Customer Analysis — Eaton Corporation (ETN)

§ 01Executive View

Eaton's customer base is the most structurally diversified in the deep-dive cohort, spanning hyperscalers, electrical contractors, industrial OEMs, utilities, defense primes, and EV OEMs across five business segments and six end-market verticals within Electrical. That diversification is simultaneously the thesis and the question: data-center revenue is growing at extraordinary rates — Q4 2025 orders roughly +3x YoY — but the data-center end-market represents an estimated 30–35% of Electrical Americas revenue and approximately 20–25% of total Eaton group revenue, meaning roughly 75–80% of the company is exposed to more prosaic industrial, utility, commercial, and aerospace cycles. The cohort's Open Question §4 — does the diversified industrial mix dilute the AI-DC re-rate? — gets a nuanced answer here: the mix is a structural strength (prevents a single-customer cliff) but a narrative headwind (the 35% of revenue in AI-DC gets absorbed into a 100% EV that looks like 3–4x less leverage than Vertiv). The single most important customer fact is the hyperscaler-directed nature of ETN's data-center order inflection: AWS, Azure, Google, Meta, and Oracle are either direct-specification customers or final end-users routed via large electrical-contractor partners (EMCOR, Comfort Systems, Parsons, Mortenson) — and Eaton's switchgear, UPS, and busway earn a specification moat that is sticky once the engineer-of-record locks the design. The main demand-quality concern is the calendar-mismatch between Q4 2025 order print (+3x YoY) and revenue recognition: Eaton's large-project electrical infrastructure carries 12–24 month conversion lead times, meaning the order surge converts to revenue primarily in 2026–2027 rather than immediately, and Schneider's "2028–2030 for 800V real impact" framing is an honest calibration of how back-end loaded the 800V-architecture-dependent revenue tail actually is.

§ 02Customer Concentration

Interpretation: ETN's customer concentration profile is a structural moat at the named-customer level. No single utility, hyperscaler, or industrial OEM controls an outsized share of the revenue base. However, the customer type that matters for the bull thesis — hyperscaler data-center operators — is concentrated in a handful of names (the Big 5 hyperscalers + CoreWeave/Oracle as emerging neoclouds), and Eaton sells into them primarily via the electrical-contractor channel rather than direct-to-end-user. This is different from Vertiv, which increasingly sells direct to hyperscaler procurement. The contractor mediation layer adds one degree of visibility buffering: Eaton sees contractor PO flow rather than hyperscaler demand schedules directly, which creates a slight lag in demand-quality signal.

Direction of concentration: the hyperscaler sub-segment is the fastest-growing portion of Electrical Americas, and as it grows it is simultaneously the most concentrated end-user set (five hyperscalers represent the majority of AI-DC investment) and the least concentrated direct-customer set (because the buy is routed through dozens of electrical contractors). This is a nuanced distinction. The practical implication: Eaton does not face a "one hyperscaler leaves" cliff at the revenue level, but does face a "if hyperscalers slow their build rate collectively" headwind because the end-market is concentrated in the aggregate.

§ 03End-Market Exposure

Eaton reports five reportable segments. Within the two Electrical segments, management breaks out six sub-end-markets. The table below maps the full revenue picture at both levels.

Segment-level revenue mix (FY25, approximate)

Source: ETN FY25 10-K segment disclosures; Q4 2025 earnings call; ETN 2024 Investor Day presentation.

End-market sub-mix within Electrical Americas (~52% of total)

Note on data-center % of total Eaton revenue: Management has been explicit that data-center is the fastest-growing vertical within Electrical Americas but has not always disclosed the precise percentage of total Eaton group revenue. The most direct disclosures (Q3 2025 earnings call; Q4 2025 earnings call; ETN 2024 Investor Day) suggest data center was approximately 28–32% of Electrical Americas in 2024, accelerating toward 33–38% in 2025 on the order surge. Electrical Americas is approximately 52–54% of total Eaton revenue. Implied math: data center is approximately 15–20% of total Eaton group revenue in FY25, trending toward 20–25% in FY26–FY27 as the Q4 2025 order backlog converts to revenue. This is the TXN-benchmark: TXN AI-DC at ~9% of revenue in FY25 is "material." By this same lens, Eaton's AI-DC exposure is already materially larger in absolute terms and as a percentage of total revenue — the cohort materiality test passes clearly.

