Eaton Corporation ETN

Reference table

1. What the business is, and how the disclosure actually works

Eaton is an Irish-domiciled (country code IE, CIK 0001551182) power-management company with approximately 97,303 employees, NYSE-listed, chief executive Paulo Ruiz Sternadt.

The segment structure changed between the vendor's last annual snapshot and the current 10-Q, and the change matters. The seg_prod block reports five segments for 2025 — Electrical Americas, Electrical Global, Aerospace, Vehicle and eMobility. The 10-Q for the June 2026 quarter reports FOUR: Electrical Americas, Electrical Global, Aerospace and Mobility, the last being Vehicle and eMobility combined. That combined Mobility segment is the one being separated into Dana (Section 3), so the recombination is not cosmetic — it is the segment being prepared for departure.

Segment results, second quarter 2026 (10-Q filed 2026-07-31, US$M)

The four segments sum to $8,531M against reported net sales of $8,531M — exact.

The order book, which is the actual thesis

And the consolidated figure from the revenue-recognition note: "total backlog at June 30, 2026 was approximately $24.1 billion. At June 30, 2026, approximately 71% of this backlog is targeted for delivery to customers in the next twelve months."

Three readings. First, orders are growing more than twice as fast as shipments in the largest segment — 41% against 18% — which is the definition of a supply-constrained market and is why the backlog grew 33% while revenue grew 18%. Second, the acceleration is new: Electrical Americas backlog growth went from +17% a year ago to +33% now, and Electrical Global from +1% to +103%. Third, $24.1B of backlog against $30.0B of trailing revenue with 71% deliverable inside a year is roughly ten months of shipments already contracted. For an industrial, that is unusual forward visibility and it is the single strongest piece of downside protection in this dive.

The margin problem, in the company's own bridges

Electrical Americas margin fell from 29.5% to 27.5%. The 10-Q: "a 470 basis point decline from higher commodity inflation, partially offset by a 260 basis point increase from higher sales."

Electrical Global margin fell from 20.1% to 19.8%, with "a 460 basis point decline from higher commodity and wage inflation, partially offset by a 240 basis point increase from higher sales, a 100 basis point increase from favorable mix, and an 80 basis point increase from operating efficiencies."

Aerospace margin ROSE from 22.2% to 22.8%, absorbing "a 300 basis point decline from higher commodity and wage inflation."

Consolidated gross margin: 33.5% against 37.0%, a 350 basis point decline.

This is a price-cost timing problem, not a competitive one, and the bridges prove it — in every segment the volume, mix and efficiency contributions are positive and the single negative line is input cost. But it is running at roughly 470 basis points and has now appeared in two consecutive quarterly comparisons. The favourable reading is that Eaton has not lost pricing power; the unfavourable one is that a company shipping into a 1.3 book-to-bill should be able to reprice, and has not yet.

Geography (seg_geo) — verified clean

The five lines sum exactly to reported 2025 revenue of $27,448M, and the 2024 lines sum exactly to $24,878M. No gap, no double-count. seg_prod for 2025 sums to $27,463M against $27,448M — a $15M difference, 0.05%, consistent with intersegment elimination. Both vendor segment tables are usable and we say so.

2. The quarter — and why GAAP and adjusted point in opposite directions

Second quarter 2026, from the 10-Q (US$M except per share):

The $1.04 per share of adjustments is the largest in this file's history — $0.49 of acquisition and divestiture charges, $0.50 of intangible amortisation, $0.05 of restructuring — and it is 33% of the adjusted figure. A year earlier the same adjustments were $0.44, or 15%.

And the corporate lines explain the whole gap:

$256M of incremental annual-run-rate amortisation and interest arrived in a single quarter, and both are direct consequences of the Boyd Thermal purchase. The interest line is the more consequential of the two: it is cash, it does not go away with an accounting adjustment, and at $201M a quarter it is $804M annualised against $444M of trailing net interest — a permanent charge against forward earnings that consensus must be carrying.

We use adjusted EPS for forward valuation because that is the basis on which estimates are struck, and we print the GAAP figure alongside every time. But we note explicitly that intangible amortisation of $255M a quarter is a real economic cost of the assets Eaton bought, and that adding it back — as both the company and the analyst consensus do — flatters the earnings power of an acquisition-built business.

The annual record

Revenue has compounded 11.3% a year since 2020 and operating margin has risen from 13.0% to 19.1%. That is a genuinely transformed industrial, and it happened before the data-centre order surge that is now driving the shares.

3. The two transactions the vendor payload does not contain

Boyd Thermal — closed, and it re-shaped the balance sheet

From the 10-Q, verbatim: "On March 12, 2026, Eaton acquired Boyd Thermal for $9.55 billion, net of cash acquired. Boyd Thermal is a U.S. based global leader in thermal components, systems, and ruggedized solutions for data center, aerospace and other end-markets. Boyd Thermal employs more than 6,000 people with manufacturing sites across North America, Asia, and Europe. Boyd Thermal is reported within the Electrical Global business segment."

And: "Eaton's 2026 condensed consolidated financial statements include Boyd Thermal results of operations, including segment operating profit of $121 million on sales of $524 million, from the date of acquisition through June 30, 2026."

$524M of sales in 110 days is a run-rate near $1.8B; a $121M segment operating profit on that is a 23.1% margin, above Electrical Global's 19.8%. The allocation is explicitly preliminary: "These preliminary estimates will continue to be revised during the measurement period... These differences could have a material impact on Eaton's preliminary purchase price allocation." Goodwill from the deal was initially $4,901M and has already been adjusted down $1,169M to $3,733M in one quarter — which tells you how movable the allocation is, and how much of the $255M quarterly amortisation is still an estimate.

Two further acquisitions sit in the same period: Ultra PCS (Aerospace, goodwill $849M) and Fibrebond (goodwill $541M, measurement period ended March 2026). Cash paid for acquisitions in the first half of 2026 was $11,079M against $1,450M in the first half of 2025.

