VRT — Research Report (2026-07-27)

Vertiv Holdings Co (NYSE: VRT) · data as of 2026-07-27 · not investment advice, a personal research note

TL;DR

• What it is: Vertiv makes the "picks and shovels" of data centers — power gear (UPS, switchgear, busbars) and cooling (air + liquid) that keep AI server halls running. Old Emerson Network Power, IPO'd 2020.
• Overall take: A genuine AI-infrastructure bottleneck with exploding orders — but the stock already prices in years of perfection.
• Educational decision: Watch (Medium conviction) — great business, demanding price; want a better entry or proof the order surge isn't a one-cycle spike.
• Revenue $10.23B, +27.7% YoY (AI data-center capex boom).
• Backlog $15B, book-to-bill 2.9× (2+ years of orders already in hand).
• FCF $1.89B, ~89% cash conversion (profits are real cash, not accounting).
• Forward P/E ~48× vs industrials ~29× (priced for flawless execution).
• Strongest bull: organic orders grew +252% in Q4 2025 — demand is accelerating, not fading.
• Strongest bear: valuation leaves no room for an AI-capex digestion year; any order air-pocket re-rates it hard.
• Watch next: Q1 2026 book-to-bill and full-year 2026 order growth — does the 2.9× hold, or was Q4 the peak?

1. Business overview

So what: Vertiv sells the power and cooling backbone every AI data center needs — it wins no matter which cloud or chipmaker wins.

Vertiv builds the physical infrastructure around the servers, not the servers themselves. Two things it does:

Revenue model: Large project-based equipment orders (lumpy, booked into backlog) plus a recurring services annuity. Customers are hyperscalers (Amazon, Microsoft, Google, Meta), "colocation" operators (they rent data-center space), and enterprises.

Geography: Americas leads growth; APAC mid-20s%; EMEA (Europe/Mideast/Africa) soft in early 2026, expected to recover H2. (Q4 2025 release)

Jargon note — "hyperscaler": the handful of giant cloud companies that build data centers at massive scale. "Colocation/colo": companies that build data-center space and rent it out. "CDU" (Coolant Distribution Unit): the box that pumps liquid coolant to the chips.

2. Financials (5-year)

So what: revenue nearly doubled since 2021, margins are climbing, and cash generation went from negative to gushing.

All figures $M, from SEC EDGAR company facts (US-GAAP, FY ends Dec 31).

Reading it:

• Revenue CAGR 2021→2025 ≈ 20%/yr, accelerating to +27.7% in the last year — this is a company hitting an inflection, not decaying. (CAGR = the smooth yearly growth rate.)
• Margin slope is the story. Operating margin went 3.9% → 17.9% in three years — that's pricing power showing up (demand > supply for AI power/cooling).
• Cash conversion is excellent. 2025 FCF $1.89B on EBITDA $2.14B ≈ 89% — profits turn into real cash. The 2022 negative FCF was a working-capital build to fund the order surge, now unwinding.

⚠️ One-off flag: GAAP net income jumped +169% (496→1,333) while operating income rose only +34%. Most of that gap is non-operating (prior-year warrant/derivative fair-value marks and tax items depressed 2024; 2025 is cleaner). Use adjusted EPS for valuation, not GAAP net income — the GAAP line is noisy. (2020's $1.3B revenue line is a SPAC-merger partial-period artifact — ignore it; real 2020 sales were ~$4.4B.)

Net debt & liquidity: Net debt = $2,892M debt − $1,728M cash ≈ $1.16B. Against EBITDA $2.14B that's ~0.5× net debt/EBITDA — very low (anything under 3× is comfortable). Balance sheet is a non-issue.

Returns on capital: ROIC ≈ 28% (NOPAT ~$1,446M ÷ invested capital ~$5,105M), far above a ~9% cost of capital. Translation: every $1 Vertiv reinvests creates real value — the sign of a moat.

3. Metrics table — with the "why this matters here"

So what: every number below is cheap-or-expensive only in context; here's the context.

