn sequentially [DISCLOSED]. Electrification's equipment backlog alone hit $40.6bn (+69% YoY). For an industrial, a backlog this size against a ~$46bn annual-revenue run-rate is close to four years of visibility — the single feature that separates GE Vernova from most AI-linked names, whose demand is a forecast rather than a signed order.

§3 The Backlog Is the Asset

Visibility is what you are really buying — and it is why the thesis says this premium is durable.

The reason GE Vernova is the single-name expression of the Power Wall thesis is the turbine book. Gas-turbine backlog and slot reservations went from 62 GW to 83 GW across 2025, to 100 GW by Q1-2026, then to 116 GW at Q2-2026, with management now targeting ≥125 GW by year-end 2026 (raised from ≥110 GW) and describing the slots as sold out through 2030 [DISCLOSED]. To fill them the company is ramping gas-turbine output to 20 GW/yr in Q3-2026, 24 GW by 2028, and a targeted 30 GW by 2030. Roughly a fifth of the book is tied to data centers; separately, Electrification data-center orders reached over $5bn year-to-date — more than double the 2025 total. The rest is the broader electrification of the economy, which means AI is bidding for turbine slots against every utility and industrial buyer at once.

This is the durable-premium mechanism from the thematic note, made into an income statement. Because only three companies on earth make these turbines and none can add capacity in under three years, the backlog is not just revenue — it is priced revenue, and the pricing is rising. Unlike a GPU order, which reprices down as packaging capacity doubles annually, a turbine slot booked today for 2029 delivery carries pricing that reflects a shortage no competitor can undercut within the delivery window. That is why the premium here is structurally more durable than in silicon, and it is the strongest argument for the stock.

But visibility cuts both ways, and the note's discipline requires saying so. A backlog booked to 2030 tells you what GE Vernova will build; it does not tell you what happens to new orders if the AI capex that drives ~20% of it plateaus — and it does not, by itself, justify any particular multiple on the earnings that backlog produces. Visibility lowers the risk to the 2028 target. It does not lower the price you are paying for it.

§4 The Financials

A genuine margin-expansion story, a cash flow flattered by its own growth, and two GAAP distortions to see past.

The operating story is real and improving. Adjusted EBITDA margin has gone 5.8% (FY2024) → 8.4% (FY2025), ran 11.3% in Q2-2026 (+280bps YoY; 10.5% H1), is guided to 12–14% in 2026 and a 20% target for 2028 [DISCLOSED] — a doubling driven by price (the shortages), mix (services), and the Electrification inflection. At Q2 the company raised 2026 guidance: revenue to $45.5–46.5bn (from $44.5–45.5bn) and free cash flow to $11.5–12.5bn (from $6.5–7.5bn), while holding the adjusted-EBITDA-margin band at 12–14% — i.e., the raise is a cash-timing story, not a step-up in earnings power. Free cash flow rose from $1.70bn (FY2024) to $3.71bn (FY2025), and the balance sheet is a fortress: after a $2.6bn senior-notes issue in February 2026 to help fund the Prolec acquisition, net cash was ~$7.3bn at Q1 and ~$10.3bn at Q2-2026 ($13.1bn cash less $2.85bn total borrowings) — modest, investment-grade (BBB/BBB+) leverage rather than debt-free [DISCLOSED].

Two cautions the careful reader needs:

• The cash flow is boosted by its own growth. Q2-2026 free cash flow was ~$5.1bn and H1 was ~$9.9bn — already more than double all of FY2025 — because a booming backlog means large customer advances (prepayments on orders). That is real cash, but it is a growth phenomenon: the fact that the FCF guide jumped ~$5bn while the margin band held tells you the raise is working-capital timing, not profit. When order growth slows, the advance tailwind fades and FCF converges back toward earnings. We normalize this in the model rather than extrapolating the advance-inflated figure.
• Two GAAP distortions. FY2025 GAAP earnings carry a large tax benefit, and H1-2026 GAAP net income of $5.4bn is inflated by a ~$4.0bn pre-tax gain on the Prolec remeasurement (plus a $0.3bn Proficy-sale gain) — so H1 diluted EPS of $19.96, and the ~31× trailing GAAP P/E, are not meaningful. Adjusted EBITDA and free cash flow are the clean metrics, and the valuation uses those.

§5 Valuation

Two methods, deliberately, because a net-cash industrial should be triangulated. Both land below the price.

Method 1 — cash flow. A three-stage FCFE with a thesis-consistent runway (six explicit years, 2026–31, reflecting the backlog's visibility rather than a short cyclical fade), discounted at ~9–9.5% cost of equity for a net-cash industrial, plus the ~$10.3bn of net cash. Base case: ~$570. It comes in below the price because even a durable, high-single-digit-growth cash-flow stream, disciplined discount rate, and a normalized terminal do not compound to ~$265bn of equity value quickly.