The Dana Reverse Morris Trust — announced 2026-06-11, absent from the payload entirely

From the 8-K, verbatim: "On June 10, 2026, the Company entered into definitive agreements with Merger Partner pursuant to which the Company will separate its Mobility segment (the 'SpinCo Business') and combine it with Merger Partner in a Reverse Morris Trust transaction." Eaton transfers the Mobility business to SpinCo, distributes SpinCo shares to Eaton shareholders, and immediately merges a SpinCo subsidiary into Dana Incorporated. "Under the terms of the Merger Agreement, the Company will receive approximately $1.1 billion in cash distribution and the Company shareholders will receive newly issued shares of the combined company such that the Company shareholders will own at least 50.1% of the combined company's outstanding shares."

What leaves: the Mobility segment, which on the vendor's 2025 figures was Vehicle $2,505M plus eMobility $618M = $3,123M, or 11.4% of revenue, running a 13.0% operating margin in the June quarter against 27.5% for Electrical Americas.

What this does to the numbers, stated as arithmetic rather than as a forecast. Removing $3,123M of revenue at a ~12-13% margin from a $27,448M base at a 19.1% margin raises the remaining group's operating margin by roughly 90 basis points and lowers the revenue base by 11.4%. Eaton receives $1.1B of cash — 5.5% of current net debt — and its shareholders retain a majority economic interest in the combined vehicle business rather than losing it.

None of this is in the vendor file. seg_prod still shows five segments including Vehicle and eMobility; est still projects revenue of $36,088M for 2027 and $39,315M for 2028 on the pre-separation base; nothing in quote, profile or km_ttm reflects a pending disposal of 11.4% of the company. A reader working from the payload alone would model a company that is about to stop existing in its current form. As of the 2026-06-30 balance sheet Mobility is not classified as held for sale, so the financial statements are also still on the pre-separation basis — correctly, because the agreements were signed on 2026-06-10 and closing conditions remain.

4. Balance sheet and the enterprise-value rebuild

As at 2026-06-30, from the 10-Q filed 2026-07-31 (US$M):

> Net debt at 2026-06-30 = $20,611M − $483M − $212M = $19,916M.

> Net debt at 2025-12-31 = $9,895M − $803M = $9,092M.

> The increase is $10,824M in six months.

The vendor file contains both the right answer and the wrong one, and this is the defect that matters most on this name. bal_a supplies annual balance sheets only, the most recent dated 2025-12-31, reporting totalDebt of $11,169M (including $637M of operating leases) and netDebt of $10,547M. But km_ttm.enterpriseValueTTM of $193,528M less the market capitalisation of $172,685M implies net debt of $20,843M — within 4.7% of the filed figure. The derived block reflects the Boyd acquisition; the balance-sheet block predates it. The file disagrees with itself by roughly $9.4 billion, and a reader who takes bal_a at face value will understate leverage by more than half.

The rebuilt enterprise value

> Market cap $172.68B (388.4M × $444.72 = $172,729M; vendor reports $172,685M on 388.3M — a 0.03% difference)

> + total debt $20.611B

> − cash $0.483B − short-term investments $0.212B

> = corrected enterprise value $192.65B

The vendor reports $193.53B. The difference is $0.88B, or 0.46% — attributable to the $715M of operating lease liabilities the vendor's totalDebt convention includes. This is one of the rare cases where the vendor's enterprise value is right and its balance sheet is wrong, and the reconciliation is close enough that we accept the vendor figure for cross-checks while using our own for every multiple.

On the leverage ratio, our method, stated so it can be checked. The vendor's 3.44x uses a trailing GAAP EBITDA of $6,058M that is itself depressed by the acquisition charges — the June quarter's ebitda of $1,345M is lower than March's $1,485M despite revenue rising 14.5% sequentially. We construct trailing adjusted EBITDA from adjusted earnings: trailing adjusted earnings of approximately $4,817M, grossed up at the disclosed 20.6% effective tax rate to roughly $6,067M of adjusted pre-tax profit, plus trailing net interest of about $444M and depreciation (excluding the intangible amortisation already adjusted out) of roughly $560M, gives approximately $7,071M. Net debt of $19,916M against that is 2.82x. No greater precision is claimed for that ratio than the method supports, and the vendor's 3.44x is not wrong so much as struck on a different and less useful denominator.

Tangible equity is deeply negative and this is a genuine risk marker rather than an accounting curiosity. Goodwill of $20,229M plus other intangible assets of $12,611M is $32,840M against total equity of $20,299M — tangible equity of approximately negative $12.5B, which the vendor correctly reports as tangibleBookValuePerShareTTM of −$32.28. intangiblesToTotalAssetsTTM of 58.5% is the same fact from another angle. For a company that has just added $11.1B of acquisition spend to a business with $30B of revenue, the question of whether those intangibles earn their carrying value is not rhetorical, and the 10-Q's warning that the Boyd allocation could change materially is the honest statement of how provisional the number is.

5. Cash flow and capital returns

Capex is only 3.3% of revenue (capexToRevenueTTM 1.49% on a trailing basis, which looks low because the trailing revenue base now includes Boyd), and capexToDepreciationTTM of 0.377 says Eaton is investing well below its depreciation charge — an asset-light growth profile that is being achieved by buying capacity rather than building it. That is the whole strategic choice: $11.1B of acquisitions in six months against $919M of annual capital expenditure.

The dividend. The 10-K: "On February 26, 2026, Eaton's Board of Directors declared a quarterly dividend of $1.10 per ordinary share, a 6% increase over the dividend paid in the fourth quarter of 2025." Forward run-rate $4.40, a 0.99% yield. The vendor's dividendPerShareTTM of $4.28 is the trailing-twelve-month total, not the declared rate; profile.lastDiv is null. Cash dividends paid were $1,626M in 2025, $1,500M in 2024 and $1,379M in 2023.

And the buyback problem, which is not small. Eaton repurchased 4.2 million shares for $1,306M in the first half of 2026 (2.3M for $698M in the first quarter, 1.9M for $608M in the second). Ordinary shares outstanding went from 387.9 million to 388.4 million over the same period — UP 0.5 million. A $1.3 billion buyback produced no net share reduction at all, implying roughly 4.7 million shares issued in six months. The file does not disclose the reason and we do not guess at one; what we do is decline to count the gross buyback as shareholder yield. Total shareholder yield on the evidence is the 0.99% dividend and nothing more, and that is a materially different figure from the 1.75% a gross-buyback calculation would produce.