4. Valuation view — is 48× defensible?

So what: VRT is priced at a ~65% premium to its own sector, justified only if 20%+ growth persists for years — a high bar, but the backlog partly funds it.

Company vs sector (always show the yardstick):

Which multiple and why: VRT is profitable and cash-generative, so P/E and EV/EBITDA are the right lenses (EV/Sales is for pre-profit names). The premium is a growth premium: VRT is guided to grow 2–3× faster than Eaton or Schneider, so a higher multiple is logical — the question is how much higher.

What a reasonable buyer might pay: On 2026 adj EPS $6.02, a "growth-industrial" multiple of 30–40× (still a premium to the sector, reflecting superior growth) implies **$180–$240**. The market at ~$290 is paying ~48×, i.e. crediting the current AI-capex pace continuing well into 2027–28. Analyst consensus 12-month target is $376 (Strong Buy, 28 analysts, MarketBeat) — so the Street is even more bullish than the current price. This is a "quality business, priced-for-perfection" setup: the disagreement is about the multiple, not the business.

5. Ownership, management & incentives

So what: this is a turnaround built by a legendary operator (Dave Cote), and the balance sheet/margin discipline shows his fingerprints.

• Chairman: David Cote — former Honeywell CEO who ran the SPAC (GS Acquisition Holdings) that took Vertiv public in Feb 2020. His operating playbook (margin discipline, cost-out) drove the 3.9%→18% margin transformation.
• CEO: Giordano "Gio" Albertazzi (since Jan 2023) — long-time Vertiv/Emerson insider, ran the EMEA and data-center businesses.
• Origin: Emerson Network Power → bought by Platinum Equity (2016) → SPAC IPO (2020). It's a private-equity-style operational turnaround that worked.
• Dilution history: Shares rose 342M→383M since 2020, mostly early SPAC warrants; now roughly flat — the silent-dilution risk has passed.
• Insider/Form 4 signals: not verified in this pass — worth checking EDGAR Form 4 before acting.

6. Market & value-chain position

So what: Vertiv is a near-bottleneck in AI power/cooling — but it's a strong duopoly-ish position, not a monopoly, and its customers (hyperscalers) hold real bargaining power.

The market: Data-center liquid cooling alone is projected to grow from $4.1B (2026) to ~$27.7B (2033), ~31% CAGR (MarketsandMarkets). Vertiv holds ~23% of precision cooling globally.

Value-chain map — who captures the margin:

Bottleneck or commodity? → Near-bottleneck. NVIDIA's Blackwell/Rubin GPUs mandate liquid cooling, and Vertiv is an NVIDIA Solution Advisor partner (Introl). That's a real advantage. But watch the competition closing in: Schneider bought Motivair, Eaton bought Boyd Thermal (Mar 2026), and Vertiv itself is acquiring Ecolab's data-center cooling assets ($4.75B). The moat is "early lead + scale + NVIDIA relationship," not a patent wall. And the buyers are concentrated giants — a handful of hyperscalers can squeeze pricing over time.

7. Bull / base / bear (named views)

So what: the Street is near-unanimously bullish; the bear case is about valuation and cyclicality, not the business itself.

I could not find a reputable named institution with an outright bear/sell case and a below-market target — the coverage is lopsidedly bullish. That itself is a risk signal (crowded long).

Catalysts: continued 2×+ book-to-bill · Ecolab cooling acquisition closing/integrating · new hyperscaler capex announcements · Rubin-generation cooling wins.

What would prove the bull thesis wrong: book-to-bill drops toward ~1× · a hyperscaler pauses/cuts a capex plan · gross margin stalls as Eaton/Schniepder cut price · orders reveal Q4 2025 was a pull-forward spike.

§5 Practitioner decision layer

5a. Reverse expectations (multiple-and-consensus approximation — NOT a DCF)

So what: the price implies ~20% growth AND margin expansion sustained for ~5 years — plausible near-term, a stretch long-term.

