Vertiv is the widest-portfolio supplier of the power and cooling that AI factories run on, and it is growing 31% organically at 24% margins with customers paying deposits up front. At $244 the market is paying for management’s own 2030 framework delivered at the top end of its range. Our model lands in the same place. That is a fairly valued stock with a deep-tail bear case, not a cheap one and not a bubble. Our band is $210–275.
The price and the model agree, which is unusual for this name. Our modal base case is $238 against $244 spot. The reverse DCF shows the market pricing 2025–30 revenue growth of roughly 25% a year with margins peaking near 28% — the top end of management’s stated 20–22% growth and 27%-plus margin framework. Given that the November 2024 framework (12–14% growth, 25% margin by 2029) was beaten by roughly double, “top of range” is a defensible modal outcome, not a stretch.
Customer deposits have replaced the backlog as the demand signal — and they doubled in six months. Vertiv stopped disclosing orders, book-to-bill and backlog after the record $15.0bn print in February 2026. What is still visible on the balance sheet is deferred revenue: $1.81bn in December 2025, $3.63bn in June 2026. Advance and milestone payments on very large projects funded 115% of first-half free cash flow. That is a genuine sign of strong intake and pricing power; it is also a working-capital tailwind that reverses the moment intake slows.
The skew is asymmetric in dollars and roughly symmetric in probability. A 2028 digestion year (revenue −8%, margin 20%) takes the equity to about $106, a 57% fall. Bull and super-bull cases reach $317 and $396. At our 25 / 45 / 25 / 5 weights the probability-weighted value is $233 — two per cent below the base, so the distribution is very slightly negatively skewed at this price. The stock’s 35% fall from its May peak has removed the premium, not created a discount.
Run the model in Section 02 backwards and $244 implies 2030 revenue of roughly $30.7bn and a terminal adjusted operating margin near 27.8% — or, equivalently, our base path discounted at about 9.35% instead of 9.5%. Vertiv’s own May 2026 framework is $26bn of organic revenue and “27%-plus”. The stock is priced for that framework to be met at the top of the range, as this management team has done before. The bear case — an AI-capex digestion year in 2028 — is worth about $106 and carries a real 25% weight. The bull case is $317. Weighted, the equity is worth roughly what it costs.
Backlog is no longer disclosed. The last print, $15.0bn at 31 December 2025, covered 107% of this year’s revenue guide. Deposits already collected cover a quarter of it. Each line below is scaled against FY2026 guided revenue of $14.0bn.
| Measure | Amount | Of FY2026 guide | What it actually is |
|---|---|---|---|
| Last disclosed backlog 31 Dec 2025 | $15.0bn | 107% | Reported once, in February 2026, alongside organic orders up 252% and a book-to-bill near 2.9x. Not updated since. Nine months stale |
| FY2026 revenue guide midpoint | $14.0bn | 100% | Raised from $13.75bn in July; 31% organic growth. Requires H2 revenue of $8.08bn, up 45% year on year |
| H1 2026 revenue booked | $5.92bn | 42% | Reported. The second half has to deliver the remaining 58% |
| Customer deposits held deferred revenue, 30 Jun 2026 | $3.63bn | 26% | Cash already received from customers against undelivered work — collected at order, at design freeze and at delivery milestones on the largest projects. The only remaining public proxy for order intake |
| Deposits held six months earlier 31 Dec 2025 | $1.81bn | 13% | The doubling in two quarters is the demand signal this report leans on, and the working-capital tailwind it discounts |
Customers do not prepay billions to a supplier they intend to switch away from. Read as an intake proxy, deposits point to order growth well ahead of revenue growth in the first half of 2026. Read as cash flow, they are an advance, not earnings: $1.82bn of deferred-revenue growth against $1.58bn of first-half free cash flow means the entire cash generation of the half, and then some, was customers’ money. When intake stops accelerating the tailwind becomes a headwind and cash conversion mean-reverts toward 85–90%. The model assumes exactly that; it does not assume the deposits reverse.
The valuation question reduces to a single variable: whether the 2025–2030 period contains a digestion year. If it does not, the stock is worth $240–320; if it does, $100–150. We weight the two outcomes 75 / 25. The dated tests that would move the rating are the Q3 print in late October — does the “timing” revenue return, and does Q4 guidance hold at $4.3bn? — hyperscaler 2027 capex guidance in late January, and whether the February 2027 release restores orders and backlog disclosure.
One more piece of context on the price itself. Vertiv fell 17.3% on 29 July on a 2% revenue miss and has fallen 35% from its May peak. That drawdown is large enough to feel like a valuation reset, and it is not one: at $244 the price sits at the 52nd percentile of our band. The stock being cheaper than it was is a statement about the past, not about the price.
2025–30 revenue CAGR 24.1% · 2035 free cash flow $8.2B · implied 2035 EV/EBITDA 10.6x · terminal value is 63% of enterprise value.
This is an unlevered discounted cash flow on a ten-year explicit path, and the shape of the path is the whole argument. Nine years run from 2027 to 2035 on top of the company’s own 2026 guidance, which is taken at face value: $14.0bn of revenue at a 23.8% adjusted operating margin, with the roughly $0.92bn of 2026 free cash flow still to arrive after the 17 September valuation date charged at the $2.5bn guide. Free cash flow to the firm is adjusted operating profit after 24% tax, less share-based compensation charged as cash at 0.4% of revenue, less capital expenditure net of depreciation, less the working capital that growth absorbs. Terminal value is a Gordon perpetuity on 2035 cash flow. On base settings that is 63% of enterprise value, which is the honest shape of any ten-year model on a company growing 30% today, and it is why the sensitivity grids in Section 03 matter more than the centre cell.
The first two sliders are the thesis. 2027 growth of 30% is backlog conversion plus new capacity — the Q4 2026 exit run-rate is already $17.3bn annualised — and it is deliberately set below the 34–36% Q3 organic guide, which is where the mega-project execution risk is charged. 2028 is the year a capex digestion would land if it lands: the base case has +22%, the bear case −8%. Note what happens when you move only that slider: the value barely changes, because the path re-anchors to the 2030 endpoint and the lost year is recovered. A digestion year is expensive only if the base it leaves behind is permanently lower — which is exactly what the bear card does, taking 2030 revenue to $16.6bn and 2035 to $21.2bn. The distance between $238 and $106 is the endpoints, not the dip. The 2030 and 2035 endpoints define how the path matures: $30.1bn in 2030 is management’s organic $26bn plus roughly $2bn of acquired revenue and $2bn of out-performance, and $42.6bn in 2035 is growth fading from 11% to 4% as data-centre capex normalises to a replacement-plus-growth market.
Two margin settings, and both are above what the company has ever earned. The margin path ramps from this year’s 23.8% onto the 2030 setting with the shape of the base case (25.3%, 26.3%, 27.0%, 27.5%) and then fades linearly onto the terminal setting by 2035. Management targets 27%-plus in 2030 with named levers — 200 basis points of operating leverage, 150 of productivity, 50 of commercial — and reached its 2029 target two years early, so the base takes 27.5% and fades 50 basis points for competition and mix. Capital expenditure, depreciation and working capital are held fixed across every scenario because the company describes them that way: capex 3–4% of sales through 2030 fading to 2.5%, depreciation rising toward 1.8%, and working capital absorbing 1% of incremental revenue while deposits fund it and 4% once they do not. The model lets the deposit tailwind fade rather than reverse.
Where each risk is priced, once. Cyclicality, execution, competition and the deposit unwind live in the scenario paths and their weights, never in the discount rate. The 9.5% rate is a 4.85% ten-year Treasury, a 4.25% equity premium and an asset beta of about 1.1 for a global electrical-equipment maker; with a net cash balance sheet it is effectively the cost of equity. The shares’ own beta is above 2 and is not used: it measures how the stock trades, and charging cyclical risk in the rate after charging it in the bear scenario would count it twice. Readers who disagree have the 10.0% and 10.5% rows of the grid, which is why they are printed. Dilution is priced once too, in the share count and the compensation charge; growing the count by 3m a year instead gives the same answer within $2.
Reading the inputs. The two revenue endpoints and the terminal margin are the settings that decide the answer. Each point of terminal margin is worth roughly $7 a share; each billion of 2030 revenue, carried through to 2035 at the base shape, roughly $8. Enterprise value is $93.1bn on base settings against a $96bn market capitalisation, with $171m of net cash — $3.11bn of cash and short-term investments against $2.94bn of fixed-rate notes — added back. The UtilityInnovation Group purchase is treated as value-neutral: $1.45bn of cash out at closing equals the asset acquired at roughly 13 times 2027 EBITDA.
The number above is not a fact about Vertiv. It is what these assumptions produce, and the most consequential of them is a judgement about the industry rather than the company: whether the hyperscalers pause. No $14bn industrial has sustained 25% organic growth for a decade, and every capex supercycle on record has contained a year in which the customers’ spending fell. The rating survives a wide range of views on margin and rate, and does not survive a digestion year — which is the point of publishing the bear case at 25% rather than hiding it in a discount rate.