6. Valuation — priced in or room?

At $444.72, on 388.4M ordinary shares (market cap $172.73B) and a corrected enterprise value of $192.65B:

Estimate coverage is good on 2026 and 2027 (13-17 analysts) and thin beyond — the 2028 EPS row rests on FOUR analysts and 2029 and 2030 on TWO each. Only 2026 and 2027 drive conclusions here; 2028 appears once, as a cross-check, and is labelled.

est.ebitdaAvg, est.ebitAvg and est.sgaExpenseAvg are fabricated, and this file additionally trips a second, worse signature. From 2026 through 2030, ebitdaAvg is exactly 20.985% of revenueAvg and ebitAvg is exactly 17.039% — identical to five significant figures in five consecutive years. And in three of those years netIncomeAvg EXCEEDS ebitAvg, which is arithmetically impossible for a taxpaying company with $800M of annual interest expense: 2028 net income $7,269M against EBIT $6,699M; 2029 $8,659M against $7,489M; 2030 $9,430M against $7,500M. Both rows are rejected in full and all forward valuation runs on epsAvg and revenueAvg.

Peer context. The vendor peer set is a large-cap industrial basket: Deere ($166.6B, also in this batch), Parker Hannifin ($125.1B), Howmet ($115.4B), Emerson ($89.0B), Cummins ($87.8B), Illinois Tool Works ($84.9B), Honeywell ($78.9B), Boeing ($187.4B), Lockheed Martin ($136.0B), Union Pacific ($176.1B). Only Emerson is a genuine electrical-equipment comparable and no peer multiples are supplied in the file, so no peer-multiple comparison is drawn. The knowledge base supplies the one useful external data point: macrovoices (2026-01-22, conviction 82) puts Siemens' transformer backorder at roughly EUR138 billion on four-to-five-year lead times, which is category-level corroboration of Eaton's own book-to-bill from an unrelated source.

6a. What today's price assumes (the inversion)

At $444.72 — 33.0x the 2026 consensus and 27.9x 2027 — the price embeds the following falsifiable claims, each with a number and a date:

• Adjusted EPS reaches $13.459 in 2026. (Consensus-derived; 13 analysts.) The first half delivered $5.96, so the second half must deliver $7.50 — $3.75 a quarter against the $3.15 just reported, a 19% sequential step. The earnings calendar's third-quarter estimate of $3.52 implies a fourth quarter of roughly $3.98. That is the single most demanding number in the price and it is testable on 2026-11-03 and again in February.
• Adjusted EPS reaches $15.926 in 2027, on revenue of $36,088M. (Consensus-derived; 14 and 17 analysts.) That revenue figure implies organic growth decelerating from the 14% just delivered to roughly 6-8%, once Boyd's annualisation of about $1.3B of incremental 2027 revenue is stripped out. Against a backlog that grew 33% organically and orders that grew 41%, we think this estimate is stale, and it is where our variant perception lives (6c).
• Gross margin stops falling. (Filing-confirmed base; our assumption on the direction.) 37.0% to 33.5% year on year, with 470 basis points of commodity inflation in both Electrical segments. Consensus net margin on the 2027 figures implies roughly 17% — above the 15.5% the trailing period actually delivered.
• Net debt of $19.9B is temporary. (Our number, from the filed balance sheet.) The price is not discounting a leveraged company, and free cash flow of roughly $3.5B plus the $1.1B Dana distribution is the deleveraging path. A downgrade or a covenant discussion is not in the price at all.
• The market keeps paying roughly 28x the forward year. (Our number.) This is the most fragile assumption in the price. At 22x the 2027 consensus the stock is $350; at 34x it is $541. Nothing in the business changes between those numbers, and Eaton traded at roughly 20x forward earnings as recently as 2021 on a materially lower growth rate.

6b. The return bridge (why the multiple moves)

Expected return over the next twelve months decomposes as: adjusted EPS growth (+18.3%, from 2026E $13.459 to 2027E $15.926) + multiple drift (COMPRESSION, from 27.9x on the forward year today to roughly 26.6x on the forward year in twelve months, −4.7%) + shareholder yield (+1.0%) ≈ +13% to +15%.

Our base case assumes mild multiple compression and it is a judgement, not a rolldown artefact. Today the market pays 27.9x on the next calendar year. Our $495 is simultaneously 31.1x the 2027 consensus and 26.6x the 2028 consensus — so in twelve months, when 2028 is the forward year, the base implies a 26.6x multiple against today's 27.9x. That is a genuine 4.7% de-rating. We assume it because the earnings growth rate itself decelerates on the consensus path (+18.3% in 2027, +16.9% in 2028) and because the Mobility separation removes 11.4% of the revenue base at some point in the window, which mechanically reduces the size of the company being multiplied.

Almost the entire base-case return is earnings growth. If the multiple simply HELD at 27.9x on the 2028 consensus, the price would be $520 (+16.9%); if it compressed to 22x, $410 (−7.8%); if it expanded to 32x, $596 (+34.0%). The thesis does not need expansion, and we would flag it as a fragile leg if it did.

6c. Variant perception (where we differ, what would surprise)

• We differ from consensus on 2027 revenue, and this is the whole reason for a Buy rather than a Watch. The street models $36,088M for 2027, +10.9%. Backlog grew 33% organically in Electrical Americas and 54% organically in Electrical Global; book-to-bill was 1.3 and 1.1; organic customer orders grew 41% and 33%. A company shipping at a 1.3 book-to-bill in its largest segment does not decelerate to single-digit organic growth in the following year unless orders roll over first. Boyd alone adds roughly $1.3B of incremental 2027 revenue on annualisation. Watchable number: the 2026-11-03 print, where consensus revenue of $8,408.4M sits BELOW the $8,531M just delivered. That estimate has not been updated for the June quarter and is the clearest single piece of evidence that the forward numbers are stale.
• We differ from every screen on the balance sheet, and in the unfavourable direction. Any screen reading bal_a sees $10,547M of net debt; the filed figure at 2026-06-30 is $19,916M. That is a 89% understatement, and it is the reason our risk score is a 6 rather than a 5. We report it because it cuts against the position.
• We differ on how to read the a16z claim, and it is the sharpest long-run bear point in the lane. a16z (2026-04-14, conviction 75) describes power transformers as "an unglamorous bottleneck unchanged since inception" and discloses that the firm invested in a transformer-manufacturing company to relieve it. Most readers will take that as bullish confirmation of scarcity. We read it as the supply response beginning. A four-to-five-year lead time is the strongest possible signal to build competing capacity, and venture capital is now funding exactly that. Watchable: whether Electrical Americas book-to-bill falls below 1.0, which would be the first evidence that supply is catching demand.
• Positive surprise that would force a re-rate: third-quarter revenue materially above $8,408M with Electrical Americas book-to-bill still above 1.2, plus any commentary that commodity inflation is anniversarying out of the margin bridge. Together those validate volume and margin at once and put 2027 consensus visibly too low.
• Negative surprise that would break the thesis: organic orders decelerating below 15%; gross margin below 33%; a material upward revision to the Boyd purchase-price allocation that raises the amortisation run-rate above $255M a quarter; a delay or renegotiation of the Dana transaction; or a fourth quarter that fails to deliver the roughly $3.98 of adjusted EPS the full-year consensus of $13.459 requires.

