Phase A — Understand the business
Lens 1 · Company Overview
Vertiv designs, sells, installs and services the critical digital infrastructure inside a data center that is not the compute itself — the power train (UPS, switchgear, busbar, power distribution, the new 800VDC architecture), the thermal train (air handlers, chillers, and increasingly direct-to-chip liquid cooling and coolant distribution units), plus racks, integrated modular/prefab systems, monitoring software, and a large global services network. Its own framing: "a global leader in critical digital infrastructure" serving three end-markets — (1) data centers (hyperscale/cloud, colocation, neocloud, enterprise), (2) communication networks, (3) commercial & industrial ``.
It is, in plain terms, the picks-and-shovels arms dealer of the AI buildout: it does not care which model wins, only that racks get denser and hotter. The 10-K names its end-customers by tier — hyperscale (it explicitly cites Amazon Web Services and Google Cloud), colocation operators, and the neocloud cohort (it explicitly names CoreWeave and Nebius) ``. That neocloud call-out is notable: Vertiv is selling into both the incumbents and the challengers.
Contract structure: order-book / backlog driven, not recurring SaaS. Orders convert to revenue over a "12 to 18 month" shipment window; a "majority" of backlog is "considered firm" but customers can defer or cancel firm orders, a risk the 10-K flags explicitly ``. Services & spares is the annuity layer riding on the installed base.
Scale (FY2025): net sales $10,229.9M, of which 62% was transacted in the Americas ``.
Lens 2 · Supply Chain
Upstream → Vertiv → end-customer, named where the filings + web allow:
- Upstream inputs: electronics/power semiconductors and components, copper and aluminum (busbar, heat exchangers), steel/sheet-metal enclosures, compressors and pumps, refrigerants/coolants, contract-manufactured sub-assemblies. The 10-K states a deliberate diversification strategy "to avoid over-concentration or significant dependence on a particular supplier or region" `` — i.e., no single named critical supplier disclosed, which is itself the disclosure.
- Vertiv (the integrator): manufactures across the Americas, EMEA and APAC; is adding factory capacity (capex stepping from a historical 2–3% of sales to 3–4%, new plants coming online)
. Bolt-on M&A is buying *thermal-chain depth*: PerchRight (fluid management, Q4 2025, part of ~$1B Q4 deal spend), **Strategic Thermal Labs** (server-side↔facility liquid-cooling interface, Apr 2026), **ThermoKey S.p.A.** (heat-exchanger/thermal manufacturing, EMEA, closed Jun 2026), and Waylay NV (AI infra monitoring software, Aug 2025) .
- Downstream chokepoint Vertiv itself sits on: it is one of only a handful of vendors that can deliver an integrated power+thermal reference design at hyperscale rack density. The most important downstream relationship is NVIDIA — Vertiv co-develops reference architectures (142 kW cooling+power for the GB300 NVL72; 800VDC for 1MW+ racks) and ships its designs as SimReady 3D assets inside NVIDIA's Omniverse blueprint ``. NVIDIA is not a customer that pays Vertiv; it is the demand-shaping standard-setter whose roadmap (GB300 → Kyber → Vera Rubin) dictates Vertiv's product cadence.
- End-buyers: the hyperscaler capex pools (Microsoft, Amazon/AWS, Alphabet/Google, Meta), colocation operators (Equinix, Digital Realty-type), and neoclouds (CoreWeave, Nebius).
Chokepoints / single-source risk: the binding constraint is increasingly external to Vertiv's own chain — grid interconnection and power availability at the data-center site (capacity is being announced faster than utilities can energize it) ``. That is the supply-chain risk that actually matters: not Vertiv's inputs, but whether its customers can plug in.
Lens 3 · Competitive Advantages (moats)
Vertiv's moat is integration + installed base + co-design proximity to NVIDIA, and it is real but not impregnable.
- Breadth/integration moat. Vertiv spans power and thermal and racks and monitoring and services. Pure-play cooling or pure-play power vendors cannot offer a single integrated reference design; the integration is what wins the hyperscaler relationship ``. This is the strongest moat.