Electrical Global sub-mix (approximately)

Weighted demand picture: Eaton is best understood as a slow-moving portfolio with one fast-moving engine. Data center (roughly 16–19% of total group revenue in FY25, growing toward 20–25%) is the fast engine. The remaining ~75–80% — utility T&D, commercial construction, industrial OEM, aerospace, vehicle — are slower-cycle exposures that smooth the revenue base but dilute the AI-DC leverage ratio. The critical implication for the cohort: Eaton's AI-DC revenue mix is materially higher than TXN's (~9%), meaningfully lower than Vertiv's (~80%+), and sitting approximately where the cohort thesis inflects from "relevant exposure" to "primary driver." This is not a verdict of "too diluted to matter" nor "pure-play." It is a genuine mixed story — a Vertiv on a slower horse.

§ 04Data Center Exposure — Quantified

This is the binding cohort question. Evidence assembled from filings, transcripts, and peer comparisons:

The AI-DC materiality test (TXN benchmark): TXN customer analyst set the benchmark at $1.5B / 9% of revenue / +90% YoY Q1'26. ETN's equivalent in FY25 is approximately $3.5–4.5B / 16–19% of total revenue / estimated +30–50% YoY growth in data-center-attributable revenue. The answer to "is this material?" is an unambiguous yes — materially larger than TXN in both dollar terms and mix percentage. The follow-on question — "is this a pure-play AI-DC story?" — answers no: at 16–19% mix, data center is the biggest single sub-vertical but is still less than one-fifth of the enterprise. The Eaton story is structurally different from Vertiv (where AI-DC is approximately 80%+ of revenue): Eaton is a large-cap diversified industrial with a very large and fast-growing AI-DC sub-segment, not an AI-DC pure-play in diversified clothing.

Q4 2025 orders-to-revenue conversion timeline:

• Q4 2025 ~3x YoY order print implies a step-change in the order backlog entering 2026
• Eaton's large-project electrical segment (switchgear, UPS 9395, busway) carries 12–24 month delivery timelines; smaller standardized products carry 4–8 week lead times
• Best estimate: approximately 30–40% of the Q4 2025 order surge converts to revenue in H1 2026; the remainder converts in H2 2026 – H2 2027
• This is consistent with Schneider's "2028–2030 for 800V real impact" calendar framing, which applies specifically to 800V-architecture-dependent revenue (a sub-set of total DC revenue); conventional 48V/AC-architecture datacenter electrical work is converting faster

§ 05Hyperscaler and Utility Customer Table

Key End-User Accounts (data center)

Key Utility / T&D Customers (Cooper-era relationships)

Cooper-era integration assessment: The 2012 Cooper acquisition ($11.8B) was the single largest transformation of Eaton's customer base in its modern history. Cooper Bussmann (circuit protection), Cooper Crouse-Hinds (hazardous-location electrical), and Cooper B-Line (cable management) brought utility/T&D and industrial/OEM relationships that Eaton's legacy power distribution lacked at scale. As of 2026, these relationships are fully integrated (10+ years of Cooper integration is complete) and are load-bearing for the AI-DC thesis in a non-obvious way: the AI-DC build is straining utility interconnect queues, which is forcing investor-owned utilities to accelerate T&D upgrade programs — which feeds directly into Eaton's utility customer base. The utility grid buildout for AI-DC load is a downstream demand driver for the Cooper-era T&D customer relationships. This is not a Cooper-specific thesis anymore; it is a structural reinforcement of Eaton's T&D franchise by the same AI-DC capex wave that is also driving direct switchgear/UPS orders.

Aerospace Customer Concentration

Aerospace is approximately 14–15% of total Eaton revenue and serves as a non-correlated diversification buffer. Defense subcomponent of Aerospace is growing as defense budgets increase; commercial OEM (Boeing, Airbus) is recovering from post-COVID supply-chain normalization. Aerospace does not contribute to the AI-DC thesis but insulates the enterprise from an AI-DC demand-cycle downturn — this is the structural diversification benefit.

Vehicle / eMobility

Both are sub-10% of total revenue and directionally weaker in NTM. Vehicle is recovering from the heavy-truck destocking cycle. eMobility is structurally positive (every EV has more Eaton electrical content than an ICE vehicle) but on a slower commercial ramp than the market hoped. Neither segment contributes materially to the AI-DC thesis. eMobility represents optionality, not a near-term catalyst.

