Vertiv Holdings

Refresh delta (since 2026-07-25)

What changed

1. Q2 2026 printed split: a clean EPS/margin/cash beat on a revenue miss. Net sales $3,274.3M, +24.1% YoY (organic +18%, M&A +5%, FX +1%) against a company guide midpoint of $3,350M and Street consensus of ~$3.37–3.39B — short of both. But adjusted diluted EPS $1.52, +60%, above the top of the guide range ($1.43) and ~6% above consensus; adjusted operating margin 22.6%, +410bps, vs a 21.2% guided midpoint; adjusted FCF $925M, +234%.
2. FY2026 guide raised across every metric — and raised above consensus. Net sales $13.8–14.2B (from $13.5–14.0B), adj EPS $6.65–6.75 (from $6.30–6.40), adj operating profit $3,285–3,365M, adj operating margin 23.3–24.3%, adj FCF $2.4–2.6B (from ~$2.2B midpoint — the largest proportional raise in the guide). Q3 guided to $3.65–3.85B and $1.77–1.83 adj EPS. ``
3. The stock fell hard anyway, and the Street cut targets faster than the stock fell. VRT dropped roughly 10% on the print ; it trades **$273.15** on 2026-08-10 , −5.9% vs the prior refresh's $290.36. Average analyst PT fell from ~$380 to $338.15 on a broad round of cuts — Evercore $425→$375, Citi $414→$358, Goldman $352→$301, Oppenheimer $353→$325, Baird $370→$320. Implied consensus upside narrowed to +23.8%, from ~+31% last pass — a reversal of the prior refresh's direction, where the stock fell and targets held.
4. Deferred revenue doubled — the demand proxy fired. Current deferred revenue $3,633.7M at 2026-06-30 vs $1,814.7M at 2025-12-31, +$1,819.0M (+100.2%) . The CFO confirmed on the call these are **"advanced deposits on orders"** plus milestone deposits on larger projects, and called it an indicator of "strong commercial performance" . With backlog still undisclosed, this is now the best orders read available.
5. The revenue miss was internal, not demand. Management attributed it to "minor timing shifts… primarily due to temporary supply chain congestion and multi-phased project execution as deployments scale in size and complexity." Albertazzi framed it as "complexity and interdependencies" on large SmartRun/OneCore projects, conceding "there is a learning curve" and that "these are the first very large projects with this level of complexity." H2 guidance "already bakes in some level of continued friction" ``.
6. Three standing open items closed (detail in §Lens 9/10): the May 2026 Investor Day multiyear framework is now sourced (20–22% five-year revenue CAGR, 27% adj operating margin ambition by 2030); the June Form 4 clusters are grants, not sells — all code-A acquisitions at $0, no open-market disposals; APAC's Q1 timing miss fully reversed (+26% organic in Q2 vs +12% in Q1).

What held

The structural thesis stands, with one amendment. Power and thermal remain the binding constraint on the AI buildout, Vertiv remains the integrated arms dealer, content-per-rack still rises with liquid cooling and 800VDC, and the balance sheet is now effectively unlevered (net debt $129.2M; net cash of $170.8M including short-term investments) . Gross margin expanded to **37.7% from 34.0%** . Demand commentary strengthened, not weakened — pipeline "accelerating" in the Americas, EMEA "further strengthening," APAC "broad-based."

The amendment: a second binding constraint has appeared, and it is inside the company. The prior dossier located the external bottleneck at grid interconnection — the customers' problem, which Vertiv had begun selling into via Bring-Your-Own-Power. Q2 adds an internal one: Vertiv's own ability to coordinate multi-factory, multi-phase integrated projects at scale. That constraint sits on exactly the product line that carries the content-expansion story (SmartRun / OneCore). The bull case and the execution risk are no longer separable.

Latest earnings vs consensus, guidance, balance-sheet flags, market reaction

The quarter against every yardstick

Gross profit $1,234.9M, 37.7% of sales vs 34.0% LY . GAAP operating profit $637.9M (19.5%); the $100M GAAP-to-adjusted gap is amortization of intangibles $73.7M (vs $46.9M LY, +57%) plus the PurgeRite earnout revaluation (below).