Method 2 — EV/EBITDA on the 2028 target. The market's own language for an industrial. The company targets $52bn of revenue at a 20% margin = ~$10.4bn of adjusted EBITDA in 2028 (a 2028 framework it left unchanged at Q2, even as it raised 2026 guidance). Apply a premium-but-defensible 22× to that target and discount back and you get ~$750 on this method; grant a secular-grower 27× (bull) and it is ~$920, apply a cyclical-aware 16× (bear) and it is ~$545.

Blend the two methods per scenario and weight them, and the central estimate is ≈ $672, on a band of $427 to $897 (bull $897 / base $659 / bear $427).

What the price assumes, stated plainly. At $985.03, GE Vernova's enterprise value is about 24.5× its 2028 target EBITDA (an EV near $254bn on ~$10.4bn of 2028 target EBITDA) — i.e., the market is paying 25 times a profit that is ~2 years away and requires the margin to double, with essentially no discount for time or execution. On a nearer-term basis it is ~42× 2026 guided adjusted EBITDA; the 2026 free-cash-flow multiple looks deceptively cheap (~22× the raised $11.5–12.5bn guide) only because that FCF is advance-inflated, as §4 explains. These are secular-growth-stock multiples on a power-equipment company. To defend them you must extend the super-cycle well past 2028 at a sustained peak multiple — which is a coherent view (it is our bull case), but it is a view, and at this price it is the only view that works.

Why we are below the Street. The 38-analyst mean target is $1,221, above even our bull case, and the low is $836, above our central estimate. The difference is almost entirely the multiple applied to the 2028 target and how far the super-cycle is assumed to run — not the operating forecast, which we largely share. We discount the target for time and execution and decline to assume a peak multiple persists for a decade; the Street, in effect, does both. A reader who is confident the power scarcity is a ten-year secular event should weight our bull case and can justify a price near today's. We think that is the less disciplined assumption, and we say so.

§6 Cyclical or Secular? The Question the Whole Price Turns On

This is the Power Wall thesis, applied to GE Vernova's own multiple.

Every dollar of the gap between our $672 and the $985 price is one question: is the AI power buildout a secular regime — a decade-long, structurally-supplied-short expansion that makes GE Vernova a compounding growth company — or a cycle — a sharp, order-driven super-cycle that will crest, normalize, and re-rate to the low industrial multiple cyclicals earn at peak earnings?

The bull, secular case is the thesis of our companion note, and it is strong: the power bottleneck (turbines sold out to 2030, interconnection waits lengthening past five years, only three OEMs) is genuinely more durable than the silicon bottleneck, and it supports a higher multiple than a normal cycle. The bear, cyclical case is the history of the industry: power-equipment demand has always mean-reverted, ~20% of the turbine book rides on AI capex whose own payback is unproven (our AI Capex Payback Clock), and turbine super-cycles have burned investors who paid peak multiples for peak orders before.

Our resolution, consistent with the thematic note: the premium is more durable than a normal cycle but not infinitely durable, and — critically — already priced as if it were. GE Vernova at ~24.5× the 2028 target is not priced as a cyclical at peak; it is priced as a secular grower with the durability question already answered in its favor. That is why the valuation, not the business, is the risk. If the power scarcity proves as durable as the thesis argues and the multiple holds, today's buyer does fine. If it proves even partly cyclical, the re-rating is severe — because the starting multiple leaves no room.

The symmetry with our NVIDIA valuation is deliberate. NVIDIA, the demand-side supplier, trades below a disciplined fair value because the price assumes the demand never plateaus; GE Vernova, the supply-side winner, trades below fair value because the price assumes the scarcity is secular rather than cyclical. Each is the best business on its side of the migration, and each is priced for its own force to break its way. The reader's guide sets the two side by side.

§7 The Risk Ledger, Both Ways

A stock up ~86% off its low, at ~24.5× a forward target — and one that just fell 8.7% on a beat-and-raise — owes the reader both tails.

What makes us wrong on the upside:

• The super-cycle is secular. If AI-driven power demand compounds into the 2030s and the interconnection/turbine bottlenecks stay tight (the thematic note's central case), GE Vernova compounds EBITDA at a high rate for a decade and the multiple is defensible.
• Margins beat the 20% target. Continued grid-equipment pricing power and services mix could push margins past 20%, and the Electrification inflection has room to run.
• The balance sheet is a weapon. ~$10.3bn of net cash plus investment-grade access funds capacity expansion, buybacks (~$3.9bn returned to holders year-to-date), and grid M&A from a position of strength — the Prolec deal was part-funded with a $2.6bn notes issue, but off a nearly-unlevered base — optionality the model does not fully credit.