§ 06Contract Structure & Switching

Long-term agreements: Eaton does not publish a formal RPO or backlog figure analogous to Vertiv's $15B disclosed backlog. However:

• Large-project switchgear and UPS contracts for hyperscaler campus builds (50–200 MW+ campuses) carry multi-year procurement frameworks. Hyperscalers typically run 2–4 year "approved vendor" programs with preferred pricing tiers; Eaton is typically on the approved vendor list for the full scope of medium-voltage switchgear, ATS, and large-format UPS. These are framework agreements, not binding take-or-pay.
• Utility T&D relationships are multi-year supply agreements for equipment categories (circuit breakers, fuses, disconnects, panelboards); often 2–3 year contract cycles tied to utility IRP/capital plans.
• Aerospace LTAs are the strongest formal contract structure in the portfolio. Boeing and Airbus LTAs for electrical/hydraulic/fuel systems content carry 5–15 year life-of-program terms; defense contracts are cost-plus or IDIQ structures with long periods of performance.
• Cooper-era distribution: Eaton's electrical distribution hardware (fuses, circuit breakers, cable management) sold via Wesco, Sonepar, Rexel, and Graybar is largely spot catalog pricing — no formal LTAs.

Switching costs:

Qualification cycle for data-center products: Eaton's 9395 UPS and Power Xpert switchgear typically undergo 6–12 month approval processes within a hyperscaler's electrical infrastructure standard (each hyperscaler has its own internal standard). Once on the approved-equipment list, displacement requires re-approval for the competing product — which hyperscaler EPC teams are highly reluctant to do mid-program. The switching cost is real and is the primary reason Eaton retains specification share once it wins a platform.

Backlog / RPO visibility: Eaton does not separately disclose backlog figures. However, management commentary implies a growing order-to-delivery lead-time situation: "orders are strong, and we're seeing lead times extend for our largest-format products" — consistent with a backlog that provides 6–12 months of forward revenue visibility in the large-project Electrical segment. This is materially less transparent than Vertiv's disclosed $15B backlog, but the underlying economic dynamic is similar.

OCP involvement: Eaton's participation in the OCP 2025 supercapacitor/BBU demonstration (cited in corpus Note 1) is significant beyond the product story: OCP membership gives Eaton insight into hyperscaler power architecture roadmaps 18–24 months before commercial deployment, which reinforces the specification-design-in advantage. Being at the table for OCP Mt. Diablo / Diablo 400 architecture decisions means Eaton products can be shaped to the specification rather than shaped to compete against it.

§ 07Demand Quality

Pull-through demand (dominant and growing):

• Hyperscaler data-center construction is genuine end-customer pull. AWS, Azure, Google, Meta, and Oracle have each committed hundreds of billions of dollars in multi-year capex programs; the switchgear, UPS, and busway go into physical buildings. There is no meaningful channel-fill mechanism in large-project electrical infrastructure — a UPS ships when a data center is ready for it.
• Utility T&D upgrade programs are similarly genuine pull-through — driven by IOU capital plans and state regulatory approvals, not distributor inventory cycles.
• Aerospace content is LTA-anchored and pull-through.

Channel fill (relevant only in catalog/distribution tier — modest):

• The Wesco/Graybar/Rexel distribution channel carries catalog-format Eaton electrical products (panelboards, circuit breakers, fuses). This channel had modest destocking in 2023–2024 during the general industrial destock cycle. As of Q4 2025/Q1 2026, distributor inventory in the Eaton electrical channel appears to be normalizing.
• Estimate: ~5–8% of Electrical Americas revenue is distributor-mediated catalog product susceptible to channel-fill dynamics. Not a material demand-quality concern for the AI-DC thesis.

Pre-buy (tariff / supply-chain driven — modest but real):

• The April 2026 tariff shock has introduced some degree of pre-buy acceleration in Eaton's data-center product categories, particularly for switchgear and UPS components with Asian manufacturing content.
• Eaton has significant North American manufacturing (Buffalo NY, Beaver PA, and other US sites); the tariff exposure is lower than for competitors with heavier Asian production footprints.
• Estimated pre-buy component of recent order strength: ~5–10% of the Q4 2025 order surge may be tariff-accelerated pre-buy rather than pure demand. This is manageable relative to the underlying demand inflection.