Each card loads into the model above and prints that path’s discounted value: $107, $238, $316 and $395. The published scenario values are $106, $238, $317 and $396, from the report’s own spreadsheet; the page model reproduces them within a dollar and is reproducible from the settings on the cards. Two notes on construction. First, the bear case carries a 25% weight because a base case that credits management’s framework at the top of its range has to hand its failure branch real probability, not token probability: every capex supercycle in history — telecom in 2000, shale in 2014, crypto in 2021 — contained a year in which the customers’ spending fell 20–40%. Second, the bear is floored by the services annuity rather than by a hardware-cycle analogy: 20%-plus margins through the trough and a $106 value rather than the sub-$80 a pure component-supplier comparison would produce. Weighting the four at 25 / 45 / 25 / 5 gives $233, two per cent below the base — the distribution is very slightly negatively skewed at this price.
| Assumption | Bear | Base | Bull | Super-bull |
|---|---|---|---|---|
| 2026 revenue / margin | $14.0bn / 23.8% in every path — the company’s 29 July guidance midpoints, taken at face value with the H2 slack management described | |||
| Revenue growth 2027 / 28 / 29 / 30 | +14 / −8 / +4 / +9% | +30 / +22 / +18 / +15% | +36 / +30 / +24 / +18% | +40 / +34 / +28 / +22% |
| 2030 revenue | $16.6bn | $30.1bn | $36.2bn | $41.0bn |
| Revenue growth 2031–35 | 7 / 6 / 5 / 4 / 3.5% | 11 / 9 / 7 / 5 / 4% | 14 / 11 / 8 / 6 / 5% | 17 / 13 / 10 / 8 / 6% |
| 2035 revenue | $21.2bn | $42.6bn | $55.0bn | $68.2bn |
| Adj. operating margin 2027 / 2030 / terminal | 23.5 / 22.2 / 23.0% | 25.3 / 27.5 / 27.0% | 25.8 / 29.0 / 28.5% | 26.2 / 29.8 / 29.5% |
| Tax rate | 22% in 2026; 24% thereafter, per company guidance for 2027–30. Applied to adjusted operating profit; the amortisation tax shield is ignored as small | |||
| Capex / depreciation, % of sales | Capex 4.0% in 2026 fading to 2.5% by 2032–34; depreciation 1.2% rising to 1.8%. Management guides 3–4% through 2030. Terminal reinvestment is 0.7% of sales plus 4% of incremental sales | |||
| Working capital | 1% of incremental revenue 2027–30, 4% thereafter; 2026 free cash flow pinned at the $2.5bn guide. Cash conversion works out at 86% in 2027 rising to 92% in 2030 — inside management’s 95–100% only if deposits keep growing, which we do not assume | |||
| Share-based compensation | 0.4% of revenue deducted as a cash cost; diluted shares held at 392.8m | |||
| Discount rate | 9.50% | 9.50% | 9.50% | 9.50% |
| Terminal growth | 3.00% | 3.00% | 3.00% | 3.00% |
| UtilityInnovation Group | Value-neutral in every path: $1.45bn of cash out equals the asset in at roughly 13x 2027E EBITDA of ~$110m. The earn-out of up to $1.15bn is paid only if EBITDA is higher, so it is roughly neutral too | |||
| Discounted value per share, page model | $107 | $238 | $316 | $395 |
| Published scenario value | $106 | $238 | $317 | $396 |
| $bn | 2026 | 2027 | 2028 | 2029 | 2030 | 2031 | 2032 | 2033 | 2034 | 2035 |
|---|---|---|---|---|---|---|---|---|---|---|
| Revenue | 14.0 | 18.2 | 22.2 | 26.2 | 30.1 | 33.4 | 36.5 | 39.0 | 41.0 | 42.6 |
| Adj. operating profit | 3.33 | 4.61 | 5.84 | 7.07 | 8.29 | 9.20 | 9.95 | 10.61 | 11.10 | 11.50 |
| Margin | 23.8% | 25.3% | 26.3% | 27.0% | 27.5% | 27.5% | 27.3% | 27.2% | 27.1% | 27.0% |
| Unlevered free cash flow | 2.50 | 3.00 | 3.89 | 4.84 | 5.78 | 6.39 | 6.97 | 7.49 | 7.91 | 8.21 |
Valuation date 17 September 2026; only the roughly $0.92bn of 2026 free cash flow still to come is discounted. Present value of 2026–35 flows $34.8bn; terminal value $58.7bn, 63% of enterprise value; enterprise value $93.5bn; equity $93.6bn; $238 per share on 392.8m diluted shares.
The base growth path sits at the top of management’s range, not the middle, and that is deliberate. The 2025–30 CAGR of 24% reported and roughly 22% organic is the top of the 20–22% framework issued in May 2026. We set it there because the November 2024 framework — 12–14% organic growth through 2029 — was beaten by roughly double inside eighteen months: 2025 delivered about 27%, 2026 is guided at 31%. A management that under-promises by half changes what “modal” means. The counter-argument is also on the record and it is why the base does not sit above the range: no $14bn industrial has sustained 25% organic growth for a decade, and the 2028 digestion risk is real. 2027 is backlog conversion plus new capacity; 2028–30 is market growth of 18–20% plus share and content, decaying toward the market.
A 27.5% 2030 margin is above management’s target and above anything the company has earned. The target is 27%-plus, with quantified levers of $125m a year of labour savings and $250m a year of material savings. We credit it and half a point more because the 2024 investor day targeted about 25% by 2029 and Q4 2026 is guided at 25.8% — the plan is running two years ahead on margin. The fade to 27.0% by 2035 is for competition and mix: the 13% APAC margin shows what the business earns where Delta and Envicool set price, and the Americas’ 27.6% is partly scarcity pricing during a build-out.
Cash conversion is held below management’s 95–100%. First-half free cash flow of $1.58bn was funded 115% by deferred-revenue growth. Deposits are real cash and good commercial terms, but they are an advance, not earnings, and when order growth decelerates the tailwind becomes a headwind. The model lets working capital absorb 1% of incremental revenue while deposits fund it and 4% thereafter, which lands conversion at 86–92% — a fade, not a reversal. A reader who believes deposits keep growing at today’s intensity should treat the base as conservative by a few dollars a share.
The terminal value is deliberately austere. A 3% terminal growth rate on 2035 cash flow implies roughly 10.5 times 2035 EBITDA, well below the 18–22 times mature electrical-equipment makers trade at today. Using a 14 times exit multiple instead would lift the base to about $268. We use the Gordon approach because an exit multiple imports today’s market regime into 2035, and because it is the main reason the base sits marginally below the price rather than above it. It is the largest single judgement in this report and it is disclosed as such in the known weaknesses below.
All three grids hold the base operating path fixed and vary two inputs at a time. The first is the discount rate against terminal growth, and the WACC 9.0–10.0% by growth 2.5–3.5% block is the published fair-value band.
| WACC ↓ / terminal growth → | 2.0% | 2.5% | 3.0% | 3.5% | 4.0% |
|---|---|---|---|---|---|
| 8.5% | $253 | $267 | $284 | $304 | $328 |
| 9.0% | $234 | $246 | $259 | $275 | $294 |
| 9.5% base | $217 | $227 | $238 | $251 | $267 |
| 10.0% | $202 | $211 | $220 | $231 | $243 |
| 10.5% | $190 | $197 | $204 | $213 | $224 |
The second grid holds the growth path and varies the discount rate against the terminal margin.
| WACC ↓ / terminal margin → | 23.5% | 25.0% | 26.5% | 28.0% | 29.5% |
|---|---|---|---|---|---|
| 8.5% | $258 | $269 | $280 | $292 | $303 |
| 9.0% | $236 | $246 | $256 | $266 | $276 |
| 9.5% base | $217 | $226 | $235 | $244 | $253 |
| 10.0% | $202 | $210 | $217 | $225 | $233 |
| 10.5% | $188 | $195 | $202 | $209 | $216 |
The third scales every post-2026 growth rate in the base path by a single factor against the terminal margin, at the 9.5% rate. It is the widest, because those are the inputs that actually decide this company: 0.8x means 2027 growth of 24% instead of 30%, and so on down the path. The bear case is roughly the 0.6x column with a lower margin; the bull roughly 1.2x with a higher one.
| Growth intensity ↓ / terminal margin → | 23.5% | 25.0% | 26.5% | 28.0% | 29.5% |
|---|---|---|---|---|---|
| 0.6x | $151 | $157 | $162 | $168 | $174 |
| 0.8x | $181 | $189 | $196 | $203 | $210 |
| 1.0x base | $217 | $226 | $235 | $244 | $253 |
| 1.2x | $260 | $271 | $282 | $292 | $303 |
| 1.4x | $310 | $323 | $336 | $349 | $362 |
Modal base case $238. Fair-value band $210–275, the WACC 9–10% by growth 2.5–3.5% block of the first grid, which is what anchors the rating; the price sits at the 52nd percentile of it. Probability-weighted value $233 across the four scenarios at our own weights. Note what the third grid says: on the base path the price is reached only with a terminal margin near 29%, or with growth intensity a notch above 1.0x — the reverse DCF below puts it at 1.03x. One notch above the modal input in either direction, and no more than that.
There are three equivalent readings of the market price. Growth: holding base margins and a 9.5% rate, the price requires the base growth path scaled by 1.03x — 2027 revenue up 31%, 2030 revenue of about $30.7bn, a 2025–30 CAGR of 24.6% and 2035 revenue near $43.8bn. Margin: holding the base growth path, the price requires a terminal adjusted operating margin of about 27.8%, 80 basis points above the base terminal — in effect, the 27.5% the base credits for 2030 held rather than faded. Rate: holding everything else, the price is the base case discounted at about 9.35% rather than 9.5%.