- Switching costs / installed base. Its global services network and installed base create a sticky annuity and a default-incumbent position on the next build. Services revenue grew alongside products in FY2025 (service & spares improved $39.0M in one segment alone) ``.
- Co-design / standard-setting proximity. Being inside NVIDIA's Omniverse blueprint and co-authoring the 800VDC and GB300 reference designs is a time-to-market moat — Vertiv's gear is the reference, so it ships first ``.
- Bargaining power: strong over suppliers (diversified, commoditized inputs); weaker over its largest customers — a handful of hyperscalers with enormous purchasing leverage. The 10-K itself lists "less leverage with large customer contracts" as a risk ``. Who needs whom more is genuinely two-sided right now (capacity is scarce), but in a downturn the hyperscalers hold the whip.
Durability verdict: scale + integration + IP/process is a 3–5-year moat, not a 20-year one. Eaton, Schneider, nVent and Modine are all buying liquid-cooling capability fast (see Lens 7/13). The moat is being out front on the next architecture, which must be re-won every NVIDIA generation.
Lens 4 · Segments
Vertiv reports on three geographic segments (Americas, Asia Pacific, EMEA) — not product lines — so the segment story is regional. All figures ``, $ millions, fiscal year = calendar year:
| Segment (net sales) | FY2023 | FY2024 | FY2025 | FY25 YoY |
|---|
| Americas | 4,500.6 | (≈5,? ) | 6,386.3 | strong double-digit |
| Asia Pacific | 1,717.8 | — | 2,019.2 | up |
| EMEA | 1,793.4 | — | 1,824.4 | roughly flat |
| Total net sales | — | 8,011.8 | 10,229.9 | +27.7% |
(Note: the 10-K presents a three-year segment table; FY2024-by-segment intermediate values were not cleanly isolatable from the run-on filing text — total FY2024 net sales $8,011.8M is -confirmed, and the FY2023/FY2025 segment columns above are . Treat the FY2024 per-segment cells as n/a — not cleanly parsed.)
Segment operating profit, FY2025 ``: Americas $1,714.3M, APAC $222.1M, EMEA $377.4M → segment total $2,313.8M, reconciling to ~$1,829.7M consolidated operating profit after corporate/FX.
Trend & cause: the Americas is the entire growth story — it is 62% of revenue and carries the AI data-center demand. In Q1 2026 Americas net sales hit $1,814.4M, +$629M / ≈+53% YoY (organic +44%), explicitly attributed to data-center demand . EMEA is the soft spot — roughly flat in FY2025 and, per Q1 2026 commentary, down sharply (see Lens 5/13). APAC is steady mid-tier. The geographic concentration in the Americas is both the bull case (it's where AI capex is) and a concentration risk.
Phase B — Measure performance
Lens 5 · Earnings Result (latest print = Q1 2026, quarter ended 2026-03-31)
The latest 10-Q is an unambiguous beat-and-raise :
- Net sales $2,649.5M, +$613M / +30% YoY. Decomposed: 23% organic + 4% acquisitions + 3% FX ``.
- Gross profit $999.7M = 37.7% of sales (up from 36.3% FY2025) — margin expanding ``.
- Adjusted operating margin 20.8%, +430 bps YoY ``.
- Net income $390.1M (vs $164.5M Q1 2025) = +137% ``.
- Diluted EPS $0.99 (vs $0.42 Q1 2025); adjusted diluted EPS +83%
.
- Operating cash flow $766.8M (+$463M YoY); adjusted FCF $653M (+147%)
.
- Americas +53% reported / +44% organic — the driver ``.
Guidance & tone (raised): management lifted FY2026 to net sales $13.5–14.0B, organic +29–31%, GAAP diluted EPS $5.60–5.70, adjusted diluted EPS $6.30–6.40 — adjusted EPS +51% at the midpoint vs FY2025 ``. Tone is confidently aggressive (capex stepping up to fund capacity).
FY2025 context (full year) : revenue $10,229.9M (+27.7%); gross margin 36.3%; operating profit $1,829.7M; **net income $1,332.8M (vs $495.8M FY2024)**; diluted EPS ≈ **$3.41 GAAP**; adjusted diluted EPS **$3.49** . OCF $2,113.8M, capex $226.4M → FCF ≈ $1,887M.
