Generac Holdings
This is the archived full research. Read the story
The American maker of the generator that sits outside a house and starts by itself when the power fails, now also building giant backup generators for data centres.
The scores
Growth3/4 ↑
Sales rose 16% in Q3 2026 and 12% in nine months after three flat years, mostly from data centres, acquisitions adding 1 to 2 points.
Business quality3/4 →
The dealer network protects home standby, but in data centres it sells the same machines as Cummins and Caterpillar and wins on delivery speed.
Financial position3/4 →
Debt of $1.33B against $0.39B of cash, covered by free cash flow of $132M in Q3 alone, with new plants funded from operations.
Execution record2/4 ↑
Missed estimates in late 2025, missed its 2023 to 2025 revenue and margin targets and cut the home guide in July; three beats since.
Per share capture3/4 →
Buybacks cut the count about 8% since 2021, but Amazon's warrant hands the customer up to 2.9% of the company as it buys.
Potential3/4 ↑
A large data-centre build on observable steps, with 2027 substantially booked through backlog and Amazon, though the orders are conditional.
Risk profile3/4 →
Product litigation and an emissions inquiry, one engine supplier for large units, and a data-centre order list resting on two or three buyers.
Valuation3/4 →
About 13 times guided adjusted EBITDA, but about 93% of a three-stage value near $189 on GAAP profit: fair, inside the 20% margin of safety.
Evidence quality3/4 →
Results and the Amazon terms are filed, but the backlog, the second hyperscaler and the margins are management's words, and the contract is unfiled.
If it works 0.9× · if it fails you keep about 84%.
The Verdict
We rate Generac Neutral. It leads home backup generators through its dealer network, and its commercial arm is growing fast on data centres, with unfilled orders of about $4.5B after the Amazon deal, so the business is solid. Those orders earn about half the home margin and can be cancelled, and calm weather stalls the higher-margin home half. Valued on its own cash, even with those orders shipping on time, the shares are worth about $189 against $177: fair, not cheap. As a bet on ecobee's thermostat programmes it is defensive.
The long caseAs data centres strain the grid, utilities pay to turn home thermostats down at peak, and ecobee already sits in millions of homes.
How we weighed it
The price against our value
Cuts both ways
Price at 93% of our value
The price is about nine-tenths of our value, with the bear case far below.
The business
In its favour
Solid
It leads home backup generators through its dealer network.
Growth against what the price assumes
Cuts both ways
Data-centre orders grow fast at about half the home margin.
What could go wrong
Against it
Orders may be cancelled, calm weather stalls home sales, rivals are adding capacity and it relies on one engine supplier.
Against competitors
Cuts both ways
It sits between its two large rivals, priced like them.
How its group is behaving
Cuts both ways
It sits in none of our ranked groups; outages and data-centre building move it.
The latest news and report
In its favour
Orders grew sharply and home sales returned to growth.
The evidence under this - 9 claims
The 10-Q names cancellation rights in some data-centre contracts (C16), so backlog is not revenue yet, and $2.9B of the rise comes from one quarter's orders, of which Amazon's $2.4B is most (C15).
- 2026-10-28editedQ3 call split the backlog by year and customer (C65); the release had not.
The 8-K says deliveries are expected, not committed, and the agreement's terms are not filed in full.
The adjusted EBITDA the market prices on leaves out about $120M of amortisation and $50M of stock pay; on that measure a 19% margin would give a higher value. If the bull case's 2027 and 2028 deliveries arrive, the value is about $307, above the price.
- 2026-11-09editedRe-struck on the upgraded method: every view read at 9%, the return on new capital set by the moat, and capital per dollar of sales read from the filings.
- 2026-11-09editedRe-struck at the 8% required return the owner chose after the roll-out screen: the market priced the median name on the roster at about 7.7%, so 9% on top of a 20% margin read most of it Unattractive.
- 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
- 2026-11-10editedre-valued on the owner's rules of 10 November: the return on new capital glides to its terminal level over years six to ten; a US company's normal tax is the federal 21% plus state, 25%
Adjusted EBITDA leaves out stock pay, guided at $54M to $58M, and $118M to $122M of acquisition amortisation; on trailing GAAP earnings the vendor's price-to-earnings is 51.
- 2026-10-28editedRe-struck at the 2026-10-27 close of $193.71 (was $224.94) on the 30 September balance sheet: company value $12.37B and 13.3 times guided EBITDA before refunds (was $14.34B and 15.4 times).
- 2026-11-06editedRe-struck at the 2026-11-05 close of $172.33 (was $193.71): company value $11.11B and 11.9 times guided EBITDA before refunds (was $12.37B and 13.3 times).
- 2026-11-08editedRe-struck under the valuation rules of 2026-11-08
The company
A home standby generator is an engine bolted on a pad outside the house, running on the gas line. When the grid fails, a switch starts it within seconds. American homes lost power for 11 hours on average in 2024, nearly double the decade before.
For data centres Generac packages large diesel engines, bought from one supplier, into generator sets that carry a whole building. Builders want backup for all their load and, with rivals quoting waits of 70 to 80 weeks, delivery speed wins orders.
The evidence under this - 7 claims
Neither Cummins nor Caterpillar states its lead times; the rival figure is Generac's.
The supplier is not named in the filings, so the term of the exclusivity cannot be checked.