None of the three is implausible; each is one notch above the modal input. The market is not wrong about Vertiv on this evidence — it is pricing the top of management’s range, and the burden this report places on itself is to say why the modal case should be at the top of that range (management’s record of under-guiding) rather than above it (no $14bn industrial has sustained 25% organic growth for a decade, and the 2028 digestion risk is real). The reverse DCF is also why the rating is fairly valued and not overvalued: a price one notch above the mode on a distribution this wide is inside the band, not outside it.
| Risk | Priced in | Not priced in |
|---|---|---|
| AI-capex digestion / cyclicality | Bear scenario at 25% weight: 2028 revenue −8%, margin 20% | Discount rate (asset beta, not stock beta); base growth path |
| Mega-project execution / timing | Base 2027 growth of 30% rather than the 34–36% Q3 run-rate; H2 2026 taken at guide, not above | Margin path; discount rate |
| Competition / architecture shift / Asian pricing | Terminal margin fade 27.5% to 27.0%; bear margin 23% | Revenue path; discount rate |
| Deposit unwind / working capital | Working capital 1% rising to 4% of incremental sales; cash conversion 86–92%, below management’s 95–100% | Margin; discount rate |
| Tariffs / price-cost | Base margin path assumes price-cost neutral to positive; bear margin covers a reversal | Separate haircut |
| Acquisitions / earn-outs | UIG value-neutral; adjusted rather than GAAP profit used, so amortisation excluded but earn-out remeasurement not credited | Discount rate; revenue path beyond roughly $2bn of acquired growth by 2030 |
| Dilution | Share-based compensation deducted as a cash cost | Share-count growth |
| Disclosure withdrawal | Not priced — a transparency issue, not a cash-flow one. Carried in the known weaknesses | |
The most expensive name in its group on this year, and in line with Eaton and Schneider on next year while growing earnings three to four times faster. Vertiv against the four suppliers it competes with most directly. Vertiv is at the 17 September intraday quote; peer prices are 9–10 September 2026 and were not refreshed; forward figures are company guidance midpoints where given and otherwise approximate consensus, flagged with an asterisk.
| Vertiv (VRT) | Eaton (ETN) | Schneider (SU) | nVent (NVT) | Johnson Controls (JCI) | |
|---|---|---|---|---|---|
| Price / market cap | $244 / ~$96bn | $415 / ~$162bn* | ~$174bn (Jun)* | $159 / ~$26bn* | $143 / ~$92bn* |
| 2026E revenue growth | +37% (31% org.) | 11–13% organic | ~6–8%* | +37–39% (32–34% org.) | ~8% organic (FY Sep) |
| 2026E adjusted operating margin | 23.8% | 24.3% (segment) | ~19% (adj. EBITA)* | ~22% (ROS) | ~17% (seg. EBITA)* |
| Free cash flow / yield | $2.5bn / 2.6% | ~$4.5bn / 2.8%* | ~€5bn / 3%* | ~$0.6bn / 2.3%* | ~$2.9bn / 3.2%* |
| P/E 2026E guide midpoint | 36.4x ($6.70) | 30.7x ($13.50) | ~30x* | 31.4x ($5.05) | 28.3x ($5.05, FY26) |
| P/E 2027E | 26.8x (cons. $9.10) | ~27x* | ~27x* | ~25x* | ~24x* |
| PEG 2026E P/E ÷ 2026–27 EPS growth | 1.0x (36% growth) | ~2.4x | ~3x | ~1.3x | ~1.9x |
| EV / revenue 2026E | 6.8x | ~5.9x* | ~3.9x (TTM) | ~6.3x* | ~3.7x* |
| EV / EBITDA 2026E | 27.4x | ~22x* | 18.4x (TTM) | ~25x* | ~18x* |
| Price / FCF 2026E | 38x | ~36x* | ~33x* | ~43x* | ~31x* |
*Approximate. Sources: Vertiv Q2 2026 release; Eaton Q2 2026 8-K (FY26 adjusted EPS $13.40–13.60, organic growth 11–13%, segment margin 24.1–24.5%); nVent Q2 2026 release (FY26 adjusted EPS $5.00–5.10, organic growth 32–34%); Johnson Controls Q3 FY26 release (FY26 adjusted EPS ~$5.05, organic growth ~8%); Schneider multiples per FactSet/Morningstar aggregation as at 9 June 2026 (P/E 31.8x, EV/EBITDA 18.4x, EV/revenue 3.9x). Peer prices: ETN and NVT 9 September close; JCI intraday 10 September; SU not refreshed. Vertiv enterprise value uses 392.8m diluted shares and $171m of net cash; 2026E EBITDA is adjusted operating profit plus roughly $170m of depreciation. Peer multiples are directionally reliable and not precise to the turn; they are not used in the valuation.
Read that two ways. On the current year Vertiv is the most expensive name in the group on every multiple — 36x earnings against a 28–31x peer range, 27x EBITDA against 18–25x, 6.8x sales against 3.7–6.3x. On the following year the premium collapses: at consensus 2027 EPS of $9.10 Vertiv trades at 27x, in line with Eaton and Schneider and above only nVent and JCI, while growing earnings three to four times faster. The PEG of 1.0x is the lowest in the set. The market is not paying a quality or franchise premium over Eaton and Schneider; it is paying for one more year of 30%-plus growth and then asking the same price per unit of earnings as the mature electricals. That is consistent with the model: the price sits at the base case. The 2027 multiple requires that 2027 revenue lands near $18bn, which the last disclosed backlog, the deposits and the $4.3bn Q4 run-rate support; that margins step up another 150 basis points; and that 2028 is not a down year.
Intrinsic value is not a price target and this report does not publish one. But in September 2027 the market will price VRT on 2028 earnings, and the multiple regime it applies is worth being explicit about. Blending consensus 2028 EPS of about $10.13 with our base-case 2028 EPS of about $11.08 gives roughly $10.60. Apply the regimes this stock has actually traded in — 20x if the cycle is questioned (30% weight), 26x at today’s forward multiple (45%), 32x on a re-rating (25%) — and the trading values are $212, $276 and $339, a weighted twelve-month trading value near $273. This is a market-regime estimate, not a fair-value estimate, and it sits inside the band because the band is where the market is. What would expand the multiple: orders and backlog disclosure restored with a book-to-bill above 1.2x; Q3 revenue at or above $3.75bn with Q4 reaffirmed; hyperscaler 2027 capex guides up 20%-plus; a buyback executed in the drawdown. What would contract it: a second timing miss; deferred revenue flat or down sequentially; any hyperscaler “digestion” language; a tariff round that turns price-cost negative.
Terminal value is 63% of the valuation. Any ten-year DCF on a company growing 30% is dominated by what happens after 2030; the 3% Gordon growth rate implies roughly 10.5 times 2035 EBITDA, well below where mature electricals trade, and a 14x exit would lift the base to about $268. The base case leans on management’s own framework. If the 2024 framework was conservative because the AI cycle surprised everyone rather than because this team habitually sandbags, the base is too high, not too low. Order data are stale. The last disclosed backlog is nine months old; deferred revenue is an imperfect intake proxy that lags orders by one or two quarters and depends on contract terms management describes only qualitatively. Peer multiples mix guidance midpoints with approximate consensus and, for Schneider, a June snapshot. Insider ownership and compensation were not re-verified against the 2026 proxy statement. UIG is treated as value-neutral — conservative if it is the “time to power” wedge management describes, generous if $1.45bn for a 2020 start-up proves rich; either way about 1.5% of enterprise value. And the market data are intraday: the $243.83 price was captured during the 17 September session, at 11:36 New York time, and the close will differ.
Vertiv reported before the open on 29 July. Adjusted EPS of $1.52 beat a $1.42 consensus; revenue of $3.27bn missed a roughly $3.37bn consensus and the company’s own guidance midpoint by $76m. The stock fell 17.3% to $223.04 on the day. When a 2% miss costs a sixth of the company, the price is telling you what it assumed.
| Metric | Q2 2026 | Q2 2025 | Change | Why it matters |
|---|---|---|---|---|
| Net sales | $3,274m | $2,638m | +24% (+18% org.) | $76m below the guide midpoint and ~$100m below consensus; the first miss in the current run. Management attributes it to multi-phase project timing and internal supply-chain congestion, not demand |
| Adjusted operating profit | $738m | $489m | +51% | $28m above guide despite the revenue shortfall — the mix and productivity story is intact |
| Adjusted operating margin | 22.6% | 18.5% | +410 bp | 140 bp above guide. Price-cost positive including tariffs; Americas 27.6%, EMEA 25.7%, APAC 13.3% |
| Gross margin (GAAP) | 37.7% | 34.0% | +370 bp | Services 40.8%, products 37.0%. Operating leverage is arriving at the gross line, not just SG&A |
| Adjusted diluted EPS | $1.52 | $0.95 | +60% | $0.12 above guide, helped by a 20% adjusted tax rate (guide 23%) from stock-compensation deductions |
| GAAP diluted EPS | $1.27 | $0.83 | +53% | Gap to adjusted: $0.19 amortisation, $0.07 PurgeRite contingent consideration |
| Adjusted free cash flow | $925m | $277m | +234% | Conversion above 150%; $451m of working-capital inflow, of which deferred revenue +$1,172m offset inventory −$664m and receivables −$587m |
| Deferred revenue | $3,634m | $1,815m 31 Dec 25 | +100% in six months | Advance and milestone payments on large projects — the only remaining public proxy for order intake |
| Net leverage | −0.1x (net cash) | 0.2x 31 Mar 26 | Net cash achieved | Liquidity $5.6bn; investment grade since February |
| FY2026 revenue guide | $14.0bn | $13.75bn April | +$250m | Organic growth 31%; implies H2 revenue of $8.08bn, up 45% year on year |
| FY2026 adjusted EPS guide | $6.70 | $6.35 April | +$0.35 | Third raise of the year; 60% growth on 2025 |
| FY2026 adjusted FCF guide | $2.5bn | $2.2bn April | +$300m | About 95% conversion for the year even after $560m of capex |
Source: Q2 2026 earnings release and results presentation, and the earnings call transcript, all of 29 July 2026. Consensus figures as reported in the financial press and not independently verified.