Synthos fair values

All three anchors are multiples of the 2027 consensus adjusted EPS distribution (mean $15.926, low $14.541, high $16.553, 14 analysts), which is the last forward year in this file with adequate coverage.

• Bear ~$330 — 20.7x the 2027 consensus mean, cross-checked at 22.7x the 2027 consensus LOW of $14.541. Roughly 4.5% above the 52-week low of $315.82. The scenario: data-centre order intake digests, commodity inflation proves structural rather than cyclical, the Boyd allocation forces higher amortisation, and a 28x multiple on a levered industrial becomes a 21x multiple. −25.8%.
• Base ~$495 — 31.1x the 2027 consensus MEAN of $15.926, equivalently 26.6x the 2028 consensus mean of $18.620. Cross-check: corrected enterprise value of $212.5B on 2027 consensus revenue of $36.088B is 5.89x sales. Sensitivity, stated openly: 28x the 2027 consensus gives $446 — essentially spot — and 34x gives $541. The answer lives in a 28-34x band on 2027, or equivalently a 24-29x band on 2028. The scenario: the backlog converts, margin recovers as input costs anniversary, Dana closes, and the multiple drifts modestly lower as the growth rate normalises. +11.3%, plus about 1.0% of dividend.
• Bull ~$600 — 36.2x the 2027 consensus HIGH of $16.553. Cross-check: 32.2x the 2028 consensus mean. The scenario: 2027 revenue comes in at $38-39B rather than $36.1B because the order book converts faster than modelled, commodity costs reverse, and the post-Dana Eaton is re-rated as a pure electrical-infrastructure asset. This bull case requires multiple expansion AND an estimate revision, and both legs are named. +34.9%.

Base is 11.3% above spot; asymmetry approximately 1.35:1 to the upside (25.8% down, 34.9% up). That ratio, plus a variant perception with a testable date on it and a director buying in the open market, is enough for Tactical. It is not enough for Core with net debt three months old and the stock at its high.

7. Knowledge base — zero entity claims, two collisions discarded, one theme lane

Raw hits: 0 entity matches. Text-sweep hits: approximately 40, of which the large majority are discarded as homograph collisions. Case-sensitive bare-ticker hits: 3, ALL discarded. Used as theme-level support: 7, from 5 channels. Used as name-level conviction: 0.

Eaton Corporation is not named as an entity in any of the 51,928 distilled claims. The entity sweep on ETN, Eaton, Eaton Corporation and Eaton Corp returned zero. That is the primary finding and everything below is secondary.

Collision class 1 — the bare ticker, exactly as warned

A case-sensitive, word-boundary sweep on the string ETN in claim text returned three hits. All three use ETN to mean exchange-traded note:

> 2025-12-30 · compound_and_friends · conviction 60 — "When the MicroSectors FANG -3x ETN (FNGD) breaks above its 50-day MA, it signals leverage leaving megacaps — something's wrong."

> 2025-12-30 · compound_and_friends · conviction 50 — "FNGD (inverse-3x FANG ETN) breaking above its 50-day MA signals leverage leaving mega-caps and market trouble."

> 2023-03-24 · odd_lots · conviction 72 — "Short-vol blowups are 'neon swans'... VIX-curve inversion made the 2018 ETN collapse foreseeable."

Not one relates to Eaton. The sweep is discarded in full.

Collision class 2 — the AI-Transformer homograph

The obvious text search for an electrical-equipment company is transformer. In this knowledge base that word overwhelmingly denotes the neural-network architecture. Of the text matches returned, the majority are claims such as "No obvious replacement for attention/transformers", "Transformers are more efficient than the human brain in some respects", "Diffusion and Transformers are converging" and "Self-driving cars (same transformer architecture) can't generalize." Every claim in which "transformer" denotes the machine-learning architecture is discarded in full. This is the UNP-railroads pattern in a different vocabulary, and it is severe enough that a naive keyword sweep on this company would return a knowledge base that appears rich and is entirely irrelevant.

What survives — 7 theme-level claims from 5 channels

> 2026-01-22 · bullish · conviction 82 · skill 0.9 · macrovoices · entities: Siemens

> "Siemens has ~€138bn transformer backorder with 4-5 year lead times, ~100% dependent on Chinese rare earths — a hard bottleneck on grid/nuclear buildout."

This is the most useful claim in the lane and the only one with a hard number in it. It is about a competitor, it quantifies category-wide backlog and lead times, and it independently corroborates Eaton's own 1.3 book-to-bill from a source with no relationship to Eaton. It also names a supply-chain dependency — Chinese rare earths — that Eaton's own filings do not quantify.

> 2026-04-14 · bullish · conviction 75 · a16z · no named speaker

> "Power transformers are an unglamorous bottleneck unchanged since inception; a16z invested in a transformer-manufacturing company to relieve it."

Two-sided, and we read the second half as the more important one. New venture-funded manufacturing capacity is the supply response to the scarcity Eaton monetises.

> 2026-02-26 · neutral · conviction 70 · no_priors · no named speaker

> "Short-term (6-12mo) bottlenecks are mundane: structural steel, electricians, transformers, chillers needed to build a powered shell."

Note the word "chillers" — that is the Boyd Thermal category, named as a data-centre bottleneck four months after Eaton bought a $9.55 billion position in it.