What makes us right, or worse:

• Cyclical normalization. Turbine orders peak ~2027 and new bookings slow; the multiple compresses to a cyclical-peak level and the stock re-rates hard from ~24.5×.
• The payback clock breaks. The ~20% of the turbine book tied to data centers softens if hyperscaler AI capex plateaus (our companion note's central risk) — with a lag, but it comes.
• Wind stays broken. Offshore losses persist or deepen, capping group margins short of the 20% target.
• Execution on the ramp. Doubling margins while ramping turbine output to 24 GW by 2028 is not free; power-equipment history is littered with warranty charges and project write-downs.
• The re-rating itself. The near-doubling off the 2025 low is its own risk: a crowded, momentum-owned stock at a peak multiple falls fastest when the narrative wobbles, regardless of the fundamentals.

The asymmetry we see: the upside cases justify roughly today's price to modestly above; the downside case (cyclical normalization) is a stock worth ~$427. Paying $985 accepts that asymmetry at its least favorable point.

§8 Recommendation, and What We Could Not Verify

Recommendation. Fair value ≈ $672, band $427–$897, versus $985.03. We are constructive on the franchise and disciplined on the entry: GE Vernova is the highest-quality name in the AI-infrastructure complex and the cleanest expression of a durable thesis, but at ~24.5× its 2028 target the price already pays for the durability in full, and the risk has shifted from the business to the multiple. Q2-2026 sharpened the point rather than changing it — a genuine beat-and-raise, yet the stock fell 8.7%, because the good news was already in the price and the raise was cash-timing, not a lift to the 2028 earnings power the valuation turns on. This is a name to own on a cyclical or market pullback, not at a full price for a secular outcome that is assumed rather than proven. We are explicitly not short — a net-cash, backlog-visible, structurally-advantaged compounder is a dangerous short, and the thesis behind it is real. We log this fair-value call against the $985.03 July-22 close to our signal ledger for out-of-sample scoring, per house practice.

What we could not verify, and how it would move the number.

• Forward consensus is wide and we treat it as a cross-check, not an anchor: sell-side puts FY2027 EPS near $24 and a price-target mean around $1,212–1,221. Our valuation is anchored on the company's own 2028 framework ($52bn revenue, 20% adjusted-EBITDA margin, ≥$22bn cumulative 2025–28 free cash flow) and runs bull/base/bear around it, so the dispersion in third-party estimates does not move the central conclusion.
• Segment-level detail. The Q2-2026 release provides full segment results, incorporated above (Power 18.8%, Electrification 18.4%, Wind −13.6% segment EBITDA margins); the group-level valuation does not hinge on the split.
• The enterprise-value bridge at the line level. We confirm the aggregate net-cash position from the Q2-2026 balance sheet (≈$10.3bn: $13.12bn cash less $2.85bn total borrowings); the difference from any line-level reconciliation is immaterial to fair value.

None of these moves the central estimate materially, because the valuation is dominated by the multiple on the 2028 target and the cyclical-versus-secular judgment — which §5 and §6 make explicit. The swing factor is not hidden in a filing. It is whether the power super-cycle is a decade or a cycle — and even after the Q2 sell-off, at $985 the market is still pricing the decade.

§9 Disclosures

Information only. TON618 Capital. This report is for information purposes only. Nothing here is an offer to sell or a solicitation of an offer to buy any security, fund interest, or digital asset, and nothing here is personalized investment advice or a recommendation regarding any instrument.

Publisher's exclusion. All research is published solely as general, impersonal information of regular circulation. It is not tailored to the objectives or circumstances of any individual and is not issued in connection with compensation from any client. The Fund has no clients and distributes all research free of charge. On that basis it publishes in reliance on the publisher's exclusion from the definition of "investment adviser" under the Investment Advisers Act of 1940 (§202(a)(11)(D); cf. Lowe v. SEC, 472 U.S. 181 (1985)).

Registration & conflicts. TON618 Capital is not registered as an investment adviser or broker-dealer in any capacity. The Fund is a Bitcoin fund and may hold or transact in the securities or digital assets it discusses. This report values GE Vernova (NYSE: GEV) and references, among others, Siemens Energy, Mitsubishi Heavy Industries, and NVIDIA (NVDA); the Fund holds no position, long or short, in any of them, and has no economic interest in the price of any security named here. The Fund receives no compensation from any party in connection with its research.

Use of AI. Artificial intelligence is used in the creation of this research. All methodology and data integrity are reviewed and approved before publication by TON618 Capital's Chief Investment Officer, Keyth Beck; errors may nonetheless occur, and readers should verify independently.

CFA. This report was prepared to align with CFA Institute analytical standards (methodology only). CFA® and Chartered Financial Analyst® are registered trademarks owned by CFA Institute. That reference describes the analytical framework applied; it does not imply the report was prepared, reviewed, or authored by a CFA charterholder, and the report is not issued, reviewed, endorsed, certified, or approved by — nor affiliated with — CFA Institute.

Risk & feedback. Past performance is not indicative of future results. Digital assets and equities are volatile and may result in total loss of capital. Corrections and feedback are welcome — please direct them to CIO Keyth Beck at keyth@ton618capital.com.