Segments — the regional picture inverted from Q1 ``:

EMEA's "return to growth" is a foreign-exchange artifact and the filing says so plainly. Sales rose $8.0M while FX contributed approximately $13.7M — so organic ex-M&A was negative, and the call confirmed **−2% organic**. This is a real improvement from Q1's −29%, but the return to organic growth has now been promised for the back half twice. Note the margin, though: EMEA delivered +380bps on declining organic volume, on "project execution improvements" — the operating fix is landing ahead of the volume recovery.

APAC is the clean win: Q1 came in below its own quarterly guide on timing; Q2 delivered +26% organic and +270bps of margin. That open item is closed.

Balance-sheet flags (2026-06-30 vs 2025-12-31) ``

Green:

• Net debt $129.2M (total debt $2,939.8M less cash $2,810.6M); including $300.0M of short-term investments, Vertiv is net cash $170.8M. Effectively unlevered against a $3.3B FY adj operating profit guide.
• Deferred revenue $3,633.7M, +100.2% — customer money in the door.
• AR $3,750.3M, +20.6% — growing slower than revenue (+24%). DSO ~104 days ``. No receivables-stuffing signature.
• Interest expense now trivial ($17.4M net in Q2; H1 was a net $13.0M after a swap settlement), following a $2.1B issue / $2.076B repayment refinancing that cost $6.2M in extinguishment.

Amber — watch these:

• Inventory $2,522.7M, +73.2% in six months against revenue +26.7%. DSI ~113 days ``. Management guided this as a deliberate capacity-ramp build and the Q1 CFO commentary pre-warned it — so it is disclosed, not a surprise. But a 73% inventory build is only prudent if the H2 revenue ramp lands; if it slips, this is where it shows up first.
• Goodwill $2,283.3M + intangibles $1,800.8M = $4,084.1M, which is 25.7% of total assets and 85.8% of book equity ``. Not a near-term issue on this cash generation, but the M&A cadence is accelerating and this ratio only moves one way.
• Contingent consideration is now a live GAAP line. PurgeRite carries $206.1M of accrued earnout (against up to $250M payable), and the fair-value revaluation ran a $28.8M loss in Q2 / $62.0M in H1 through "Other operating expense" ``. Read it correctly: a loss here means PurgeRite is beating its earnout targets. It is simultaneously good news about the asset and a growing drag on GAAP earnings and on the GAAP-to-adjusted wedge (~$0.25/share in Q2 alone).
• PurgeRite was not a tuck-in. Total consideration $1,138.3M net of cash ``. The base dossier's framing of the M&A programme as "technology-led tuck-ins" is wrong for this deal by an order of magnitude and should be corrected. Three further acquisitions closed within Q2 for $331.8M combined — liquid cooling, heat-exchange/heat-rejection, and custom structural fabrication.

The earnings-quality point that matters most. H1 operating cash flow was $1,866.6M, of which the deferred-revenue swing supplied $1,822.7M. Strip that single line and H1 free cash flowFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. is roughly negative $245M ``. This is not an accusation — advance deposits are real customer money against real orders, and on a project business they are the correct way to fund a ramp. But it does mean the headline FCFFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. is currently an orders indicator, not a profitability indicator, and it will not repeat at this magnitude once the deferred balance stops doubling. The raised FCFFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. guide ($2.4–2.6B) leans on it.

Market reaction

Down ~10% on the print ; other outlets reported "more than 9% premarket" . Source conflict flagged and resolved: one low-quality outlet reported −26% . That figure is inconsistent with the observable price path — $290.36 on 2026-07-24 to $273.15 on 2026-08-10 `` is −5.9% total, which a −26% single-day drop would require a ~+26% recovery in eight sessions to reconcile. Use ~−10%; the −26% figure is not credible.