One: the revenue miss is an execution signal, not a demand signal. Management named the mechanism — SmartRun and OneCore mega-projects with “significant interdependencies”, multiple Vertiv factories feeding other factories — and said the slipped revenue ships in the second half. It also said second-half guidance assumes the congestion partly persists. Both statements can be true; the test is the third quarter. The slipped amount was not quantified, only described as “the majority” of the miss.
Two: margins are running ahead of the plan by roughly two years. The 2024 investor day targeted about 25% by 2029; Q4 2026 is guided at 25.8% and 2030 at 27%-plus. Each 100 basis points on a $30bn revenue base is worth roughly $300m of operating profit, and the company is delivering it at the gross line rather than through SG&A leverage alone.
Three: customers are financing the growth. Deferred revenue rose $1.17bn in a single quarter. The chief financial officer confirmed there is no structural change in revenue recognition — the mix has simply shifted to project-based contracts with deposits at order, design and delivery milestones — and that Vertiv aims to stay “ahead of the cash curve” on every project. It drove a large part of the working-capital benefit, and the CFO said so.
Four: orders and backlog have left the disclosure set. Neither figure appeared in the first- or second-quarter 2026 release or on either call; the language is now “pipeline”, “another year of robust orders growth” and “very strong backlog coverage”. The last quantified backlog is $15.0bn at year-end 2025. The withdrawal coincided with the largest order quarter in the company’s history, so it is unlikely to be hiding weakness; it does remove the metric investors used to test the growth path.
Five: capital is being deployed upstream and into thermal. ThermoKey and Strategic Thermal Labs closed in the quarter for $278m of cash; BMarko in April; and five weeks after the call Vertiv agreed to buy UtilityInnovation Group for up to $2.6bn. Capex is at the top of the range at 4% of sales. The buyback authorisation, live through 2027, was described as “opportunistic” — and has not been used in 2026.
| Q3 2026 | Q4 2026 (implied) | FY2026 | |
|---|---|---|---|
| Net sales | $3,650–3,850m | $4,225–4,425m | $13,800–14,200m |
| Organic growth | 34–36% | n/g | 30–32% |
| Adjusted operating profit | $898–938m | $1,096–1,136m | $3,285–3,365m |
| Adjusted operating margin | 24.0–25.0% | 25.3–26.3% | 23.3–24.3% |
| Adjusted diluted EPS | $1.77–1.83 | $2.17–2.23 | $6.65–6.75 |
Regional organic growth, Q3: Americas high-30s, APAC high-30s, EMEA mid-teens. Full year: Americas high-30s, APAC low-30s, EMEA low single digits. Adjusted tax rate 23% for Q3 and 21% for the year; diluted shares about 392.8m. FX translation adds about $89m to full-year sales and roughly nothing to profit.
On tone: confident to the point of exuberance from the executive chairman — “the future is so bright we have to wear shades” — and more measured from the chief executive, who used “complexity”, “learning curve” and “prudent” repeatedly and refused the word “disruption”. The chief financial officer was precise on the mechanics of deferred revenue and declined to quantify the slipped revenue. Analyst questions concentrated on whether project complexity makes quarterly timing shifts permanent (complexity is permanent; the ability to manage it is improving), how much revenue slipped (not quantified), whether deferred-revenue accounting changed (no), whether on-time delivery is opening share to competitors (“if anything improving”), the buyback threshold (opportunistic, no commitment), 800 VDC timing and a rumoured customer push-out to 2029 (gradual; Vertiv is agnostic across AC, 400 VDC and 800 VDC), and the M&A pipeline (active, disciplined, core first). The overall posture — raise the year on every line while conceding a quarter — is that of a team that believes demand is not in question and wants credit for saying so. What to check in the third-quarter release, in order: revenue against the $3.75bn midpoint, whether Q4 is reaffirmed at $4.3bn, the sequential change in deferred revenue and inventory, and any restoration of orders or backlog disclosure.
Vertiv makes the equipment that sits between the electricity grid and the server rack, and between the server chip and the outside air. On the power side that is uninterruptible power supplies (Liebert, Trinergy), switchgear and busway (E+I Engineering, acquired 2021), power distribution units, DC power systems (NetSure) and battery energy storage (EnergyCore). On the thermal side it is chillers, heat-rejection units, computer-room air handlers, in-row and rack cooling, and — increasingly — the liquid-cooling chain: coolant distribution units (CoolChip), cold plates (Strategic Thermal Labs, acquired April 2026), heat exchangers (ThermoKey, June 2026) and the fluid-management services that commission them (PurgeRite, 2025). It also sells racks and integrated IT enclosures (Great Lakes), monitoring and controls software (Unify, Avocent), and prefabricated modular data halls (SmartRun, OneCore) that package all of the above into a factory-built product. Around all of it sits a service business of roughly 5,300 field engineers in 130 countries.
By product, FY2025 revenue was roughly 35% power management, 30% thermal management, 20% services and spares, 8% IT systems and 7% infrastructure solutions. By end market it is about 85% data centres, 10% communication networks and 5% commercial and industrial. By geography it is 62% Americas, 20% Asia-Pacific and 18% EMEA. Dell’Oro ranks Vertiv first in data-centre thermal management; Omdia ranks it first in large three-phase UPS and in power-distribution infrastructure.
Vertiv is a project-and-product manufacturer with a growing annuity attached. Products, 80% of sales, are sold to hyperscale cloud operators, colocation developers, neo-cloud and AI-compute operators and enterprises, either directly or through consulting engineers, contractors and distributors. Large projects are increasingly sold as engineered systems with advance and milestone payments, which is why deferred revenue has doubled in six months. Services, 20%, are roughly three-quarters lifecycle work — maintenance contracts, spares, retrofits, monitoring — and one-quarter project work such as commissioning and start-up; management expects them to grow 20%-plus a year as the installed base compounds. Gross margin was 37.7% in Q2 2026 (products 37.0%, services 40.8%), up from 34.0% a year earlier.
The business is the former Emerson Network Power, carved out to Platinum Equity in 2016 and listed in February 2020 through a merger with GS Acquisition Holdings, the special-purpose vehicle chaired by former Honeywell chief executive David Cote, at about $10 a share. From 2022 to 2026 revenue has compounded at roughly 25% a year, from $5.7bn to a guided $14.0bn, while the adjusted operating margin has gone from 7.7% to a guided 23.8%; adjusted EPS has risen twelvefold, from $0.53 to $6.70. Vertiv received investment-grade ratings from Moody’s (Baa3) and S&P (BBB-) in February 2026, refinanced into $2.1bn of senior unsecured notes and a $2.5bn revolver in March, and joined the S&P 500 the same month. Total shareholder return from end-2022 to May 2026 was about 2,600%.
Management’s served-market estimate from May 2026, sourced to Omdia, 451 Research, IDC and Dell’Oro, is about $62bn in 2025 growing at 16–18% a year to 2030, of which data centres are about $49.5bn growing 18–20% and the cloud-and-colocation sub-segment about $29.5bn growing 23–25%. Adjacencies added through acquisition and new products — fluid services, converged infrastructure, DC power, chilled water — take the addressable market to roughly $75bn. On these numbers Vertiv’s 2025 revenue of $10.2bn is a ~16% share of its legacy served market, and management targets roughly 15% share growth over the plan. The external pacing items are not demand but power availability, permitting and skilled commissioning labour — which is precisely why Vertiv is buying its way upstream to the grid interconnect: UtilityInnovation Group, announced 2 September 2026, at $1.45bn plus up to $1.15bn of earn-out.
Vertiv’s own estimate of its content opportunity is $3.25–3.75m per megawatt of IT load, up 10–15% from the prior range, and management argues the migration to medium-voltage and 800-volt DC architectures — customer validation in 2026 and deployment in 2027 for rack and pod, a year later for the data-hall level — raises rather than lowers that figure. Industry forecasts cited at the investor day have 20–35 GW of new data-centre power added per year through 2030, roughly 140 GW cumulative. None of these market figures is used in the model; the revenue path is built from the company’s own guidance, backlog and deposits.