At ~$290 and ~48× forward earnings, back-of-envelope: to "grow into" a normal ~25× multiple in ~4 years without the stock falling, earnings must roughly double — i.e. ~18–20%/yr EPS growth held for 4+ years, which needs revenue ~15–20%/yr and margins pushing toward the mid-20s%.

• Near-term (2026–27): plausible — the $15B backlog and 2.9× book-to-bill largely underwrite it.
• Sustained to 2029–30: stretched — few industrials hold 20% growth and 25% margins that long (base rate is low; growth usually decays as the law of large numbers bites and rivals add capacity).

5b. Variant perception

1. Consensus believes: AI data-center buildout is a multi-year secular wave and Vertiv is the prime power/cooling beneficiary — Strong Buy, $376.
2. This report's differentiated view: The business is consensus-right; the risk is that orders are cyclical and partly pulled-forward, so the 2.9× book-to-bill is a peak, not a run-rate — and the price has no cushion for that.
3. Evidence: +252% organic order growth and 2.9× book-to-bill are statistically extreme prints that historically mate-revert; competitors (Eaton/Boyd, Schneider/Motivair) are adding cooling capacity that will pressure the current supply-tightness.
4. Why the market may be wrong: the Street extrapolates a peak order quarter into a durable trend and pays 48× for it — classic late-cycle multiple on an early-cycle number.

Honest caveat: this is a valuation/cyclicality variant, not a claim the business is weak. On fundamentals I largely agree with consensus.

5c. Probability-weighted expected value

(scenario discipline, not a forecast; 12–18 mo)

EV = 0.35×376 + 0.45×300 + 0.20×170 = $131.6 + $135 + $34 = ~$300. ≈ current price. Read: at ~$290 the risk/reward is roughly balanced — the easy money was made on the way up; from here you're paid for the base case with real downside if orders cool.

5d. Pre-mortem — it's 2028 and VRT has halved

Three most plausible causes:

1. AI-capex digestion. Hyperscalers pause a build cycle → orders drop, book-to-bill falls under 1×. Early signal: any hyperscaler cutting/deferring 2027 capex guidance; VRT sequential order decline. De-risker: diversified customer base + services annuity.
2. Margin give-back from competition. Eaton (Boyd) and Schneider (Motivair) flood cooling capacity → price war. Early signal: gross-margin guidance flattening despite volume growth. De-risker: NVIDIA spec-partnership + installed-base lock-in.
3. Multiple compression alone. Business is fine but the market de-rates growth industrials from 48× to 25×. Early signal: peer multiples (ETN, SU) rolling over; rising rates. De-risker: earnings growing fast enough to "catch" the falling multiple.

5e. KPI driver tree & sensitivity (rough, labeled estimate)

Earnings ≈ Orders → Backlog → Revenue × Operating margin.

The single number the business runs on: book-to-bill ratio — above 1× the backlog (and future revenue) grows; below 1× it shrinks. At 2.9× it's flashing green; the whole thesis is whether it stays above ~1.3–1.5×.

5f. Scuttlebutt proxies

• Hiring trend: Vertiv has been expanding manufacturing capacity (new plants in the Americas/EMEA for power & liquid cooling) through 2025–26 — consistent with the order boom (management commentary, Q4 slides). Indicates: management believes demand is durable enough to add fixed capacity — a positive tell (but also raises the stakes if demand cools).
• Employee-review tone: not verified in this pass — unavailable/unassessed.
• Customer/supplier call mentions: hyperscaler capex guidance (Microsoft, Meta, Amazon, Google) has been rising through 2025–26, and NVIDIA's Blackwell/Rubin ramp explicitly requires liquid cooling — the demand signal from Vertiv's customers is strongly corroborative. Indicates: the order surge is grounded in real downstream capex, not channel-stuffing.

5g. Decision & review contract

• Watch, not chase. The business quality is high; the price already reflects it. A better entry (multiple <35× forward, ~$200–230) or one more quarter proving book-to-bill durability would flip this to a buy candidate.