---

Call-sentiment shift vs prior calls

Comparing the Q2 2026 call to Q1 2026 (the only other transcript on the shelf) plus the Q4/Q3 2025 tone carried in the base dossier:

What is unchanged or stronger:

• Demand language escalated. Q1 said "pipeline generation is robust." Q2 says pipeline momentum is "accelerating… in all regions and across all customer categories" — hyperscalers, colos, enterprises, and a new cohort management called "new cloud players" ``. Americas pipeline "actually accelerating." EMEA "further strengthening."
• Cote's framing held: "moving at the speed of technology," investing ahead of demand, "strengthening pipelines globally" ``.
• Capital deployment posture unchanged and confident: M&A pipeline "is active" (CFO Chamberlin), capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. pushed to the high end of the range, 4% of 2026 sales — Johor (Malaysia), large Americas plant expansions, EMEA chiller capacity ``.

What shifted — and this is the whole story of the reaction:

• A new vocabulary of internal friction entered the call. Q1 had no equivalent: "minor timing shifts," "temporary supply chain congestion," "multi-phased project execution," "complexity and interdependencies," and most tellingly Albertazzi's "there is a learning curve" and "these are the first very large projects with this level of complexity." Chamberlin: "When it becomes a larger supply chain, it becomes more confounding" when multiple factories depend on each other ``.
• Management hedged its own back half. Asked directly about H2 prudence (Chris Snyder, Morgan Stanley), management said "we have assumed some of that congestion continues" and that they remain "prudent in our guidance" ``. Q1 carried no such hedge. This is the correct disclosure behaviour — and it is also an admission that the friction is not modelled as one-quarter noise.
• EMEA's promise was re-issued rather than delivered. Q1 guided EMEA to flat for the full year with a back-half-loaded return to growth; Q2 delivered −2% organic and re-promised H2 growth. Improving trajectory, but a second deferral of the same commitment.

Net sentiment read: demand conviction up, delivery conviction down, and management was candid about the second rather than papering it. The Q1 call was a pure beat-and-raise with an APAC asterisk. The Q2 call was a beat-and-raise with a structural execution caveat attached to the flagship product line. The market repriced the caveat, not the demand.

---

Catalysts and ±5% moves

Moves since the last refresh:

Forward catalyst calendar:

• {type: EARNINGS, what: Q3 2026 print — the single test of whether the timing shifts reversed. Guide is $3.65–3.85B / $1.77–1.83 adj EPS / 24.0–25.0% adj op margin. Watch: does revenue land at or above midpoint, does EMEA finally post positive organic, does deferred revenue keep building, when: late Oct 2026}
• {type: LAUNCH, what: 800VDC — under customer validation in 2026, deployment 2027. Named deployments now include Taiwan's first AI datacenter with NVIDIA GB300 (with VisionBay.ai / Foxconn) and "the world's first AI data center adopting 800 VDC architectures at the rack and pod level," plus NVIDIA Vera Rubin. Management believes Vertiv's 800VDC content sits toward the HIGH end of the $3.25M–$3.75M per megawatt range, when: validation 2026 / ship 2027} ``
• {type: LAUNCH, what: production-grade digital twin + gigawatt-scale reference architectures for the NVIDIA Omniverse DSX Blueprint, when: announced 2026} ``
• {type: OTHER, what: BACKLOG DISCLOSURE — undisclosed for two consecutive quarters. Any refreshed figure is a first-order catalyst in either direction, when: unknown, possibly FY2026 10-K only}
• {type: MACRO, what: hyperscaler FY2027 capex guidance, when: late-2026/early-2027}
• {type: OTHER, what: M&A — pipeline "active," net cash balance sheet, Investor Day targets 1–2x leverage vs ~0x today. A large deal is live optionality, when: ongoing}

---

Forward framework (Investor Day, now sourced — prior open item CLOSED)

The prior refresh flagged the May 19–20 2026 Investor Day (Greenville, SC) multiyear outlook as "likely the single biggest missing piece for the forward-projection lens." It is now sourced, though secondary only — I could not obtain the primary deck ``:

• Five-year revenue CAGR raised to 20–22%
• Adjusted operating margin ambition 27% by 2030 (vs 23.8% guided for 2026)
• Adjusted FCF conversion 95–100% through 2030
• Capex framework 3–4% of revenue annually (2026 tracking to the high end, 4%)
• Target leverage 1–2x — against roughly zero today
• ~$24B of available deployment capital after dividends; $28B targeted capital deployment by 2030