> 2026-01-22 · bullish · conviction 85 · skill 0.9 · macrovoices — "Copper's physical intensity in AI data centers, grid cables and electrification is enormous; Chinese negative-cost smelting bankrupts Western refiners, tightening supply."

> 2024-11-11 · neutral · conviction 65 · luke_gromen — "US no longer makes grid transformers (Germans/Japanese/French/Chinese do); China retains leverage over critical US infrastructure supply chains we've barely started reshoring."

> 2022-11-05 · neutral · conviction 76 · skill 1.1 · lyn_alden — "Developed-world grid/infrastructure is chronically underinvested and slow to build; full electrification concentrates all energy into one channel that must be extremely robust."

> 2024-05-01 · bullish · conviction 65 · macrovoices — "The US grid is far behind China and must be rebuilt/hardened (e.g., Texas); electrification is fundamental to running the planet."

Concentration sensitivity test. macrovoices supplies 3 of the 7 used claims — 42.9%, above the one-third threshold. Stripping it leaves 4 claims from 4 independent channels (a16z, no_priors, luke_gromen, lyn_alden), all pointing the same direction. The lane survives its own concentration, and the surviving claims include both the most recent (2026-04-14) and the highest-skill (lyn_alden, 1.1).

Note also that the copper claim is DOUBLE-EDGED for Eaton and we count it that way. Copper scarcity supports the demand thesis and is simultaneously the 470-basis-point commodity headwind in Eaton's own margin bridges. A claim that is bullish for the theme is bearish for this company's gross margin, and we do not get to use only the half we like.

Conclusion, stated as a finding. The Synthos knowledge base has no view on Eaton Corporation. It has a well-corroborated, multi-channel, recent structural view that grid and power-equipment supply is a hard bottleneck — which is the demand backdrop for Eaton's order book, is independently confirmed by Eaton's own disclosures, and would apply equally to Schneider, ABB, Siemens, Vertiv or Hubbell. Conviction rating: Low. Breadth: 5. No conviction premium is applied to the fair value.

8. Data integrity — what we rejected from the vendor file and why

Ten findings. ETN is a file with two excellent segment tables and a balance sheet that is six months and $10.8 billion out of date.

1. bal_a predates the largest event in the company's recent history — REJECTED and rebuilt from the 10-Q. The most recent balance sheet in the payload is 2025-12-31, three months before Boyd Thermal closed. It reports totalDebt of $11,169M and netDebt of $10,547M. The 2026-06-30 10-Q reports short-term debt $2,091M, current long-term debt $11M and long-term debt $18,509M — $20,611M — against cash of $483M and short-term investments of $212M, i.e. net debt of $19,916M. Every balance-sheet figure in this dive comes from the 10-Q. Total assets went from $41,251M to $56,181M in six months; goodwill from $15,769M to $20,229M; other intangibles from $5,054M to $12,611M.

2. The file disagrees with itself by roughly $9.4 billion. km_ttm.enterpriseValueTTM of $193,528M less the market capitalisation of $172,685M implies net debt of $20,843M — within 4.7% of the filed $19,916M and 97% above the $10,547M in bal_a. The derived block reflects the Boyd acquisition; the raw block does not. This is the ABT pattern precisely, and it is the reason the data contract requires an 8-K scan on every name.

3. An entire transaction is absent from the payload — the Dana Reverse Morris Trust. The 8-K of 2026-06-11 discloses definitive agreements to separate the Mobility segment (Vehicle $2,505M + eMobility $618M = $3,123M, 11.4% of 2025 revenue) and merge it with Dana Incorporated, with Eaton receiving approximately $1.1 billion in cash and Eaton shareholders receiving at least 50.1% of the combined company. seg_prod still shows five segments; est still projects 2027 and 2028 revenue on the pre-separation base; nothing in quote, profile, km_ttm or ratios_ttm reflects it. Note that the financial statements are also correctly still on the pre-separation basis — Mobility was NOT classified as held for sale at 2026-06-30 — so this is a disclosure gap in the vendor file rather than an accounting one.

4. The segment structure in seg_prod is superseded — corrected in presentation. The vendor reports five segments for 2025; the 10-Q reports four, with Vehicle and eMobility combined into Mobility. We use the four-segment structure for all 2026 figures and the five-segment structure for 2025 and earlier, and label which is which.

5. est.ebitdaAvg and est.ebitAvg are fabricated AND arithmetically impossible — REJECTED in full. From 2026 through 2030, ebitdaAvg is exactly 20.985% of revenueAvg and ebitAvg exactly 17.039%, identical to five significant figures across five years (sgaExpenseAvg is likewise fixed). Worse, netIncomeAvg exceeds ebitAvg in 2028 ($7,269M vs $6,699M), 2029 ($8,659M vs $7,489M) and 2030 ($9,430M vs $7,500M) — impossible for a taxpaying company carrying roughly $800M of annualised net interest. All forward valuation runs on epsAvg and revenueAvg.

6. epsdiluted is NULL on every row and eps is the basic figure. The vendor's 2025 eps of 10.49 on 389.5M shares is basic; the 10-Q's filed diluted EPS for the first half of 2026 is $4.33 against the vendor's implied 2.23 + 2.11 = $4.34, and for the first half of 2025 $4.96 against 2.46 + 2.51 = $4.97. The differences are rounding, but the field is systematically absent and every diluted figure here is filing-sourced.

7. GAAP earnings contain large, disclosed, non-recurring acquisition effects — flagged rather than smoothed. Second-quarter GAAP diluted EPS of $2.11 contains $0.49 of acquisition and divestiture charges, $0.50 of intangible amortisation and $0.05 of restructuring — $1.04, or 33% of the $3.15 adjusted figure, against $0.44 (15%) a year earlier. We use adjusted EPS for forward valuation because estimates are struck on it, and we state plainly that adding back $255M a quarter of amortisation on assets bought for $9.55B three months ago flatters the earnings power of an acquisition-built business.

8. netDebtToEBITDATTM of 3.44x is struck on a charge-depressed EBITDA — recomputed. The vendor's trailing GAAP EBITDA of $6,058M includes a June-quarter figure ($1,345M) that is lower than the March quarter's ($1,485M) despite 14.5% sequential revenue growth, because acquisition charges land in it. On our constructed trailing adjusted EBITDA of approximately $7,071M — method disclosed in Section 4 — net debt of $19,916M is roughly 2.82x.