Balance-sheet flags: Term Loan principal $2,076.1M, due 2032, ~5.61% rate ``; cash ~$1.7B; net leverage modest (~0.3–0.5x EBITDA range). The flag to watch is working capital: a $15B backlog being built into a capacity ramp can swell inventory and receivables — monitor whether they outrun revenue (they did not in Q1, OCF surged). One provenance conflict to flag: some web sources frame "FY2024 net income $1.32 / up 169%" — this conflicts with the 10-K's GAAP FY2024 net income of $495.8M. The 10-K GAAP figure governs; the "$1.32" web framing appears to confuse a prior-period or per-share basis. Cited the filing.
Lens 6 · Earnings Calls (sentiment trend)
No transcripts on the shelf (transcripts=0), so this is ``. Across the last ~4 prints the management narrative has escalated monotonically from "AI is a tailwind" to "thermal is now the gating constraint and we are the answer":
- CEO Albertazzi (Mar 2026): liquid-cooling capacity growing "really, really, really rapidly"; "thermal performance is now a critical enabler of capacity and efficiency" ``.
- Recurring phrases: "market share gains," "deployment speed," "thermal chain," "capacity investment," "reference architecture with NVIDIA."
- What they started saying in 2025–26: explicit capex step-up (2–3% → 3–4% of sales), neocloud as a named channel, 800VDC roadmap.
- What's notably absent / downplayed: EMEA. Management leans on a second-half regional recovery to hit the FY guide — a tell that EMEA softness is real and they'd rather talk about the Americas (see Lens 13).
Tone: bullish, increasingly operational (capacity/throughput) rather than demand-evangelical — consistent with a company that has already won the demand and is now execution-constrained.
Lens 7 · Comps
Peer set = global electrical/thermal infrastructure names with data-center exposure. Multiples are ``, dated; where I could not source a clean figure I mark n/a. I did not fabricate any multiple.
| Company | Ticker | Mkt cap | EV/EBITDA | P/E (fwd) | Div yield | 5-yr avg ROE | DC exposure |
|---|
| Vertiv | VRT | ~$125–130B | n/a | ~82x trailing / fwd lower on guided EPS | ~0.07% | high (post-recap, rising) | ~pure-play |
| Schneider Electric | SU / SBGSY | large-cap (~$34.2B rev) | ~18–20x | ~25x fwd / ~35x trailing | yes | mid-teens | ~15% of biz |
| Eaton | ETN | large-cap | n/a | n/a | yes | mid-teens | growing (Boyd Thermal) |
| nVent Electric | NVT | mid-cap | ~22x TTM | ~26–28x fwd | yes | mid-teens | ~30% |
| Modine | MOD | mid-cap | n/a | n/a | minimal | improving | growing |
Sources: ``.
Read: VRT trades at a large premium to every peer — ~82x trailing P/E vs a peer average cited around 40x and a US-electrical-industry average of ~39.7x ``. On forward guided EPS ($6.30–6.40 adjusted) the multiple compresses to roughly ~52x, still ~2x the peer group. The premium is "fastest organic grower + pure-play + NVIDIA-proximate." The risk is that the premium is the whole thesis — see Lens 11/12. VRT's own trailing EV/EBITDA was not cleanly sourced; n/a rather than invent it.
Lens 8 · Stock-Price Catalysts (what moves >5%)
``, dated. Pattern over the last ~3 years:
- Up moves are almost always orders/backlog + guidance raises, not the revenue print itself. Mid-2024: organic orders +57% YoY, book-to-bill 1.4x → guidance raise → rally
. Q1 2026: +11.8% on the raised guide .
- The single biggest signal the market rewards is the ORDER number / book-to-bill, because it forward-prices the backlog. Backlog $7.2B → $15.0B (2024→2025) is the headline that drove the re-rating ``.