---

Red flags, governance, insiders, regulatory

Insider activity — prior open item CLOSED, and the resolution is benign. I parsed the June Form 4 clusters directly from EDGAR XML. All eleven filings since 2026-06-01 are transaction code A (acquisition) at $0.00 per share — grants, 401(k) match and dividend-equivalent accruals, not open-market purchases or sales. Named: Albertazzi (9.53 sh), Gill (4.58), Karlborg (3.55), Sanghi (3.38), Poncheri (2.74), Mikkilineni (3,076 — a post-AGM director grant) ``. No open-market disposals in the period. The base dossier's "65 sells: 0 buys over six months" stands as a historical read of an earlier window but has not extended — the selling has stopped, or at least paused, through the Q2 print.

Governance — carried, unchanged. The 2026-06-17 AGM withhold votes (van Dokkum ~46%, Fradin ~30%, DeAngelo ~25%, ~12% against say-on-pay) remain the standing yellow flag ``. Nothing new since.

Regulatory — clean, re-verified today. regulatory-findings.md was regenerated 2026-08-10: zero SEC Litigation Releases, zero AAERs over 2021-08-10 → 2026-08-10 ``. Non-SEC agencies remain unsearched; no new litigation surfaced in the Q2 10-QForm 10-QThe quarterly version of the annual report. Lighter, and not audited. or the July 8-KForm 8-KA filing for something that happened between reports and matters enough to tell shareholders now..

New this quarter (see Lens 5 for detail): the intangibles-to-equity ratio (85.8%), the contingent-consideration line now running $28.8M/quarter through GAAP operating expense, and the widening amortization drag (+57% YoY) as the deal cadence builds.

EDGAR completeness check: only two filings since 2026-07-20 — the 10-Q and the earnings 8-K, both 2026-07-29 ``. Nothing missed.

---

Three-year forward EPS path

Model integrity — read this before any number below

our model was re-run after the CSV update. The workbook produces no usable output and I will not cite it. our model reports: "No computed values. The workbook has 51 formulas and no cached results." Two further seed gaps are flagged by the builder itself: only 2 quarters of financials (LTM figures are partial) and, because there are fewer than 8 quarters, revenue growth seeded at a generic 8% — which for a company guiding +37% is meaningless. Share price is blank; the balance check is uncomputed. One improvement: Opening balance sheet sourced: yes, now that the balance-sheet block is populated from the Q2 and Q1 10-Qs.

Note also that our figures was empty before this pass — the two quarters I added (2026-Q1, 2026-Q2, both primary from the 10-Qs) are the entire history. Everything below is my own arithmetic, shown in full, not model output.

Anchors

• FY2026 company guide: revenue $14.0B mid, adj operating profit $3,325M, adj EPS $6.70 ``
• Implied adj-operating-profit → adj-net-income conversion: 0.791 ``, absorbing tax and interest
• Diluted shares 392.7M at Q2 ``
• Investor Day frame: 20–22% revenue CAGR, 27% adj op margin by 2030 ``

The back-half arithmetic that drives all three cases

This is the crux, and it is pure subtraction from the company's own guide:

• Implied Q4 revenue $4,326M ``
• That is +15.4% sequential on Q3, and +32.1% above the Q2 quarter that just missed
• Implied Q4 adj operating profit $1,118M `` → Q4 adj operating margin 25.8%, i.e. +320bps above the 22.6% just delivered
• Implied Q4 adj EPS $2.21 ``, +22.8% sequential
• H2 revenue $8,076M vs H1 $5,924M, +36.3% half-on-half; H2 adj EPS $4.01 vs H1 $2.69, +49.1%

The guide has cushion on revenue but not on EPS. If Q4 lands 2.3% below its implied midpoint — exactly the way Q2 did — FY revenue is ~$13.9B, still inside the $13.8–14.2B range. The EPS low end of $6.65 is the tighter constraint: it needs H2 EPS of $3.96 against H1's $2.69.

Three-year paths (FY2027 / FY2028 / FY2029)

All ``, arithmetic shown; FY2026 anchored to guide.