Decision (Educational — not investment advice)

• Decision: Watch
• Conviction: Medium
• Core assumption: VRT's business is excellent, but at ~48× forward earnings the stock prices in sustained 20%+ growth with no digestion year — an asymmetry that favors patience over chasing.
• Review date: Q1 2026 earnings (est. late April 2026)
• Controlling question: Did book-to-bill stay above ~1.5× and did management raise (not just reaffirm) full-year order guidance?
• Review trigger: Buy interest if forward P/E falls below 35× ($210–230) OR two consecutive quarters of book-to-bill >1.5× confirm the order surge is a run-rate, not a spike.

📚 What this company teaches

1. "Picks and shovels" beats betting on the winner. You don't need to know whether OpenAI or Anthropic or Google wins the AI race — Vertiv sells the power and cooling to all of them. When a value chain has one obvious bottleneck (§0 lens), the arms-dealer often has a better risk/reward than the prospectors.
2. The margin slope is the pricing-power tell. Vertiv's operating margin going 4%→18% in three years told you demand exceeded supply before the headlines did. Watch the direction of margins, not just the level.
3. A great business and a great stock are different questions. VRT is unarguably a great business; at 48× forward earnings, whether it's a great investment depends entirely on the price you pay. Separating "is this good?" from "is this cheap?" is the core discipline.

Self-test question: Vertiv's Q4 book-to-bill was 2.9×. Explain in one sentence why a falling book-to-bill (say to 1.1×) could still mean revenue is growing — and why the stock might drop anyway.

(Answer: book-to-bill >1× means orders still exceed shipments, so backlog and future revenue keep growing — but a sharp fall signals decelerating demand, and a 48× multiple built on acceleration can compress even while revenue rises.)

📖 Glossary

• UPS (uninterruptible power supply): battery-backed power gear that keeps servers running through outages.
• Precision / liquid cooling, CDU: systems that remove heat from dense server racks; a CDU pumps coolant directly to the chips.
• Hyperscaler: a giant cloud operator (Amazon, Microsoft, Google, Meta) that builds data centers at massive scale.
• Colocation (colo): a company that builds data-center space and rents it to others.
• Backlog: the dollar value of orders received but not yet shipped/recognized as revenue — future revenue already contracted.
• Book-to-bill: orders received ÷ revenue shipped in a period. >1× = backlog growing; <1× = shrinking.
• Organic growth: growth excluding acquisitions and currency — the "real" underlying rate.
• P/E (price-to-earnings): share price ÷ earnings per share; how many dollars you pay per $1 of annual profit.
• Forward P/E: same, using next year's expected earnings.
• EV/EBITDA: enterprise value (market cap + debt − cash) ÷ EBITDA; a debt-neutral whole-business price; the standard for capital-heavy firms.
• EBITDA: earnings before interest, tax, depreciation & amortization — a rough cash-earnings proxy.
• FCF (free cash flow): operating cash flow − capex; the cash left over after keeping the business running.
• Cash conversion (FCF/EBITDA): how much of accounting profit becomes actual cash; ~89% is excellent.
• Net debt/EBITDA: debt minus cash, divided by EBITDA; a leverage/fragility gauge (>3× = stretched).
• ROIC: return on invested capital; profit generated per dollar of capital employed; above cost of capital = value creation.
• CAGR: compound annual growth rate — the smoothed average yearly growth.
• Adjusted EPS: earnings per share excluding one-off items (like warrant marks); often more representative than GAAP.
• Reverse expectations / DCF: working backwards from the price to see what growth it implies (reverse); a DCF projects future cash flows and discounts them to today.

Sources: SEC EDGAR (primary, FY2020–25 US-GAAP) · Vertiv Q4 2025 release · Investing.com Q4 slides · MarketBeat consensus · Morgan Stanley/JPMorgan targets · WallStreetZen valuation · MarketsandMarkets cooling TAM · Introl competitive comparison. Market prices as of 2026-07-24 (~$290). Not investment advice.