The demand-quality picture is strong by industrial standards: Eaton's revenue mix is dominated by long-cycle large-project pull-through (data center builds, utility T&D programs, aerospace LTAs) with a small catalog-distribution tail. Channel-fill reversal risk is contained. Pre-buy risk is present but modest. The primary demand-quality risk is not reversal of orders already placed — it is the calendar-mismatch between backlog and revenue recognition (see below).

Calendar-mismatch (the most important demand-quality issue for ETN):

• Q4 2025 orders ~3x YoY is a backlog event, not a revenue event.
• Large-project switchgear delivery timelines are 12–24 months.
• Large-format UPS (9395) delivery timelines are 8–16 months.
• Revenue recognition follows delivery and commissioning, not order placement.
• Best-estimate revenue conversion of the Q4 2025 surge: 30–40% in H1 2026, 40–50% in H2 2026, 10–20% trailing into H1 2027.
• Schneider's "2028–2030 for 800V real impact" framing is not a contradiction of this — it applies specifically to the 800V-architecture switchgear and UPS products (which require hyperscaler facilities to be 800V-DC-native, a transition that is 2028-era per Schneider). Eaton's near-term (2025–2027) data-center revenue is primarily conventional 48V/AC-architecture electrical infrastructure going into buildings that are preparing for 800V but are not yet 800V-native. The 800V-specific product revenue layer is genuinely 2027–2030.

§ 08Reference-Design Partner Naming — The Binding Cohort Question

The NVTS customer analyst and C-NVTS-1 refinement log established this as "the load-bearing mechanism translating box-builder orders into power-semi P&L." The question: does Eaton name Infineon/TI/Navitas/other as the power-semi inside its 9395 UPS, modular UPS, and power-distribution skids?

Findings as of May 2026:

Verdict: Eaton stays vendor-agnostic on power-semi partnerships in its product literature and customer-facing materials. This is standard practice for an electrical OEM that sources components from multiple vendors across its product portfolio. The implication for the cohort:

1. Eaton is not a named reference-design partner for any power-semi vendor's 800V GaN/SiC product — confirming the NVTS customer analyst's finding that "no box-builder has a publicly named Navitas-anchored reference design" and extending it to: no box-builder (Eaton, Vertiv, Schneider, Delta) publicly names any GaN/SiC power-semi vendor in their customer-facing product materials.
2. The vendor-agnostic posture is a sourcing strategy, not a gap: Eaton's multi-source procurement gives it leverage over power-semi vendors. When the UPS 9395 or modular UPS transitions to GaN/SiC topologies (likely 2027–2029), Eaton will conduct a competitive sourcing process rather than committing to a named partner publicly.
3. ADI as a disclosed partner at the supercap BBU level is a meaningful precedent — it confirms that OCP-context collaborative demos can surface component-level naming. The equivalent GaN/SiC naming may emerge in an OCP 2026 or Computex 2026 demo context.
4. The binding question for the cohort (which power-semi vendor captures the UPS/PDU GaN transition revenue?) is genuinely open as of May 2026. Based on current market positioning — Infineon's 300mm GaN scale, TI's NVIDIA-GTC-named reference architecture, and Navitas's density demos — Infineon and TI are better positioned than Navitas to win Eaton's internal GaN sourcing competition when it occurs, but this is not yet confirmed.

§ 09Demand Durability — Three Time Horizons

NTM (next 12 months, 2026 calendar)

• Q4 2025 order surge converts to revenue 2026 H1/H2. The strongest near-term revenue visibility signal is the backlog from Q4 2025 and prior quarters.
• Electrical Americas revenue growing at estimated 20–30% YoY organically in H1 2026 on the data-center build ramp.
• Utility T&D orders stable and growing as IOUs advance grid upgrade programs; IRA grid credits support the multi-year capex cycle.
• Aerospace steady — defense growing, commercial recovering; modest contribution to NTM growth.
• Vehicle/eMobility headwinds persist — HD truck destocking not fully resolved; eMobility ramping slower than plan.
• Tariff uncertainty (April 2026) introduces NTM pricing complexity; Eaton's US manufacturing base is a competitive advantage vs Schneider (French manufacturing) and Vertiv (mixed). ETN may benefit from tariff-driven specification preference for domestically manufactured equipment.

Composite NTM demand: positive. Data-center pull-through dominates the growth story. Backlog conversion provides near-term revenue visibility. Non-DC segments are mixed but not deteriorating.