Base — Investor Day CAGR delivered mid-range, margin marching toward the 27% ambition:

Bull — top-end CAGR sustained, 800VDC lifts content per megawatt to the high end of $3.25–3.75M/MW, margin beats the 2030 ambition early, buybacks hold shares flat:

Bear — the execution friction proves structural rather than transitional, the integrated-project ramp is coordination-limited, growth decelerates to mid-teens then low-double-digits, margin stalls near current levels on tariffs and complexity costs, dilutionDilutionIssuing new shares, so each existing share owns a smaller slice of the same company. resumes:

What the market is paying, at $273.15 ``

The de-rating continued through a raise. Last pass: 45.7x on a $6.35 guide. This pass: 40.8x on a raised $6.35 → $6.70 guide. The stock is cheaper on a higher number. Market capMarket capitalisationThe share price multiplied by the number of shares. What the market says the equity is worth. $105.16B on 384.99M shares ; 28.1% below the $379.94 high .

Read the bear column carefully. In the bear case you are paying 31x FY+3 earnings three years out — i.e. the current price only works if something close to base or bull happens. That is the cost of a 40x entry, and it is the honest framing of the risk.

---

Bull vs bear, and the pre-mortem

Bull

Demand is not in question and the evidence is on the balance sheet, not in the narrative: customers wired $1.8B of incremental deposits in six months. Vertiv raised every line of guidance above consensus in the same breath as the miss — companies with demand problems do not do that. Margin is compounding faster than revenue (+410bps adjusted, +375bps gross), APAC has inflected, and EMEA is delivering +380bps of margin before its volume recovery, which means the operating fix is already in place and the volume is the free option. The balance sheet is net cash with an Investor Day mandate to lever to 1–2x, giving several billion of deal capacity into a consolidating thermal/power supply chain where Vertiv is the natural acquirer. 800VDC ships into the Rubin generation with management "pretty convinced" its content sits at the high end of $3.25–3.75M per megawatt. And the multiple has now compressed from ~82x (June) to ~73x to 40.8x on forward guide while the earnings estimate went up — a genuine de-rating without an earnings break.

Bear

You are paying 40.8x a guide whose back half requires a Q4 that is 32% larger than the quarter Vertiv just failed to deliver, at a margin 320bps higher, from a company that has just told you its constraint is its own project coordination. The miss was not a demand air-pocket you can wait out; it was "the first very large projects with this level of complexity" — and those projects (SmartRun, OneCore) are precisely the integrated, margin-accretive, content-expanding line the entire content-per-rack thesis rests on. The growth engine and the failure mode are the same product. Meanwhile the visibility claim is unauditable: backlog has gone undisclosed for two straight quarters and every secondary source is still recycling a $15.0B figure from a 10-K filed 2026-02-13. Headline FCF is flattered by a deferred-revenue swing that cannot repeat at this rate — strip it and H1 FCF is about −$245M. Inventory is up 73% in six months against a ramp that just slipped. Intangibles are 86% of book equity and rising with a deal cadence that now includes a $1.14B acquisition whose earnout is being marked up through the income statement. And the Street has begun marking the name down faster than the price is falling — five houses cut targets, the average went $380 → $338.

Pre-mortem — it is 2027-07 and VRT is 40% below here. What happened?

1. Most likely — the ramp slipped again and the multiple, not the earnings, did the damage. Q4 2026 or Q1 2027 missed revenue on the same coordination complexity. Management conceded the constraint is structural to integrated projects rather than transitional. EPS still grew 30–40%; the market simply stopped paying 40x for a business whose quarters it cannot predict and re-rated toward 25x. Nothing in the thesis broke — the price of the thesis changed. This is the path the Q2 reaction was already rehearsing.
2. The demand cycle turned and the visibility defence was not there. A hyperscaler signalled capex digestion in late 2026. Vertiv's answer would have been the backlog — but it had not disclosed one in three quarters, so there was nothing to point at, and the name traded with the complex instead of above it.
3. Capital allocation. The Investor Day 1–2x leverage target was executed into a hot market. A large deal at a full price, on top of PurgeRite's rising earnout, pushed goodwill and intangibles well past book equity; the first integration stumble produced an impairment and a credibility reset on the "technology-led M&A" story.
4. Tail — margin was borrowed, not earned. Management flagged "favorable price cost in 2026, including the current impact of tariffs." If a chunk of the +410bps was price capture ahead of cost, it reverses when the supply chain loosens or tariffs step up, and the 27%-by-2030 ambition quietly disappears.