Schneider Electric, via APC, Uniflair and its EcoStruxure software, is the only rival with a comparably complete power-plus-thermal portfolio and is stronger in Europe. Eaton is the scale competitor in power — UPS, switchgear, busway — but has no thermal business. nVent competes in liquid cooling, enclosures and cable management and is growing as fast as Vertiv from a smaller base. Johnson Controls and Trane are the chiller incumbents now being pulled into data-centre cooling. Delta Electronics, Envicool and Huawei are the price competitors in Asia; Supermicro, Boyd and CoolIT compete at the rack and cold-plate level. The competitive question that matters most is architectural: as NVIDIA publishes reference designs for 800 VDC and the hyperscalers standardise, does value migrate toward the integrator that can ship a validated system — Vertiv’s thesis — or toward commoditised components? Section 06 takes this up.
| $m, FY Dec | 2022 | 2023 | 2024 | 2025 | 2026E, guide mid |
|---|---|---|---|---|---|
| Net sales | 5,692 | 6,863 | 8,012 | 10,230 | 14,000 |
| Reported growth | 14% | 21% | 17% | 28% | 37% (31% organic) |
| Adjusted operating profit | 439 | 1,054 | 1,552 | 2,090 | 3,325 |
| Adjusted operating margin | 7.7% | 15.3% | 19.4% | 20.4% | 23.8% |
| GAAP operating profit | 223 | 872 | 1,367 | 1,830 | 2,957 |
| Adjusted diluted EPS | 0.53 | 1.77 | 2.85 | 4.20 | 6.70 |
| GAAP diluted EPS | (0.04) | 1.19 | 1.28 | 3.41 | 5.87 |
| Adjusted free cash flow | (260) | 778 | 1,135 | 1,887 | 2,500 |
| Capex | 100 | 128 | 167 | 220 | ~560 (4.0% of sales) |
FY2022–25 from the 2026 Investor Conference appendix (19 May 2026); 2026E from the Q2 2026 release guidance midpoints (29 July 2026). GAAP 2026E figures are the company’s own reconciliation at midpoint. Adjusted operating profit excludes amortisation of acquired intangibles, contingent-consideration remeasurement, restructuring and M&A costs.
| $m | Q1 25 | Q2 25 | Q3 25 | Q4 25 | Q1 26 | Q2 26 | Q3 26E | Q4 26E |
|---|---|---|---|---|---|---|---|---|
| Net sales | 2,036 | 2,638 | 2,676 | 2,880 | 2,650 | 3,274 | 3,750 | 4,325 |
| Organic growth | n/d | 34% | n/d | 19% | 23% | 18% | ~35% | ~50%* |
| Adj. operating margin | 16.5% | 18.5% | 22.3% | 23.2% | 20.8% | 22.6% | 24.5% | 25.8% |
| Adj. diluted EPS | 0.64 | 0.95 | 1.24 | 1.36 | 1.17 | 1.52 | 1.80 | 2.20 |
| Adj. free cash flow | 265 | 277 | 436 | 910 | 653 | 925 | — | — |
Q3 and Q4 2026 are guidance midpoints (29 July 2026). *Q4 2026 reported growth of about 50% at the midpoint; organic not separately guided. n/d: not disclosed in the material reviewed. The second half must deliver $8.08bn of revenue, up 45% on H2 2025, for the full-year guide to be met.
Cash and short-term investments of $3.11bn against $2.94bn of long-term debt — fixed-rate notes, no maturities until the 2030s — leave a net cash position of about $0.17bn; liquidity including the undrawn $2.5bn revolver is $5.6bn. The working-capital picture is unusual for an industrial: receivables of $3.75bn and inventories of $2.52bn, up $1.07bn in six months to feed the second-half ramp, are financed almost entirely by payables of $2.47bn and customer deposits of $3.63bn, so net trade working capital is only about 1% of annualised revenue. Goodwill of $2.28bn and intangibles of $1.80bn are 26% of total assets; equity is $4.76bn. The UtilityInnovation Group purchase will consume about $1.45bn of cash at closing in Q4 2026, taking the company temporarily to a modest net debt position before free cash flow refills it.
| Driver | Evidence |
|---|---|
| Backlog conversion and capacity | The last disclosed backlog of $15.0bn exceeded FY2026 guided revenue; five large Americas plant expansions, the Johor (Malaysia) plant, EMEA chiller capacity and a South Carolina infrastructure-solutions site targeting a roughly seven-fold regional capacity increase come online through 2027 |
| Content per megawatt | Densification (GB300, Vera Rubin), liquid-cooling attach and the move to medium-voltage and DC power trains raise Vertiv dollars per MW from about $3m toward $3.25–3.75m |
| Converged infrastructure | Factory-built SmartRun and OneCore modules pull site-integration spend into Vertiv’s revenue line; management cites up to 50% shorter deployment and 25% lower total cost of ownership |
| Services annuity | Every megawatt shipped adds lifecycle service revenue at higher gross margin for a decade or more |
| Upstream expansion | UIG adds microgrid controls, on-site generation orchestration and behind-the-meter switchgear — the “time to power” bottleneck that now gates AI builds |
| EMEA recovery | The region shrank 15% organically in H1 2026; guidance has it growing mid-teens in Q3 and low single digits for the year, with Frankfurt-type campus wins (Data4) as evidence |
| Risk | Evidence |
|---|---|
| Hyperscaler capex cycle | About 85% of revenue depends on data-centre construction; a 2027 or 2028 digestion year is the dominant risk and the whole of the bear case |
| Execution on very large projects | Q2 revenue missed the guide midpoint by $76m on “timing shifts” caused by multi-phase mega-projects and internal factory interdependencies; management calls it a learning curve and has left “wiggle room” in H2 |
| Disclosure | Orders, book-to-bill and backlog were dropped from the quarterly release in 2026; investors now infer demand from deferred revenue and management’s “pipeline” language |
| Deposit-driven cash flow | $1.82bn of the $1.58bn first-half free cash flow came from deferred-revenue growth; conversion will fall toward earnings when intake stops accelerating |
| Architecture shift | 800 VDC could move value to power shelves and solid-state transformers where Eaton and Delta are strong; Vertiv’s SST is still in development |
| Acquisition pace and adjusted earnings | Six acquisitions in 18 months; adjusted EPS excludes $0.78 of amortisation and $0.16 of contingent-consideration remeasurement in 2026 (GAAP $5.87 against adjusted $6.70) |
| Tariffs and price-cost | Positive so far, including tariff countermeasures; a reversal would show in gross margin within two quarters |
| Source | Score | Assessment | Versus competitors |
|---|---|---|---|
| Distribution and service network | 9 | Roughly 5,300 field service engineers in 130 countries, the largest dedicated critical-infrastructure service force; fluid-management services (PurgeRite NearZero) have no scaled competitor | Clear leader; Schneider close in Europe; Eaton smaller in service; nVent and the chiller makers rely on third-party contractors |
| Brand and reference position | 8 | Liebert, NetSure and Avocent carry sixty-plus years of critical-power and cooling reputation; Vertiv is the named power and thermal partner in NVIDIA reference architectures for GB300 and Vera Rubin, and is first in thermal (Dell’Oro) and large UPS (Omdia) | Schneider/APC comparable; Eaton comparable in power only; nVent, JCI and Trane are component or chiller brands, not system brands |
| Scale | 8 | $14bn of revenue, 32 factories, about 37,000 staff, roughly $10bn of projected annual material spend by 2030. Capacity is being added at a pace — seven-fold in South Carolina, a new Johor plant — that only a handful of suppliers can fund from operating cash | Schneider ($40bn-plus group) and Eaton ($27bn) are larger groups but smaller in data-centre thermal; nVent ($4bn) is a fraction of the size |
| Switching costs | 7 | Installed equipment is serviced for 10–20 years on Vertiv contracts; controls (Unify) and monitoring integrate across the estate; prefabricated OneCore and SmartRun modules lock the whole power train and thermal chain into one vendor’s design. Switching mid-programme means re-qualifying a reference design | Higher than component suppliers; similar to Schneider at the system level |
| Customer loyalty | 7 | Customers prepaying $3.6bn and co-developing architectures (Foxconn/VisionBay.ai, Hut 8, Compass) indicates partnership rather than tender relationships; hyperscale and colocation sales grew 45% in 2025 | Hyperscalers deliberately multi-source; loyalty is strong at the programme level, weak at the bid level |
| Cost advantage | 6 | Purchasing leverage and a best-cost-country footprint (Mexico, Malaysia, India, China) deliver a targeted $375m a year of productivity; a 27.6% Americas margin implies a cost position competitors have not matched at scale | Delta and Envicool are cheaper on components in Asia — Vertiv’s APAC margin is only 13%; Eaton’s Electrical Americas margin of 27.5% is comparable |
| Intellectual property and technology | 6 | Deep application know-how in high-density liquid cooling, medium-voltage power, battery storage and 800 VDC; roadmap co-developed with chip vendors. Patent moats in this industry are modest; advantage comes from being first to a validated system | Ahead of the chiller makers and nVent on system integration; behind Eaton and Delta on solid-state transformers, where Vertiv’s is in development |
| Data and software | 4 | The Unify control layer, NextPredict predictive maintenance and Waylay digital services are real but small; the installed base gives a data advantage that is not yet monetised | Schneider’s EcoStruxure is the more mature software franchise |
| Network effects | 3 | Limited: more installed Vertiv equipment makes Vertiv service cheaper to deliver and reference designs more common, but there is no user-to-user effect | None of the peers has a meaningful network effect either |
| Regulation | 3 | Safety certification (UL, IEC) and utility interconnection rules raise entry barriers modestly; water-use rules are turning closed-loop thermal designs into a requirement, which favours Vertiv’s portfolio | Neutral versus incumbents; a barrier only to new entrants |
Composite: 6.1 out of 10, unweighted. A strong, service-anchored operating moat with weak software and network characteristics. The combination that matters is portfolio breadth plus service reach plus scale: Vertiv is the only supplier that can ship the medium-voltage switchgear, the UPS, the battery storage, the busway, the chiller, the coolant distribution unit, the cold plate, the rack and the module they all sit in — and then commission and maintain them in 130 countries. As projects become gigawatt-scale and customers ask for validated, repeatable infrastructure delivered in months, the value of a single accountable integrator rises. The deposit-backed contract structure is itself evidence: customers are financing Vertiv’s working capital because they need its delivery slots.