1–3yr (2026–2028)

• AI-DC revenue continues to scale. Hyperscaler capex commitments ($600B+ announced for 2026 across the Big 5) are multi-year programs. Eaton's position in the data-center electrical infrastructure specification gets reinforced with each successive build.
• 800V architecture transition begins to matter commercially in 2027–2028. Eaton's participation in OCP Mt. Diablo and Diablo 400 architecture discussions positions it for the switchgear and UPS product refresh required when buildings transition from AC-backbone to 800V DC. This is a product-refresh tailwind, not headwind.
• Utility T&D upgrade cycle extends. PJM $4.8B transmission upgrade approval (Feb 2026 — per corpus Note 2, the first-ever underground HVDC line for Dominion Energy zone) is a multi-year program feeding Eaton's Cooper-era T&D product relationships.
• Cooper-era IOU relationships reinvigorated by data-center interconnect demand. Every GW of AI data center load that comes online requires interconnect capacity; utilities are now running accelerated capital programs to serve it.
• International growth — Electrical Global AI-DC ramp (Europe/APAC hyperscaler build) is lagging North America by 12–18 months; begins to contribute more meaningfully by 2027.

Composite 1–3yr demand: strongly positive. The multi-year alignment of AI-DC build + utility T&D grid upgrade + 800V architecture product refresh creates a durable multi-cycle demand backdrop for Eaton's Electrical segments.

3–5yr (2028–2030)

• Schneider's "2028–2030 real impact" framing is the right calibration for the 800V-native product revenue layer. The switchgear, UPS, and distribution products designed natively for 800V DC architecture (solid-state transformers, 800V-compatible MV switchgear, direct-DC UPS) are 2028–2030 revenue drivers, not 2025–2027.
• Total addressable market expands as the voltage upgrade cycle progresses. Every existing data center eventually needs an electrical infrastructure refresh for 800V; the installed base represents a multi-decade re-specification opportunity.
• Aerospace defense programs provide long-duration earnings stability. F-35, tanker, and next-generation defense programs carry 10–20 year component content windows.
• Structural headwind: vehicle and eMobility remain smaller-than-hoped. If EV ramp continues to disappoint, eMobility stays a sub-5% segment indefinitely.
• Competitive threat: Schneider Electric (the primary comparable at the same value-chain layer) is investing heavily in 800V DC product development; Siemens and ABB are also active in the MV switching market. Eaton's moat is its US manufacturing base, its engineer-of-record specification relationships, and its Cooper-era utility channel — not technological differentiation alone.

Composite 3–5yr demand: positive, with the 800V product-refresh cycle as the structural expansion driver and eMobility as the underperforming optionality.

§ 10AI-DC Mix Dilution — Honest Assessment

The question: is Eaton's diversified industrial mix a strength (multi-cycle stability) or a weakness (dilutes the AI-DC re-rate)?

The honest answer is: both, with a clear hierarchy.

As a strength:

• The diversified base makes Eaton immune to the "one customer loss = revenue cliff" risk. Vertiv, at ~80% AI-DC, is a pure-play that could see a violent downward revision if hyperscaler capex slows. Eaton, at ~16–19% AI-DC, sees a significantly smaller absolute revenue impact from the same event.
• The utility T&D business and aerospace defense base generate stable, long-cycle cash flows that support the balance sheet and dividend through any AI-DC cycle. This is not optionality — it is the reason Eaton trades at a premium to pure-play industrial cyclicals.
• The Cooper-era T&D customer relationships are directly reinforced by the AI-DC build (utilities upgrading grid for hyperscaler load), meaning the diversification is not dilutive but additive — two separate exposures to the same underlying AI-DC capex wave.

As a weakness:

• The AI-DC story at 16–19% of total revenue does not re-rate the entire enterprise the way it re-rates a Vertiv. A 50% growth rate in Eaton's data-center segment (impressive) translates to approximately 8–10% organic revenue growth for the total enterprise, which is not radically different from what a well-positioned diversified industrial would deliver in a good cycle.
• The market's instinct — buying Vertiv for pure-play AI-DC leverage, and Eaton for "lower-risk AI-DC-adjacent diversification" — is correct in the sense that Eaton's P/E re-rate from AI-DC tailwinds will be smaller than Vertiv's.
• The Vehicle and eMobility segments (~11% of revenue combined) are headwinds in NTM that further dilute the total-enterprise growth story.