9. lastDividend is absent and dividendPerShareTTM of $4.28 is the trailing total — corrected from the 10-K. "On February 26, 2026, Eaton's Board of Directors declared a quarterly dividend of $1.10 per ordinary share, a 6% increase over the dividend paid in the fourth quarter of 2025." Forward run-rate $4.40, a 0.99% yield, against the vendor's stated 0.96%.

10. quote.yearHigh of $451.955 and tech.hi52 of $444.72 differ — and the difference is instructive rather than a defect. quote.yearHigh equals today's intraday high; tech.hi52 equals today's close, and tech.pct_from_hi and tech.max_dd_from_peak both read exactly 0.0. Read together they say the stock set a new intraday and closing 52-week high on 2026-08-04. We use tech throughout, per the standing rule, and note that a 0.0% drawdown from peak is a genuine reading here rather than a missing value.

Not defects — and three of them are verifications worth recording:

A. seg_prod and seg_geo both reconcile. The 2025 geographic lines (US $17,122M, Europe $5,078M, Asia-Pacific $2,704M, Latin America $1,461M, Canada $1,083M) sum exactly to reported revenue of $27,448M, and the 2024 lines sum exactly to $24,878M. The 2025 product lines sum to $27,463M — a $15M (0.05%) difference consistent with intersegment elimination. Both tables are used.

B. The share count reconciles. The 10-Q states 388.4 million ordinary shares outstanding at 2026-06-30; the vendor's market capitalisation of $172,684,776,000 divided by $444.72 implies 388.3 million. A 0.03% difference. No share-class or partial-unit error.

C. enterpriseValueTTM is current, not stale. At $193,528M it is within 0.46% of our rebuild, the difference being the $715M of operating lease liabilities the vendor's debt convention includes. The enterprise value is right even though the balance sheet feeding it is not — which is precisely why the internal contradiction in finding 2 is worth stating.

Also correctly reported: revenuePerShareTTM of $77.284 × 388.4M gives $30,017M against our filing-derived trailing revenue of $30,026M (0.03%); netIncomePerShareTTM of $9.858 matches the filing-derived trailing GAAP diluted figure exactly; interestCoverageRatioTTM of 11.95x is consistent with the trailing interest and income figures; effectiveTaxRateTTM of 20.6% matches the 10-Q's discussion; and tangibleBookValuePerShareTTM of −$32.28 is arithmetically correct and economically informative.

Non-equity tripwire — checked and passed. ETN is ordinary shares of an Irish public limited company, NYSE-listed, 388.4 million outstanding. Beta is 1.181; the dividend is variable and was raised 6% in February 2026; the 52-week band of $315.82 to $444.72 is a 40.8% range; volume was 2.49 million shares (roughly $1.11B of turnover). This is common equity, not an ADR (isAdr false) and not a fund.

9. Technicals

• Price $444.72 — the highest close of the trailing 252 sessions. pct_from_hi 0.0, max_dd_from_peak 0.0. +40.8% above the 52-week low of $315.82. Position within the annual range: 100th percentile.
• Both moving averages are well below and rising through: +9.6% above the 50-day at $405.87, +18.6% above the 200-day at $374.92, with the 50-day comfortably above the 200-day.
• RSI 59.3 — mid-range, notably NOT overbought for a stock at its high, which suggests the advance has been steady rather than parabolic. MACD +3.38 — positive.
• Relative performance, and the shape tells you when the story changed: 3-month +5.3% vs SPY +7.6% — slightly behind; 6-month +23.7% vs SPY +11.1% and QQQ +15.7% — a 12.6-point outperformance; 12-month +16.6% vs SPY +24.3% and QQQ +30.8% — 7.6 points BEHIND. Read together: Eaton underperformed for the first half of the trailing year and has outperformed sharply for the second half. The re-rating is roughly six months old, which lines up with the Boyd announcement and the acceleration in the order book.
• Sentiment: 26 buy, 13 hold, 0 sell across 39 analysts. Consensus $496.20 (+11.6%), median $500, high $512 (+15.1%), low $480 — 7.9% ABOVE spot. The full target range spans only 6.7%, from $480 to $512 — the narrowest distribution in this batch. We flag that as a caution rather than a comfort: a consensus that tight usually means most targets were set to the same framework at the same time, and it leaves no dispersion to trade against.

Today's move

ETN closed 2026-08-04 at $444.72, up $6.49 or 1.48% from $438.23. It OPENED at $451.29 — above the eventual close — made the day's high of $451.955 and the day's low of $434.30, and finished 1.5% below the open on 2,494,507 shares (roughly $1.11B of turnover). A 52-week-high close that fades 1.5% from the open is not a confirmation, and we say so rather than presenting the up-day as an endorsement. No company-specific news appears in this file for 2026-08-04; the last event was the 2026-07-31 earnings release and 10-Q, four days earlier.

10. Insiders — a genuine open-market purchase, which is rare

Eight transactions across May 2026:

The reading. Most of this table is routine: Gregory Page's four entries are a single annual restricted-stock-unit vest with tax withholding, and Dorothy Thompson's 167-share sale is trivial. Antonio Galvao's 494-share sale at $405.86 is small.

What is not routine is Gerald Johnson. Two separate open-market PURCHASES — 746 shares at $402.29 on 2026-05-08 and 215 shares at $419.02 on 2026-05-11, together 961 shares for approximately $390,000 — which took his direct holding from 668 to 1,629, an increase of 144%. P-Purchase transactions are the only insider code that carries information, because they are discretionary uses of the insider's own money, and they are absent from every other insider table in this batch.

How much weight to put on it: some, but bounded. $390,000 is a meaningful personal commitment and a trivial corporate signal, and it was made at prices 5-10% below the current one, before the June quarter's order data was public. We treat it as a mild positive and an input to the Tactical rather than Watch decision — not as a thesis.

11. Verdict, kill-criteria and flip conditions

Buy — Tactical.