- Down moves are macro/AI-sentiment, not company-specific misses: the stock fell ~25% in June 2026 on an AI-data-center oversupply panic (sector-wide), and carries a 52-week range of $107.38–$379.94 with a beta ~2.04 ``. It is a high-beta AI-capex proxy.
- What the tape reveals: the market trades VRT as a leveraged bet on hyperscaler capex direction. Orders up = re-rate; any whiff of capex deceleration = violent de-rate. Idiosyncratic execution matters less than the AI-capex narrative.
Recent dated moves: +4.9% on 2026-06-19, −3.85% on 2026-06-16 `` — high day-to-day volatility around the June pullback.
Phase C — Judge people & books
Lens 9 · Management
- CEO Giordano Albertazzi — appointed CEO 2023 (rose internally; ran EMEA / was COO before the top job). An operator, not a founder. Track record: he has overseen the operating-margin transformation — adjusted operating margin from low-teens to 20.8% in Q1 2026 (+430 bps YoY) and revenue roughly tripling off the 2020 de-SPAC base
. Strong delivery on the demand wave.
- Capital allocation: reinvest-and-bolt-on. ~$1B of M&A in Q4 2025 (PerchRight + others), plus Strategic Thermal Labs, ThermoKey, Waylay — all tuck-ins buying thermal-chain capability, on-strategy
. Capex stepping to 3–4% of sales to fund capacity . Capital returns are modest but rising: dividend raised 67% to $0.25/yr (Nov 2025), and opportunistic buybacks (a ~$599.9M repurchase referenced in FY2024) . Reinvestment >> return, which is correct at this point in the cycle.
- Skin in the game / insider behavior — a yellow flag. Over the trailing 6 months insiders logged 65 open-market sales and 0 purchases; director Roger Fradin sold
203,333 shares ($51M, ≈57% of his direct holding) at ~$253 in Feb 2026 ``. Some of this is post-de-SPAC/option-driven diversification, but zero insider buying into a 30%-organic-growth story is not a vote of confidence at this multiple.
- CEO comp: $18.3M for 2025 (+33% YoY) `` — rich but not egregious for a ~$125B-cap company; mostly equity.
- Former owner: Platinum Equity (the PE sponsor that took it public via the Cote/GS SPAC de-SPAC in 2020) fully exited by 2023 `` — overhang gone.
- Archetype: professional manager executing an inherited tailwind extremely well. The risk is the flip side of that — a professional manager has less founder-style conviction to hold capacity investment through a demand air-pocket.
Lens 10 · Forensic Red Flags
Ground: +.
- Revenue recognition / backlog: backlog ($15.0B) is the single most scrutiny-worthy number — it is an estimate of firm + non-firm orders, customer-cancellable, with a 12–18 month conversion window; the 10-K explicitly warns it "may not be fully realized" ``. Not a red flag per se, but the metric the bull case leans on is the metric with the most management discretion.
- Cash vs earnings: clean — OCF $2,113.8M FY2025 exceeds net income $1,332.8M; Q1 2026 OCF $766.8M vs net income $390.1M. Earnings are cash-backed, the opposite of a red flag ``.
- Working capital: the thing to watch as capacity ramps — inventory and receivables against a $15B backlog. No divergence yet (Q1 OCF surged), but a capacity-ramp + backlog combination is where receivables/inventory can outrun revenue. Monitor.
- SBC / non-GAAP gap: the company guides both GAAP ($5.60–5.70) and adjusted ($6.30–6.40) EPS — a ~$0.70 / ~12% gap, driven substantially by SBC and acquisition/amortization add-backs ``. Material but disclosed; the adjusted/GAAP wedge is worth tracking as M&A amortization rises.
- Goodwill/intangibles: rising with the M&A spree (PerchRight, ThermoKey, Waylay, Strategic Thermal Labs). Acquisition accounting and future impairment risk grow with the deal cadence — standard for a serial acquirer, flag for monitoring.
Regulatory findings (required sub-section):
- SEC Litigation Releases: None. EDGAR EFTS (LR) returned 0 findings for "Vertiv Holdings" since 2021-06-22 ``.
- SEC AAERs: None. EDGAR EFTS (AAER) returned 0 findings ``.