Verdict on dilution: The diversified mix does not weaken the cohort thesis. The cohort thesis is "chip-to-grid value chain benefits from AI-DC build"; Eaton is a G1/G2/G3 layer participant whose revenue is growing at double-digit rates driven by the thesis. What the diversified mix does is limit the multiple expansion upside relative to Vertiv. This is a conviction-sizing question, not a thesis-invalidation question. The correct portfolio-construction response is: Eaton is a lower-multiple, lower-volatility AI-DC infrastructure exposure vs Vertiv — appropriate for investors who want the AI-DC infrastructure thesis with a margin of safety from the diversified base. ETN is not the right position for an investor seeking maximum AI-DC leverage; VRT is. ETN is the right position for an investor who believes in the infrastructure thesis but wants the earnings durability of a $20B+ revenue diversified industrial.

§ 11Bull Points

• Data center already ~16–19% of total revenue and growing at 30–50% YoY — by TXN's benchmark this is "material" by a factor of 2x, and the Q4 2025 +3x YoY order print implies the growth rate is accelerating, not decelerating.
• Engineer-of-record specification moat — once Eaton switchgear and 9395 UPS are specified into a hyperscaler campus electrical design, the switching cost for the end-customer (re-approval, re-commissioning, re-training) is substantial. The specification is the moat, not the product itself.
• Cooper-era T&D utility relationships are doubly reinforced: the utility is simultaneously Eaton's direct T&D customer (for grid equipment) and the enabler of the hyperscaler campus's power supply (for the AI-DC build). Two revenue streams from one underlying demand wave.
• US manufacturing advantage in the tariff era — Eaton's domestic manufacturing base in electrical products (Buffalo, Beaver, multiple US sites) is a structural advantage versus Schneider (French manufacturing, US-sold) and ABB/Siemens (European manufacturing, US-sold) under the April 2026 tariff regime.
• OCP participation (supercap BBU demo, 2025) puts Eaton at the table for next-generation DC power architecture discussions 18–24 months before commercial deployment — reinforcing specification-design-in advantage.
• Diversified mix insulates from single-cycle risk — Aerospace defense, utility T&D, and industrial OEM segments provide earnings durability through any AI-DC demand moderation.
• Backlog conversion from Q4 2025 provides 12–18 months of revenue visibility in the large-project Electrical segment; meaningful near-term forward revenue confidence.
• 12–24 month qualification moat on large-format products — every competing UPS or switchgear vendor that wants to displace Eaton mid-program faces a re-approval cycle that the hyperscaler EPC team will not accept on an active build.
• Aerospace LTAs provide a structurally non-correlated revenue layer (~14–15% of revenue) that is inherently un-threatened by any AI-DC cycle dynamics.

§ 12Bear Points

• AI-DC mix at 16–19% limits total-enterprise re-rate. Investors seeking maximum AI-DC leverage will prefer Vertiv (higher mix, higher multiple expansion potential). Eaton is the "lower beta, lower reward" expression of the same thesis.
• Calendar-mismatch between orders and revenue — Q4 2025 +3x YoY orders converts to revenue over 12–24 months; 2026 revenue will show strong growth, but the full Q4 2025 order surge is a 2026–2027 revenue event.
• Schneider's 2028–2030 framing is a real calibration for 800V-native product revenue — the highest-value product categories (800V switchgear, direct-DC UPS) are not yet in commercial deployment; the current order surge is primarily conventional AC-architecture electrical infrastructure.
• Vendor-agnostic power-semi posture means Eaton does not generate the downstream named-partner benefit for cohort power-semi companies (NVTS, TXN) — the GaN/SiC transition inside Eaton products will occur but will be competitively sourced and undisclosed.
• Vehicle and eMobility (~11% of revenue) are headwinds in NTM — HD truck destocking, EV ramp slower than 2022-era forecasts; these segments dilute total-enterprise organic growth.
• Competition from Schneider, ABB, and Siemens in the MV/LV electrical infrastructure space is intense; Eaton is not the sole beneficiary of AI-DC infrastructure demand.
• No disclosed backlog / RPO figure — unlike Vertiv ($15B disclosed) or AVGO (formal RPO), Eaton's forward visibility is inferred from lead-time commentary rather than disclosed. This is a transparency gap relative to peers.
• Distribution channel (Wesco, Graybar, Rexel, Sonepar) carry modest channel-fill risk for the catalog electrical product tier — not material to the AI-DC thesis but a headwind in any industrial destock cycle.
• eMobility structural underperformance — if EV adoption ramp continues to run below forecast, eMobility stays a sub-5% sub-scale segment with ongoing investment drag.
• No power-semi named partner — for the cohort's "box-builder reference-design = power-semi P&L" question, Eaton's vendor-agnostic posture means the GaN/SiC transition inside Eaton products does not flow to a named cohort power-semi partner (NVTS, TXN) in any disclosed way. The linkage is real economically but invisible in disclosed form.