Biggest threats, in order. Architectural disintermediation: if 800 VDC and NVIDIA-published reference designs turn the power train into standard building blocks, value could migrate to the power-shelf, rectifier and solid-state-transformer suppliers — Eaton, Delta, Infineon’s customers — and to the hyperscalers’ own designs. Vertiv’s answer, that it supports every architecture and content per megawatt rises in each, is plausible but unproven at scale. Asian price competition: the 13% APAC margin against 27% in the Americas shows what happens where Delta, Envicool and Huawei set price. Customer power: five or six hyperscalers and a dozen colocation developers account for most of the growth, and their multi-sourcing discipline caps pricing in the long run even while capacity is scarce today.
Is the moat expanding, stable or shrinking? Expanding at the system level, stable at the component level. Every move of the last two years — E+I, CoolTera, Great Lakes, BSE, PurgeRite, BMarko, ThermoKey, Strategic Thermal Labs and now UIG upstream to the grid — extends the integrated offer and the service attach. Component by component, however, competitors are catching up in liquid cooling (nVent, JCI, Trane) and remain ahead in some power electronics. The moat’s durability therefore depends on the industry continuing to buy systems rather than parts. Current deposit behaviour says it is.
| Concern | Severity | Evidence and why it matters |
|---|---|---|
| Disclosure | 6 · MED-HIGH | Orders growth, trailing-twelve-month orders, book-to-bill and backlog were reported every quarter through Q4 2025 (organic orders +252%, book-to-bill ~2.9x, backlog $15.0bn). None of the four appears in the Q1 or Q2 2026 release, presentation or call. The withdrawal coincided with the largest order quarter in the company’s history, so it is unlikely to hide weakness — but it removes the metric investors used to test the growth path, and it means a slowdown would first appear in deferred revenue and guidance rather than in a headline number |
| Capex-cycle dependence | 6 · MED-HIGH | About 85% of revenue is data centres; hyperscale/cloud and colocation sales grew roughly 45% in 2025. Americas is 62% of revenue and grew 44% organically in Q1. This is the bear case in its entirety and is priced in the scenarios at 25% weight, not in the discount rate. Vertiv does not disclose customer-level concentration |
| Execution on mega-projects | 5 · MEDIUM | Q2 revenue $76m below the guide midpoint; management cites multi-phase execution and internal factory interdependencies; H2 requires +45% reported growth and Q4 a 25.8% margin. A second consecutive timing miss would convert “learning curve” into “structural lumpiness” and pressure the multiple even with demand intact |
| Deposit-driven cash flow | 4 · MEDIUM | Deferred revenue +$1.82bn in H1 2026 against adjusted free cash flow of $1.58bn; inventory +$1.07bn; receivables +$0.64bn. FY2025 conversion was 115%. Deposits are real cash and good commercial terms, but they are an advance, not earnings. When order growth decelerates the tailwind becomes a headwind and conversion mean-reverts toward 85–90%. The model assumes this |
| Acquisition pace and adjusted earnings | 4 · MEDIUM | Six acquisitions in 18 months; UIG at $1.45bn plus up to $1.15bn contingent (~13x 2027E EBITDA) for a firm founded in 2020. PurgeRite contingent consideration remeasured up $62m in H1 and excluded from adjusted profit. 2026 adjusted EPS $6.70 against GAAP $5.87. Roll-ups compound integration risk and make adjusted margins flatter than GAAP. The 12% gap is moderate and mostly amortisation, but earn-out remeasurement being excluded means overpaying is invisible in the headline number |
| Insider activity and buyback | 3 · LOW | Director Jan van Dokkum sold about 38,600 shares (~$9.8m) and EVP Anders Karlborg about 30,500 (~$7.5m) under pre-arranged plans in 2026. No repurchases in H1 2026 despite the authorisation and the stated “opportunistic” stance. Planned sales at this scale are routine; the more telling signal is that the company has not bought stock in a 35% drawdown |
| EMEA | 3 · LOW | H1 2026 organic sales −14.8%; Q1 margin fell to 16.6%. Q2 improved to −2.4% organic with a 25.7% margin, helped by an easy Ireland-switchgear comparison. Management guides an H2 recovery; failure would suggest the European AI build is slower than the pipeline language implies |
| Tariffs and price-cost | 3 · LOW | Management states price-cost is positive in 2026 inclusive of tariffs and countermeasures; about 40% of revenue is outside the Americas; production in Mexico, Malaysia, China, Italy and Northern Ireland. A tariff escalation would compress gross margin with a one-to-two-quarter lag; the pricing mechanism has absorbed two rounds so far |
| Legal | 2 · LOW | At least five plaintiffs’ firms announced “investigations” after the 29 July drop, framed around Q1 2026 statements about backlog and deployments running smoothly. No complaint had been filed in the sources reviewed. Post-drop solicitations are near-universal for S&P 500 names; risk becomes material only if a court finds the Q1 statements misleading. These are the firms’ own statements soliciting potential claimants; no allegation has been tested and we take no view on their merit |
Overall red-flag score: 4 out of 10. No accounting or governance problem was found. The two material concerns — the withdrawal of order disclosure and the concentration in one capital-expenditure cycle — are strategic and cyclical rather than forensic, and both are priced explicitly in the scenario weights rather than being charged again in the discount rate.
| Strength | Weight | Evidence and why it matters |
|---|---|---|
| Deposits as demand signal | 7 · STRONG | Deferred revenue $1.81bn, $2.46bn, $3.63bn across the last three balance sheets, while revenue rose 24%. Deposits are collected at order, at design freeze and at delivery milestones on the largest projects. Customers do not prepay billions to a supplier they intend to switch away from; read as an intake proxy, deposits point to order growth well ahead of revenue growth in H1 2026 |
| Margin ahead of plan | 7 · STRONG | November 2024 target: about 25% by 2029. Q4 2026 guide: 25.8%. New 2030 target: 27%-plus, with named levers (+200 bp leverage, +150 bp productivity, +50 bp commercial) and quantified savings ($125m a year labour, $250m a year material). Two years ahead of schedule with the levers specified. The base assumes 27.5% in 2030 and a fade to 27%; the bull uses 29% |
| Services annuity | 7 · STRONG | About 20% of revenue; roughly 75% lifecycle and 25% project; 5,300 field engineers; 31-plus training academies; services gross margin 40.8% against products at 37.0%. Growing 20%-plus with the installed base. Provides the through-cycle floor most AI-hardware suppliers lack — in the bear case services are what keep margins above 20% |
| Guidance conservatism | 6 · NOTABLE | November 2024 framework: 12–14% organic CAGR 2024–29. Delivered 2025 organic about 27%, 2026 guided 31%. FY2026 EPS guide raised from $6.02 in February to $6.35 in April to $6.70 in July. A management that under-promises by half changes what “modal” means; our base sits at the top of the new 20–22% range, not the middle, for this reason |
| Balance-sheet optionality | 6 · NOTABLE | Net cash, investment grade, $2.5bn undrawn revolver, $28bn of projected deployable capital 2026–30 at 1.5x leverage; buyback authorisation live through 2027. Capital deployed at 13x EBITDA (UIG) while the acquirer trades at 28x is accretive on multiple as well as EPS; a buyback in a drawdown would be the same trade |
| Content per megawatt | 5 · MODERATE | $3.25–3.75m per MW, up 10–15%; management “pretty convinced” 800 VDC lands at the high end; first GB300 (Taiwan) and Vera Rubin 800 VDC (Foxconn/VisionBay.ai) reference wins. Content growth compounds on top of megawatt growth. Not yet in filings, so carried as upside in the bull case at the same evidence standard as the architecture-shift risk in the bear |
Overall green-flag score: 6 out of 10. The underappreciated items are all about durability — deposits, services, margin levers — rather than about faster growth. They lift the bear-case floor more than the bull-case ceiling, which is why the bear case in this report has 20%-plus margins and a $106 value rather than the sub-$80 a pure hardware-cycle analogy would produce.