What is genuinely exceptional, and it is order data rather than narrative: total backlog of $24.1 billion at 2026-06-30 with 71% deliverable within twelve months; Electrical Americas organic customer orders +41% against organic sales +18%, on a 1.3 book-to-bill and backlog +33%; Electrical Global backlog +103% (organic +54%) with orders +33%; Aerospace book-to-bill 1.2; consolidated organic growth of 14% and total growth of 21%; adjusted EPS $3.15 against $2.95; segment operating margins of 27.5%, 19.8% and 22.8% in the three continuing businesses; Boyd Thermal annualising near $1.8B at a 23.1% segment margin after 110 days of ownership; and a director buying in the open market twice.

What we are underwriting rather than ignoring: net debt of $19,916M, up $10,824M in six months, at roughly 2.8x our constructed trailing adjusted EBITDA; goodwill and intangibles of $32,840M against total equity of $20,299M, leaving tangible equity around negative $12.5B; gross margin down 350 basis points with 470 basis points of commodity inflation in both Electrical segments; GAAP diluted EPS down 16%; a preliminary purchase-price allocation the filing says could move materially; a pending Reverse Morris Trust that removes 11.4% of revenue; a $1,307M buyback that produced zero net share reduction; 27.9x the 2027 consensus; and a close at the 52-week high that faded 1.5% from the open.

The distinction that matters. We are not paying for a story. We are paying for a contracted backlog of $24.1 billion, disclosed in a 10-Q filed four days ago, in a company whose orders are growing at more than twice its shipment rate — and our variant perception is narrow, specific and testable: consensus 2027 revenue of $36,088M implies organic deceleration from 14% to roughly 6-8% while the backlog grew 33% organically, and the third-quarter revenue estimate of $8,408.4M is BELOW the $8,531M the company just delivered. That estimate is stale, and stale estimates on a name at a 52-week high are exactly where a research edge lives. Tactical rather than Core because the leverage step is three months old, the margin bridge has a 470 basis point hole in it, and the corporate structure changes materially before this thesis matures.

Pre-registered KILL criteria — what would take this to Watch or Avoid:

• Electrical Americas book-to-bill below 1.0 in any quarter. That is the first evidence supply has caught demand, and it is the number the entire thesis rests on.
• Organic customer order growth below 15%, from 41%.
• Gross margin below 33.0%, which would mean the commodity headwind is widening rather than anniversarying.
• A material upward revision to the Boyd purchase-price allocation that pushes quarterly intangible amortisation above $255M, or any impairment discussion against $20,229M of goodwill.
• Delay, renegotiation or termination of the Dana Reverse Morris Trust, which would remove the $1.1B cash distribution and the deleveraging path with it.
• Fourth-quarter adjusted EPS materially below the roughly $3.98 the full-year consensus of $13.459 requires — the single most demanding number in the price.

Pre-registered FLIP conditions — what would take this to Buy — Core:

• A 2027 revenue consensus revision above $38B, which is what a 1.3 book-to-bill and 41% order growth imply and which would take our base case toward the bull anchor.
• Net debt below $16B with the Dana cash received, taking constructed leverage under 2.0x and removing the single largest reason this is not already Core.
• Two consecutive quarters of gross margin expansion, confirming the commodity drag is cyclical.
• A price below approximately $390 (roughly 24.5x the 2027 consensus, near the 50-day moving average) with the order book intact, which would put the base case at +27%.
• A name-level knowledge-base claim appearing. At present there are zero, and the entire lane is theme-level. That absence is worth roughly one tier on its own.

Where ETN fits in the Synthos Framework Portfolio. The electrification / data-centre-infrastructure sleeve, at 0% today with a 2-3% target initiated here. On batch overlap: Eaton is the only name in this batch with direct, disclosed, quantified exposure to data-centre electrical demand, and it is a different trade from the semiconductor names — Eaton sells to the building, not to the rack, and its revenue is contracted a year in advance rather than booked on a purchase order. The overlap risk is with the theme rather than within the batch: an AI capital-expenditure pause hits Eaton's order book with a lag of one to two quarters and its revenue with a lag of three to four, which is later than the semiconductor complex but not immune to it. Logged as a tracked Synthos call (Buy — Tactical) as of 2026-08-04 at $444.72.

Single biggest risk: price-cost. The volume question is answered — $24.1 billion of contracted backlog answers it. The margin question is not. Gross margin fell 350 basis points, both Electrical segments took a 470 basis point commodity-inflation hit, and the same copper thesis that supports the demand case is the reason the input cost is rising. Consensus needs adjusted EPS of roughly $3.98 in the fourth quarter against $3.15 just delivered, and the arithmetic that gets there requires either a sharp margin recovery or a volume step the order book supports but the shipping capacity may not. At 27.9x the 2027 consensus with net debt at $19.9 billion, there is no valuation cushion under a margin miss.