§ 13Conviction (1–5)

3 / 5.

The customer dimension on ETN is structurally sound: data center is material (~16–19% of revenue), growing rapidly (30–50% YoY estimated; +3x orders Q4 2025), and supported by genuine engineer-of-record switching costs. The utility T&D customer base is reinforced, not diluted, by the AI-DC build. The Aerospace defense segment provides non-correlated earnings durability. What keeps conviction at 3 rather than 4:

1. Calendar-mismatch — the order surge is 12–24 months from full revenue recognition; 2026 will be a strong year but the full print arrives in 2027.
2. Mix dilution narrative headwind — at 16–19% AI-DC mix, the total-enterprise re-rate is real but not a Vertiv-class multiple expansion.
3. Corpus thinness — confirmed by synthesis and refinement log; this analysis relies substantially on public filings and earnings transcripts rather than user-note-grounded confirmations.
4. 800V product revenue back-end loaded — Schneider's 2028–2030 framing is an honest calibration for the highest-ASP 800V-native product categories.

A 4/5 conviction would require: (a) a disclosed Eaton backlog / RPO figure confirming $8–10B+ in data-center orders; (b) a named GaN/SiC power-semi partnership for the UPS 9395 or modular UPS 800V-transition product; (c) explicit management guidance for data-center % of total Eaton group revenue reaching >25%.

A 2/5 conviction would require: hyperscaler capex revision materially lower (i.e., AI-DC build rates cutting 30%+), which would cascade to significantly lower data-center order intake.

§ 14Key Risks to This Read

1. Calendar-mismatch / backlog conversion — the Q4 2025 order surge is the bull thesis; if lead times extend further or hyperscaler projects slip (permitting, interconnect queue), revenue recognition stretches into 2027–2028 and the near-term earnings case weakens.
2. Hyperscaler capex revision — if the Big 5 hyperscalars cut 2026 build rates (recession risk, ROI concerns on AI), the data-center order pipeline stalls. This is the single highest-leverage macro risk on the customer dimension.
3. Schneider / ABB / Siemens specification wins — Eaton does not win every hyperscaler data-center specification; where Schneider or ABB wins the engineer-of-record approval, Eaton loses the specification lock-in. Competitive intensity in this segment is high.
4. Tariff complexity — while US manufacturing is a structural advantage, tariffs on component inputs (copper, steel, transformers) could raise Eaton's own input costs even as tariffs protect finished product share.
5. 800V UPS/switchgear product readiness — Eaton's 800V-native product line (direct-DC UPS, 800V-compatible switchgear) must be ready for the 2027–2028 commercial inflection; if Schneider or Vertiv ship credible 800V products earlier, Eaton risks being displaced in the next-generation specification.
6. eMobility underperformance continues — ongoing investment drag from a sub-scale segment; if EV ramp does not accelerate, eMobility creates recurring earnings headwind.
7. Industrial and commercial cycle softness — if the non-DC 75–80% of Eaton's revenue faces a macro downturn, the total-enterprise earnings story weakens materially even if data-center grows strongly.
8. Utility regulatory lag — utility T&D capex is gated by rate-case approvals; a hostile regulatory environment for IOUs (rate freezes, disallowed spending) could slow the T&D customer build.
9. Cooper-era product commoditization — Bussmann fuses, circuit breakers, and catalog electrical products face Chinese competition at the commodity tier (CHINT, Delixi, Schneider's low-cost range). The commodity tier is not the growth driver, but commoditization pressure compresses the distribution channel's contribution margin.
10. The box-builder named-partner gap — until Eaton publicly names a GaN/SiC power-semi partner in its 800V reference products, the linkage to cohort power-semi P&L (NVTS, TXN) remains economically real but analytically undisclosed.

§ 15Sources