- Non-SEC (FTC/DOJ/etc.): web search returned no material enforcement actions, settlements, fines, or class actions against Vertiv in 2024–2026 ``.
- 10-K Item 3 (Legal Proceedings): Vertiv states it is "not a party to any material, pending legal proceedings or claims" as of Dec 31, 2025, beyond incidental, normal-course matters (general/product liability, commercial, IP, labor) for which it accrues contingencies ``.
- Conclusion: No material regulatory or legal findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 10-K Item 3 as of 2026-06-22.
Phase D — Project & stress-test
Lens 11 · Forward Projection
Bottom-up from FY2025 actuals + the raised FY2026 guide. Anchor: FY2025 adjusted diluted EPS $3.49 ; FY2026 guided adjusted diluted EPS **$6.30–6.40** (GAAP $5.60–5.70) . Fiscal year = calendar year; shares ~388–392M diluted, ~0.5% annual net dilution assumed.
FY2026 (guided year):
- Base: adjusted diluted EPS $6.35 [research basis: management guide midpoint, ``]; revenue ~$13.75B (organic +30%).
- Bull: $6.70.
- Bear: $5.60.
FY2027:
- Base: $8.25.
- Bull: $9.20.
- Bear: $6.00.
FY2028:
- Base: $9.50.
- Bull: $11.50.
- Bear: $5.50.
Valuation cross-check: at ~$333, the base path implies ~52x FY26 / ~40x FY27 / ~35x FY28 adjusted EPS. The market is paying for the bull trajectory. If the bear FY2027 ($6.00) prints, a re-rating to even 25x = ~$150 — the "~50% drawdown" the bears cite is arithmetically consistent ``.
(Per --watchlist rules, no forecast.ts create logged — breadth mode produces the dossier only. The base FY2027 call "VRT FY27 adjusted diluted EPS ≥ $8.25" is the natural Brier candidate if/when this is promoted to a thesis.)
Lens 12 · Bull vs Bear
Bull case. Vertiv is the best-positioned pure-play on the single biggest capex super-cycle of the decade. The $15.0B backlog (+108% YoY) forward-prices ~1.5 years of growth that is already booked, not hoped-for . The structural tailwind — racks going from ~40kW air-cooled to 140kW+ liquid-cooled, then 800VDC at 1MW+ — is a **multi-year content-per-rack expansion** where Vertiv's dollar-share *rises* every generation, and its NVIDIA co-design seat means it ships the reference design first . Margins are still climbing (20.8% adj op, +430 bps) with operating leverage and mix; FCF is real ($1.9B FY2025). Management is executing and buying thermal-chain depth on-strategy. The contrarian bull point: the market is currently mispricing this as "cyclical AI-capex beta" when a chunk of the backlog is non-discretionary maintenance/efficiency retrofit on an installed base that only grows.
Bear case (2–3 things that could permanently impair or de-rate).
- It's priced for perfection. 82x trailing / ~52x forward vs ~40x peers, beta 2.04, "no margin of safety" ``. The valuation is the thesis.
- Customer concentration into a capex pool that must decelerate. Hyperscaler 2026 capex ~$700B vs ~$410B prior `` — that growth rate is unsustainable; when it normalizes, Vertiv's order growth de-rates first and hardest. A handful of customers; one deferral is outsized.
- EMEA is already cracking — organic orders/sales down sharply (Q1 2026 EMEA organic ~−29%) and the FY guide needs a 2H recovery ``. That's a falsifiable near-term crack in the story.
Pre-mortem (18 months out, thesis broke): Hyperscaler capex guidance was cut in late-2026/2027 on AI-ROI doubts and grid-interconnection delays; Vertiv's book-to-bill fell below 1.0x; a chunk of "firm" backlog was deferred; EMEA never recovered; and a stock priced at 50x+ re-rated to 25x on the first sub-guide quarter. The stock halved not because the business broke but because expectations did.