Provenance & disclosures

• Traceability: ZERO entity-level knowledge-base claims name Eaton across 51,928 distilled claims (entity sweep on ETN, Eaton, Eaton Corporation, Eaton Corp: 0 hits). Two collision classes identified and discarded with reasons. (i) Bare-ticker collision: a case-sensitive word-boundary sweep on ETN returned 3 hits, all using the string to mean exchange-traded note (compound_and_friends ×2 on the MicroSectors FANG -3x product FNGD, 2025-12-30; odd_lots on the 2018 XIV collapse, 2023-03-24) — discarded in full. (ii) Homograph collision: the text sweep on transformer is dominated by the AI neural-network architecture across ml_street_talk, no_priors, latent_space, dwarkesh, cognitive_revolution and others — discarded in full. 7 theme-level claims from 5 channels are used: macrovoices ×3 (2026-01-22 conviction 82 on Siemens' ~€138bn transformer backorder and 4-5 year lead times; 2026-01-22 conviction 85 on copper intensity; 2024-05-01 conviction 65 on US grid rebuild), a16z (2026-04-14, conviction 75), no_priors (2026-02-26, conviction 70), luke_gromen (2024-11-11, conviction 65) and lyn_alden (2022-11-05, conviction 76, skill 1.1). Concentration sensitivity test run and reported: macrovoices supplies 3 of 7 used claims (42.9%); stripping it leaves 4 claims from 4 independent channels pointing the same way, and the lane survives. None is a view on Eaton as a security; none carries a management speaker_role; six of seven carry no named speaker and are attributed only to a channel. The copper claim is counted as double-edged — bullish for the demand theme, bearish for Eaton's gross margin — rather than only in the favourable direction. All quotes are verbatim from the stored claim text.
• Data as-of: fundamentals — second-quarter 2026 income statement and balance sheet as at 2026-06-30, from the 10-Q filed 2026-07-31; annual figures from the 10-K filed 2026-02-26 (fiscal 2025) · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873600 = 2026-08-04T20:00:00Z ($444.72, +1.48%; 50-DMA $405.87; 200-DMA $374.92; RSI 59.3; MACD +3.38) · knowledge-base claims 2026-08-04. Fiscal year = calendar year. All figures come from the Synthos vendor data file for ETN or from the SEC filings in the ETN archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents: 10-K filed 2026-02-26 (fiscal 2025, 374,554 characters, financial-statement tables preserved); 10-Q filed 2026-05-05 (March 2026 quarter); 10-Q filed 2026-07-31 (June 2026 quarter, 149,202 characters — the primary source for this dive); 8-K filed 2026-05-05 (Item 2.02, first-quarter results, cover only, Exhibit 99.1 not extracted); 8-K filed 2026-06-11 (Items 7.01 and 8.01 — the Dana Reverse Morris Trust definitive agreements, 17,436 characters, present in substance); 8-K filed 2026-07-31 (Item 2.02, second-quarter results, cover only). The two earnings-release 8-Ks contain no financial data, but the corresponding 10-Qs are both present, so nothing is lost.
• Where the filings contradicted or corrected the vendor (detailed in Section 8): the balance sheet rebuilt entirely from the 10-Q — bal_a stops at 2025-12-31 and reports net debt of $10,547M where the 2026-06-30 filing supports $19,916M, an 89% understatement across the $9.55B Boyd Thermal acquisition that closed 2026-03-12; the vendor's own enterpriseValueTTM implies $20,843M of net debt, so the file disagrees with itself by roughly $9.4B; the Dana Reverse Morris Trust (8-K, 2026-06-11), which separates 11.4% of revenue, is entirely absent from the payload; the segment structure in seg_prod (five segments) is superseded by the 10-Q's four; est.ebitdaAvg/ebitAvg/sgaExpenseAvg rejected as fabricated on fixed revenue ratios of 20.985%/17.039% across five years, compounded by netIncomeAvg exceeding ebitAvg in three forward years; epsdiluted NULL on every row, with all diluted figures taken from the 10-Q; netDebtToEBITDATTM of 3.44x recomputed to approximately 2.8x on a constructed adjusted EBITDA, with the method disclosed; dividendPerShareTTM of $4.28 identified as the trailing total, with the 10-K giving a declared quarterly rate of $1.10 and a forward run-rate of $4.40 (0.99% yield); and quote.yearHigh reconciled against tech.hi52 as intraday versus closing high. Where vendor and filing AGREED — worth recording: the share count (388.4M filed against 388.3M implied, 0.03%); seg_geo, which sums EXACTLY to reported revenue in both 2025 and 2024; seg_prod, which sums to within 0.05%; enterpriseValueTTM, within 0.46% of our rebuild; revenuePerShareTTM and netIncomePerShareTTM, both within 0.03% of filing-derived trailing figures; and effectiveTaxRateTTM.
• Basis note: analyst estimates are struck on Eaton's adjusted earnings per ordinary share, which excludes acquisition and divestiture charges, restructuring charges and intangible asset amortisation. The GAAP-to-adjusted gap is unusually large and widening — $1.04 per share, 33% of the adjusted figure, in the June 2026 quarter against $0.44 (15%) a year earlier — and it is disclosed here rather than smoothed. Forward multiples use consensus adjusted EPS; the trailing multiple is shown on both bases (36.0x adjusted, 45.1x GAAP diluted). We state explicitly that adding back $255M a quarter of amortisation on assets purchased for $9.55B three months ago flatters the earnings power of an acquisition-built business, and that a reader who prefers the GAAP series is looking at a 45x stock.
• Estimate coverage: 9 analysts on 2026 revenue and 13 on EPS; 17 and 14 on 2027; 10 and FOUR on 2028; 5 and TWO on 2029; 5 and TWO on 2030. The fair value is struck on the 2027 distribution (14 analysts on EPS); the 2028 figure appears only as a cross-check and is labelled thin; 2029 and 2030 are excluded from every conclusion.
• Peer note: the vendor peer set is a large-cap industrial basket — Boeing ($187.4B), Union Pacific ($176.1B), Deere ($166.6B, also in this batch), Lockheed Martin ($136.0B), Parker Hannifin ($125.1B), Howmet ($115.4B), Emerson ($89.0B), Cummins ($87.8B), Illinois Tool Works ($84.9B), Honeywell ($78.9B). Only Emerson is a genuine electrical-equipment comparable, and no peer multiples are supplied in the file, so no peer-multiple comparison is drawn. The knowledge base's Siemens backorder claim is used instead as the one available external category data point, and is labelled as such.
• Fair-value caveat: the $330 / $495 / $600 anchors are multiples of the 2027 consensus adjusted EPS distribution — 20.7x the mean of $15.926 for the bear (cross-checked at 22.7x the low of $14.541), 31.1x the mean for the base, and 36.2x the high of $16.553 for the bull. Stated arithmetic, not a discounted cash flow. The base is sensitivity-disclosed: 28x the 2027 consensus gives $446, essentially spot, and 34x gives $541. The base assumes mild multiple compression measured on a like-for-like forward-year basis (27.9x today against 26.6x on the 2028 estimate in twelve months) and explains why; the bull case requires BOTH expansion and an upward estimate revision, and both legs are named.
• Timing: second-quarter 2026 results were released 2026-07-31 and the 10-Q filed the same day, four days before this dive, so the entire first half is filing-verified. The next print is 2026-11-03, 91 days away — no near-term company-specific catalyst, though the Dana transaction's registration and proxy filings are an unscheduled event risk inside that window. 2026-08-04 saw the shares set a new 52-week closing high while giving back 1.5% from the open, which is noted as a caution rather than a confirmation.
• Accessibility note: no information in this dive is conveyed by colour. All emphasis is carried by bold text, table structure and explicit labelling.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.