What would falsify the bear case fastest: a Q3 print at or above the $3.75B midpoint with EMEA positive organic, deferred revenue still building, and a disclosed backlog figure.

---

Devil's advocate

Against the bulls: the visibility premium rests on a number Vertiv will not print.

The entire justification for a 40x multiple on a cyclical hardware business is that this cycle is visible — that a $15B backlog with 12–18 month delivery windows converts a capital-goods company into something closer to a subscription. But Vertiv does not disclose orders, does not disclose book-to-bill, and has now declined to disclose a backlog dollar figure for two consecutive quarters. The last hard number was filed 2026-02-13 and is six months stale, yet it is still being quoted as current by essentially every secondary source I read this pass — including sell-side-adjacent coverage published after the Q2 print.

The honest reading is not that management is hiding a bad number. The deferred-revenue doubling is strong, unambiguous, primary evidence that orders are excellent, and it is the CFO who volunteered the explanation. The point is narrower and harder: you cannot audit the claim you are paying the premium for. A company sitting on a genuinely accelerating $15B+ backlog has every commercial incentive to print it into a quarter where the stock just fell 10% on a revenue miss. It chose not to. That leaves two readings — a shift to annual-only disclosure, or a composition/duration story less flattering than the headline — and no way to distinguish them from outside. At 40x, the burden of proof sits with the premium, not with the sceptic.

Against the bears: the miss was $76M on a $3.35B guide, and the guide went up.

The market treated a 2.3% shortfall against midpoint — still inside the guided range — as a thesis event, on a quarter that beat on adjusted operating profit, beat on margin by 140bps, beat EPS by 6%, tripled free cash flow, and was accompanied by a raise above consensus on every single line including a $300M raise to the FCF guide. If the demand signal were deteriorating, the deferred-revenue balance would not have doubled and the FY guide would not have gone up. A 10% drawdown on that combination is a multiple-compression event in a crowded AI-infrastructure trade, not new information about Vertiv.

Against my own bear framing: I have leaned on "the growth engine and the failure mode are the same product." That is real but it cuts both ways — an execution learning curve on genuinely novel gigawatt-scale integrated deployments is the signature of a company doing something competitors cannot yet attempt. Schneider and Eaton are not missing quarters on OneCore-class complexity because they are not selling OneCore-class systems. The complexity Vertiv is struggling to coordinate is the moat, and first-mover friction is the price of it. Whether that is generous or accurate is exactly what Q3 tests.

Against everyone, including me: the honest position on this name today is that one quarter cannot settle it. Q2 was the first miss in the covered period, management's explanation is specific, testable and pre-committed ("we have assumed some of that congestion continues"), and Q3's guide is the test they set themselves. Anyone converting a single print into conviction in either direction — up or down — is trading the narrative, not the evidence.

---

Carried by reference (not re-derived this pass)

• Business overview. Unchanged, see the previous dossier. Picks-and-shovels arms dealer of the AI buildout; power + thermal + integrated infrastructure across three geographic segments.
• Supply chain. Unchanged, see the previous dossier and §3 of the previous dossier. NVIDIA co-design; grid interconnection as the external binding constraint; Bring-Your-Own-Power (Caterpillar, Oklo, C Power Energy) as the workaround Vertiv now sells into.
• Moats. Unchanged in substance (integration + installed base + co-design proximity, 3–5yr not 20yr), but see Lens 13 for the argument that the Q2 execution friction is evidence for the moat rather than against it.
• Segment structure. Structurally unchanged; Q2 figures refreshed in Lens 5.
• Comps. Unchanged framework (premium to Schneider / Eaton / nVent / Modine). Not re-sourced this pass — no fresh peer multiples.
• Management questions. Unchanged, see the previous dossier. Add one: what changed in the backlog disclosure policy, and will a figure be given before the FY2026 10-K?

---