Contrarian view of what the market refuses to see: right now (post the −25% June pullback) the market may be over-weighting the oversupply/deceleration narrative and under-weighting that content-per-rack is rising faster than rack count — so Vertiv revenue can grow even if data-center unit growth slows. The 800VDC transition is a Vertiv-specific dollar-share catalyst the AI-capex-bears are conflating with generic capex.
Lens 13 · Devil's Advocate (short-seller)
Dismantling the bull case.
- Where revenue is concentrated: ~62% Americas, heavily hyperscale/neocloud — a bet on 4–5 buyers' capex plans
. If two hyperscalers trim AI capex growth from +70% to +20%, Vertiv's order line inverts. The 10-K's own risk factor: "less leverage with large customer contracts" .
- The moat is shallower than bulls think. Eaton bought Boyd Thermal (~$1.7B 2026 contribution), Schneider is the largest electrical player (~$34B rev) leaning in, nVent has ~30% DC exposure and a deepening NVIDIA tie, Modine is scaling ``. Vertiv's "integration" edge erodes as every diversified giant assembles the same stack via M&A. The most dangerous competitor bulls underestimate: Schneider Electric — it can out-spend Vertiv on R&D and capacity and already owns the electrical relationship at most sites.
- Capital-allocation skeptic's note: a serial acquirer (4+ deals in 12 months) buying into a hot sector at peak multiples builds goodwill that becomes impairment risk if the cycle turns; and insiders are selling 65:0 — they are not buying their own "generational" story ``.
- What must hold for today's price: ~30% organic growth for multiple years, margins to ~22%+, no hyperscaler air-pocket, EMEA recovery, and multiple staying >40x. That's a lot of "ands."
- If growth disappoints 20–30%: FY2027 EPS to ~$6.00 (bear) and a re-rate to 25x → ~$150, roughly −55%.
- Single scenario that permanently impairs: not bankruptcy (balance sheet is fine) but a structural AI-capex reset (model-efficiency gains + ROI disappointment cut hyperscaler buildout) that turns a $15B backlog into deferrals and resets the multiple for years. Plausibility: moderate — the binding question of the whole AI trade.
Lens 14 · Management Questions (ordered by information value)
- Of the $15.0B backlog, what share is firm/non-cancellable vs. cancellable, and what is the realistic cancellation/deferral rate you'd see in a hyperscaler-capex deceleration?
- What is the current book-to-bill, and what does the order-trend look like month-over-month in Q2 2026 — is it still above 1.0x?
- EMEA: organic sales/orders fell sharply — what specifically broke, and what gives you confidence in the 2H 2026 recovery the FY guide depends on?
- Quantify content-per-rack today (air → 140kW liquid → 800VDC): what is Vertiv's dollar-share of a GB300 NVL72 rack vs. a Vera Rubin rack — is your served-content actually rising per generation?
- What is your customer concentration — what % of revenue and of backlog is the top-1, top-3, top-5?
- As Schneider, Eaton (Boyd Thermal) and nVent scale liquid cooling, where do you lose share first, and how do you defend the integration premium?
- On the 800VDC portfolio (H2 2026 launch) — what's the revenue ramp, the margin profile vs. legacy power, and is NVIDIA's roadmap a contract or a collaboration?
- Capex is stepping to 3–4% of sales — what utilization/payback underwrites the new factories, and what happens to that capex if orders soften?
- Walk through the GAAP-to-adjusted EPS bridge ($5.60–5.70 → $6.30–6.40) — how much is SBC vs. M&A amortization, and where does that wedge go in 2027?
- M&A: four deals in twelve months — what's the integration risk, the cumulative goodwill, and your discipline on multiples paid into a hot sector?
- What is your pricing power right now — how much of organic growth is volume vs. price, and does price hold if capacity catches up to demand?
- Services & spares as a % of revenue and its margin — how big can the installed-base annuity get, and how counter-cyclical is it?
- Grid/interconnection is the customers' binding constraint — how much of your pipeline is gated by power availability you don't control?
- Capital returns: with
$1.9B FCF, why is the **dividend still tiny ($0.25/yr)** and buybacks modest — is this permanent reinvestment, or will returns scale?
- Insiders sold 65:0 over six months — how should shareholders read zero insider buying into a story management calls generational?