| Dimension | Score | Assessment |
|---|---|---|
| CEO track record | 5 | Giordano Albertazzi, chief executive since January 2023 and a two-decade Emerson Network Power and Vertiv operator (previously EMEA and Americas president), has taken the adjusted operating margin from 7.7% to a guided 23.8% while revenue compounded about 25% a year, fixed a 2022 price-cost crisis and a 2025 Ireland switchgear problem, and built the converged-infrastructure offer from scratch |
| CFO credibility | 4 | Craig Chamberlin, in post since 10 November 2025 after Wabtec Transit and fourteen years at GE, has been precise and unpromotional on three calls and presented a coherent 2030 framework in May. Held back only by tenure. David Fallon’s eight-year stewardship — refinancing, deleveraging, the 2024 buyback — set a high bar; he consults through 2026 |
| Guidance accuracy | 4 | Raised full-year guidance every quarter of 2025 and 2026 to date; beat the adjusted EPS midpoint in each of the last six quarters. The Q2 2026 revenue print fell $76m short of the midpoint, still inside the range — the first shortfall of the current run. Long-range frameworks have been conservative by a factor of about two |
| Transparency | 3 | Segment, product and organic-growth reconciliations are exemplary; the guidance bridge is detailed. Against that, the 2026 withdrawal of orders, book-to-bill and backlog — after years of quarterly disclosure and immediately after the largest print — is a step backwards, and the slipped Q2 revenue was not quantified |
| Capital allocation | 4 | Organic capacity first (capex 4% of sales), then technology bolt-ons at sensible multiples, then a $600m buyback in Q1 2024 at about $66 a share. Investment grade achieved and the balance sheet de-risked into fixed-rate notes. Dividend nominal |
| Acquisitions | 4 | E+I (2021) is the template: bought for switchgear and busway, now core to the power train. CoolTera, Great Lakes, BSE, PurgeRite, BMarko, ThermoKey and STL are all technology or capacity fits. UIG at up to $2.6bn is the first deal large enough to matter and the first to move upstream of the data centre; the earn-out structure puts roughly half the price at the seller’s risk |
| Buybacks | 3 | One well-timed programme (2024). None in 2026 through a 35% drawdown while describing the approach as opportunistic. Capital has gone to capacity and M&A instead — defensible, but the “opportunistic” language has not been backed |
| Dilution | 4 | Diluted shares 386m (2023) to 393m (2026E); about 3m a year of equity grants, 0.8%. Warrants from the 2020 SPAC structure were fully exercised or redeemed by 2024 |
| Insider ownership | 3 | Executive chairman David Cote holds a substantial personal stake dating to the 2020 sponsor structure; executive and director ownership is otherwise modest for a founder-less industrial. Two planned sales in 2026, about $17m combined. Detailed percentages should be checked against the 2026 proxy statement; not verified here |
| Compensation | 3 | Long-term incentives are weighted to adjusted EPS, adjusted free cash flow and relative TSR per past proxies; the adjusted metrics exclude amortisation and earn-out charges, which is standard but means acquisition costs do not reduce incentive pay. Not re-verified against the 2026 proxy |
| Board quality | 4 | Cote’s Honeywell operating record is the board’s defining asset and the origin of the Vertiv Operating System; independent directors bring industrial and technology backgrounds. The executive-chairman structure concentrates influence; the 2026 annual meeting passed all items without shareholder questions |
| Communication style | 4 | Candid on the mechanics of problems (Ireland in 2025, project timing in 2026) and willing to say “prudent” and “wiggle room”. The chairman’s rhetoric runs hot; the chief executive’s runs cool. Analysts get direct answers except on quantities management prefers not to disclose |
Composite: 3.75 out of 5. Largely yes, they act like long-term owners. The pattern — invest in capacity ahead of demand, buy technology rather than revenue, refinance into patient fixed-rate debt, buy stock only when it is cheap, set public targets and beat them — is the behaviour of managers who expect to be judged on a five-year record. The two exceptions are the retreat on order disclosure, which owners would want reversed, and the growing gap between adjusted and GAAP earnings as acquisition intensity rises, which owners should track.
| Window | Event | Upside | Downside | Confidence |
|---|---|---|---|---|
| 16–17 Sep 2026 | FOMC, with the ten-year at 4.85% and oil above $100; a hike would raise the discount rate for every long-duration growth stock | Hold or dovish guidance: multiple support | Hike: roughly $10–15 a share of DCF value per 25 bp at Vertiv’s duration | Certain date; outcome open |
| Late Oct 2026 | Q3 2026 results. Guide: revenue $3.65–3.85bn, adjusted EPS $1.77–1.83, margin 24–25% | Revenue at or above $3.75bn with Q4 reaffirmed at $4.3bn; deferred revenue still rising; EMEA organic positive. The “timing” narrative is closed | Second miss to midpoint or a Q4 trim: the multiple compresses toward peers (~30x) regardless of demand | High; date within a week of 28 Oct |
| Q4 2026 | UIG acquisition closes; $1.45bn cash out; first disclosure of UIG revenue and EBITDA | Early revenue synergies on “time to power”; earn-out shown to be well designed | Regulatory delay or thin disclosure of what was bought | Medium-high |
| Nov 2026 | SC26 (St Louis, 15–20 Nov) and NVIDIA data-centre events; Vera Rubin ramp; 800 VDC customer-validation milestones due in 2026 | Named 800 VDC sidecar and pod wins with 2027 deployment dates | Public push-out of 800 VDC timelines by a major customer, as rumoured on the Q2 call | Medium |
| Ongoing | Buyback authorisation live through 2027; no repurchases year to date | Repurchases begun in the drawdown: signals management’s own valuation view | Continued inaction while insiders sell under plans | Low |
| 31 Dec 2026 | Former CFO David Fallon’s consulting term ends; first full year for Craig Chamberlin | Smooth; disclosure practices refreshed | None specific | Certain |
| Late Jan – early Feb 2027 | Hyperscaler Q4 results and 2027 capex guidance (Microsoft, Alphabet, Amazon, Meta) | Aggregate capex guided up 20%-plus: validates the 2027 revenue step to about $18bn | “Digestion”, “optimisation” or flat guides: the bear case gains weight immediately | High on dates; content open |
| ~11 Feb 2027 | Q4 and FY2026 results, FY2027 guidance; 10-K mid-February | FY27 revenue guide at or above $18bn and margin at or above 25%; orders and backlog disclosure restored | FY27 guide below $17.5bn or margin flat; continued silence on orders | High |
| Feb 2027 | Anniversary of investment-grade ratings; agency reviews after UIG | Upgrade trajectory confirmed (Baa3/BBB- toward Baa2/BBB) | Negative outlook on acquisition pace | Medium |
| Q1 2027 | Solid-state transformer and MV DC UPS: 2027 customer validation begins per roadmap | Named validation partners | Slippage while Eaton and Delta ship | Medium-low |
| Apr / Jul 2027 | Q1 and Q2 2027 results | Organic growth holds 25%-plus; margins step toward 26%; deferred revenue stable or rising | Organic growth below 20% with deferred revenue falling — the working-capital tailwind reversing | High |
| 2027 | First 800 VDC rack and pod deployments (Foxconn/VisionBay.ai Vera Rubin site; others); UIG earn-out first measurement period | Content per MW measured at the top of $3.25–3.75m; earn-out paid because EBITDA beat | Content lands at or below the old range; earn-out not paid | Medium |
| Through 2027 | Capacity additions: Johor, five Americas plants, Ironton (+45% by Q2 2027), South Carolina | Capacity ahead of demand ends the timing-shift story | Capacity ahead of demand becomes under-absorption if orders slow | High |
| Any time | Plaintiffs’ investigations convert to a filed complaint; Ohio or Delaware | None | Headline risk; low expected cost | Low-medium |
Confidence reflects the timing of the event, not the direction of its outcome. The Q3 date is inferred from the company’s reporting pattern. Sources: company releases and the Q2 2026 call of 29 July 2026, the 2026 Investor Conference of 19–20 May 2026, and the published FOMC calendar. The ex-dividend date of 14 September 2026 ($0.062 quarterly, a 0.1% yield) has passed since the data date and is immaterial.
BULLFourth-quarter 2025 orders were up 252%; book-to-bill was 2.9x; backlog was $15bn against a year that will deliver $14bn. Since then deposits have doubled to $3.6bn — customers do not prepay for capacity they intend to cancel. The Q4 2026 exit run-rate is $17.3bn annualised before a single 2027 order. Thirty per cent growth in 2027 is arithmetic, not hope.
BEARThe company stopped telling you the orders number the quarter after it peaked. Deposits are a cash-flow fact, not an order fact; a slowing book would still show rising deferred revenue for two quarters as milestone payments arrive on old projects. And every capex supercycle in history — telecom 2000, shale 2014, crypto 2021 — ended with a year in which the customers’ spending fell 20–40%. Hyperscaler capex is about $600bn; the question is not whether it plateaus but when.
BULLTwenty-seven times 2027 earnings for a company growing EPS 36%, with a PEG of 1.0 against 2–3 for Eaton and Schneider. The reverse DCF says the price needs a 25% five-year CAGR — this management guided 12–14% in 2024 and delivered 27–31%. Their 20–22% is the new floor, and the stock is priced at the floor.
BEARThirty-six times this year’s earnings, 27x EBITDA, 38x free cash flow for a maker of chillers and switchgear — every one the highest in its peer group. The reverse DCF also says the price needs a 27.8% terminal margin or a 9.35% discount rate, on a stock that fell 17% in a day on a 2% revenue miss. Priced at the top of the range means no room for the range to be wrong.
BULLNumber one in thermal and in large UPS, the only complete power-plus-thermal-plus-service portfolio, 5,300 field engineers, and now a factory-built module that takes site-integration spend into Vertiv’s revenue line. Gross margin up 370 basis points in a year; Americas margin 27.6%. This is a compounding industrial franchise, not a component supplier.
BEARAPAC margin is 13% — that is what the business earns where Delta and Envicool compete on equal terms. The Americas margin is scarcity pricing during a build-out. NVIDIA is publishing the reference architectures; the hyperscalers write the specifications; when 800 VDC standardises the power train, Vertiv is an assembler competing with Eaton and Delta on rectifiers and solid-state transformers it does not yet ship.
BULLNet cash, investment grade, $5.6bn of liquidity, 95% cash conversion, capex fully self-funded. Services are 20% of revenue at a 41% gross margin and growing 20%-plus: that is the floor under a downturn. Inventory is up because the second half is up 45%; it is working capital for booked revenue, not speculation.
BEARFirst-half free cash flow was $1.58bn and deferred revenue rose $1.82bn — the entire cash flow, and then some, was customers’ money. Inventory up 73% and receivables up 21% in six months are what a company looks like right before the music stops, and the same balance sheet with flat deposits would have shown a cash outflow. Add $1.45bn out the door for a six-year-old microgrid firm with half its price contingent.
BULLA management that has beaten every target it set, fixed every operational problem within two quarters, and tells you when it is being prudent. The Q2 miss was $76m on $3.27bn, and they raised the year on every line the same morning. That is not a team losing control. And the next five months bring three dated events — Q3 at $3.75bn with Q4 reaffirmed, hyperscaler 2027 guides in January, an $18bn 2027 guide in February and very possibly the orders number back — each capable of taking the stock back through $300.
BEARA management that withdrew the one metric that would let you check the story, declined to quantify the slippage, and has not bought back a share in a 35% drawdown while calling the approach opportunistic. A chairman telling investors the future is so bright he needs sunglasses, on the day the stock fell 17%. The same three events, each capable of confirming that the deposit surge was the peak — and a Federal Reserve debating a hike with the ten-year at 4.85% and oil over $100, the worst macro backdrop for a 63%-terminal-value stock since 2022.
Neither side wins outright, and that is the finding. The bull has the stronger evidence on the next twelve months: backlog coverage, deposits, capacity and guidance all point to 2027 revenue near $18bn, and the peer comparison shows Vertiv is not carrying a franchise premium on forward earnings. The bear has the stronger argument on the five-year distribution: the price assumes no digestion year, and the history of capex supercycles says one is likely somewhere in 2027–2030. The bear’s forensic points — deposit-driven cash flow, inventory build, disclosure retreat — are real but are consequences of growth, not evidence of its absence. What is uncertain: the timing and depth of any hyperscaler capex pause; whether 800 VDC raises or lowers Vertiv content; whether management’s 2030 framework is a floor, as its 2024 framework proved to be, or a ceiling. Verify next: Q3 revenue and Q4 reaffirmation in late October; the sequential change in deferred revenue and inventory; hyperscaler 2027 capex guidance in January; the FY2027 guide and whether orders disclosure returns in February. At $244 the shares are fairly valued, with the bull holding the near-term tape and the bear holding the tail.
Every data centre needs two things besides computers: clean, uninterrupted electricity, and a way to get rid of the heat the computers make. Vertiv builds both. It makes the big battery-backed power systems, the switchboards, the cooling units, the pipes and pumps that carry liquid coolant to the chips, and the metal racks the computers sit in. Increasingly it builds whole pre-assembled data-hall modules in its factories and ships them to the site. It then sends its own engineers — about 5,300 of them — to install, test and maintain everything for years.
Mostly by selling that equipment to the companies building data centres: the big cloud operators, the developers who build data centres and rent them out, and the newer AI-computing firms. About a fifth of revenue is service work, which is steadier and more profitable. Because the projects are now enormous, customers pay a good part of the bill up front, which is why Vertiv had $3.6bn of customer money on its balance sheet in June against work not yet delivered.
Artificial intelligence needs far more computing power per building than anything before it, and that means far more electrical and cooling equipment per building. Vertiv is the market leader in cooling and in large power systems, so its sales are growing about 30% a year, its profit margins have tripled in four years, and its share price has gone from about $10 in 2020 to $244 today. It joined the S&P 500 in March 2026.
Vertiv used to publish its order book every quarter. It stopped this year, right after reporting the largest one in its history. What you can still see is the customer deposits, which doubled in six months — a strong sign that orders are still coming in fast. But deposits tell you about cash, not about orders, and they would keep rising for a while even after new orders slowed, as old projects hit their payment milestones. If you take one thing from this report, take that the number to watch is no longer published, and the substitute lags.
Here is the whole disagreement. Work the cash flows honestly and the business is worth about $210–275 a share to us, with a central estimate of $238, against a $244 price. That sounds like agreement, and it is — but at a level that already assumes the growth plan is delivered at the top of its range. There is no discount for the risk that it is not. If the AI build-out keeps growing for another four or five years the shares are worth $300 or more. If the handful of companies driving the spending pause for a year, as customers in every previous technology build-out eventually did, the shares could fall by more than half.
Right: the build-out does not pause, Vertiv keeps taking share, each megawatt of computing needs more Vertiv equipment as the technology gets denser, margins reach the 27%-plus the company is targeting, and the service business becomes a large, steady annuity.
Wrong: the big cloud companies pause their spending while they use what they have built; Vertiv’s sales fall, its factories are under-used, and the deposits that have flattered its cash flow stop arriving. Smaller risks: very large projects are proving hard to schedule, the company has stopped publishing its order book, and it is spending heavily on acquisitions.
The October results, for whether the delayed sales show up and the fourth-quarter guide holds. What the big cloud companies say in late January about their 2027 spending. And whether the February results bring the order book back.
| Status | What it covers |
|---|---|
| Verified from Vertiv filings and releases | All revenue, organic growth, adjusted and GAAP operating profit, gross margin by segment, adjusted and GAAP EPS, adjusted free cash flow, working-capital lines, deferred revenue, cash, debt, liquidity, goodwill and intangibles, share counts, regional margins, every guidance range, the 2030 framework and its levers, the served-market and content-per-megawatt estimates as management stated them, and the acquisition terms — taken from the Q2 2026 earnings release, results presentation and call transcript (29 July 2026), the Q1 2026 release (22 April 2026), the Q4 and FY2025 release (11 February 2026), the 2026 Investor Conference presentation (19–20 May 2026), and press releases on the appointment of Craig Chamberlin (13 October 2025) and the UtilityInnovation Group agreement (2 September 2026). The FY2022–25 financial record is from the Investor Conference appendix. The Q1 2024 release for the buyback. |
| Verified from market data | The $243.83 intraday price of 17 September 2026 and the daily close series from 17 September 2025 to 16 September 2026 behind the drawdown, peak and volatility figures; the 9 September closes for Eaton and nVent and the 10 September intraday price for Johnson Controls. Peer guidance from the Eaton Q1 and Q2 2026 8-K exhibits, the nVent Q1 and Q2 2026 releases and the Johnson Controls Q3 FY2026 release. |
| Third-party, attributed | Consensus figures — 2026E EPS $6.71, 2027E $9.10, 2028E about $10.13, 28 covering analysts, and the Q2 revenue and EPS consensus — as reported in the financial press on 3 and 10 September 2026, secondary and not independently verified. Schneider Electric multiples via FactSet/Morningstar aggregation as at 9 June 2026. Market rankings are Dell’Oro’s and Omdia’s as cited by the company. Plaintiffs’-firm notices of 11–21 August 2026 via press-release wires. Nothing in the valuation depends on any of these. |
| Our estimate or judgement | Enterprise value and every multiple in Sections 01 and 03; the deposit-funded share of first-half cash flow; the entire ten-year cash-flow path, the margin ramp and fade, the working-capital and capex paths, the four scenario settings, all three sensitivity grids, the reverse DCF, the $210–275 band and the $238 central estimate; the 25 / 45 / 25 / 5 scenario weights; the twelve-month regime values; the value-neutral treatment of UIG; and all scores in Sections 06, 07 and 08. The model on this page is the model that produced those figures — the scenario cards load the published settings and the first sensitivity grid is reproducible from them to within a dollar. |
| Still open | Orders, book-to-bill and backlog have not been disclosed since February 2026, so the intake picture rests on deferred revenue, which lags orders by one or two quarters and depends on contract terms described only qualitatively. The slipped Q2 revenue was not quantified. Insider ownership percentages and the incentive-plan structure were not re-verified against the 2026 proxy statement and are scored on prior-year structures. Peer multiples mix guidance midpoints with approximate consensus and, for Schneider, a June snapshot; they are directionally reliable and not precise to the turn. UIG’s 2027 EBITDA of roughly $110m is inferred from the disclosed multiple, not reported. Customer-level concentration is not disclosed. |
Method. Unlevered discounted cash flow, methodology version 1.2, in the operator form described in Section 02: a ten-year explicit forecast from the 2026 guidance midpoints to 2035, discounted mid-period at 9.5% in the base case from the 17 September 2026 valuation date, with a Gordon-growth terminal value at 3.0%. Free cash flow to the firm is adjusted operating profit after 24% tax, less share-based compensation charged as cash, less capital expenditure net of depreciation, less working capital absorbed by growth. Adjusted operating profit excludes amortisation of acquired intangibles and contingent-consideration remeasurement; the amortisation tax shield is ignored. Net cash of $171m at 30 June 2026 is added to enterprise value and the result is divided by 392.8m diluted shares. Scenario weights are owned by the analyst, the fair-value band is read from the sensitivity grid, and the reverse DCF is run against the market price. Ratings are undervalued, fairly valued or overvalued according to where the price sits relative to the band; Sonde does not issue buy, sell or hold recommendations or price targets. The reference price used throughout is the $243.83 intraday quote of 17 September 2026, and every multiple in this report scales directly with it.