How it makes money
It earns money by selling equipment. In 2025 home products were 54% of sales, commercial and industrial products 35%, and parts, warranties and subscriptions the rest.
The home business is the profitable half: in 2025 it kept about 22 cents of each sales dollar as profit before interest, tax and amortisation, against about 11 for the commercial half. Each unit sold through its 9,700 dealers also brings service and repeat business.
The evidence under this - 5 claims
No rival publishes a dealer count to compare; Rehlko and Briggs & Stratton are private.
Where the money comes from, stream by stream
| Stream | Share of sales (2025) | How it earns it |
|---|---|---|
| Residential products | 54% | Home standby and portable generators, ecobee thermostats, batteries, microinverters and outdoor power equipment, sold through dealers, retailers and online. |
| Commercial and industrial products | 35% | Generators from 10 kilowatts to over 3 megawatts, transfer switches, controls, mobile generators and light towers, for data centres, businesses, telecom and rental fleets. |
| Other products and services | 12% | Parts, extended warranties, service, Mobile Link subscriptions and other recurring revenue. |
Who runs it
Aaron Jagdfeld has been chief executive since 2008 and chairman since 2016; he built the home standby business and led the move into data centres. York Ragen has been finance chief as long.
Insiders own 2.4%. Half of the executives' share awards pay on sales growth, profit margin and cash, and the last three-year award missed its sales and margin goals. Insiders bought no shares in the past year.
The evidence under this - 3 claims
Why it wins, and who could beat it
Why customers come back
The protection is the dealer network. Home standby is quoted, permitted and installed by local electricians, and Generac has about 9,700 of them, nearly 400 more than a year ago. A rival with a good machine still needs installers in every town, and the main rivals are private firms that publish no comparable network.
The evidence it works: home standby sales still grew in mid-2026 with outages more than 30% below normal. It does nothing for the data-centre business.
The evidence under this - 5 claims
No rival publishes a dealer count to compare; Rehlko and Briggs & Stratton are private.
The company does not publish its outage index; the EIA's annual series runs only to 2024 (C25).
The threat
In data centres Cummins and Caterpillar sell the same kind of generator sets and both report strong data-centre demand. Abroad, other packagers use the same engine, since Generac's exclusive deal covers only the US. In homes, batteries from Tesla and Enphase can replace a generator for short outages.
The evidence under this - 4 claims
The supplier is not named in the filings, so the term of the exclusivity cannot be checked.
The rivals: TSLA, ENPH, GOOGL, REZI, CMI, CAT, ALRM, and 3 more that are private, in-house or listed abroad.
Who else does this, rival by rival
| Ticker | Rival | Where it competes | How far along | Priced against it |
|---|---|---|---|---|
| Private | Rehlko, Briggs & Stratton and Champion, private makers of home standby and portable generatorsPriced against it: Private | Residential: home standby and portable generators, ecobee thermostats, home batteries and solar | On sale | Private |
| TSLA · ENPH | Home batteries from Tesla and EnphasePriced against it: One segment only | Residential: home standby and portable generators, ecobee thermostats, home batteries and solar | On sale | One segment only |
| GOOGL · REZI | Google's Nest and Resideo's thermostatsPriced against it: One segment only | Residential: home standby and portable generators, ecobee thermostats, home batteries and solar | On sale | One segment only |
| CMI · CAT | Cummins and Caterpillar, which sell the same large generator sets to data centresPriced against it: Yes | Commercial and Industrial: generators, transfer switches and controls for data centres, businesses, telecom and rental fleets | On sale | Yes |
| MTU, the generator business of Rolls-RoycePriced against it: Listed abroad | Commercial and Industrial: generators, transfer switches and controls for data centres, businesses, telecom and rental fleets | On sale | Listed abroad | |
| Not listed | Other packagers abroad using the same engine, such as Atlas Copco, Himoinsa and AKSAPriced against it: Not listed | Commercial and Industrial: generators, transfer switches and controls for data centres, businesses, telecom and rental fleets | On sale | Not listed |
| ALRM | EnergyHub, Alarm.com's virtual power plant platformPriced against it: One segment only | ecobee grid services: utility payments for enrolled thermostats | On sale | One segment only |
| GOOGL | Google's Nest thermostatsPriced against it: Different product | ecobee grid services: utility payments for enrolled thermostats | On sale | Different product |
How big it could get
How big this could get
American data centres used about 4.4% of the country's electricity in 2023; a national laboratory expects 6.7% to 12% by 2028. That means roughly 17 to 46 gigawatts of new average load, and each gigawatt needs at least as much backup.
Nobody publishes the price of large backup sets or how much new load will use batteries instead, so the market cannot be put in dollars.
The evidence under this - 4 claims
The range is wide because chip supply, power hook-ups and buyers' spending are all uncertain; it is a scenario range, not a forecast.
Facilities are built for peak, not average, load, so installed backup may exceed this; and some new sites may use batteries, fuel cells or on-site turbines instead of diesel standby.
Who has signed
Amazon expects $2.4B of deliveries in 2027 and 2028, and holds a warrant that vests as it pays. Unfilled data-centre orders were about $4.5B at the end of September, up from $1.6B in June, with no split by customer or year. A third operator signed a framework with no volume yet. Some contracts carry cancellation rights.
The evidence under this - 5 claims
The 10-Q names cancellation rights in some data-centre contracts (C16), so backlog is not revenue yet, and $2.9B of the rise comes from one quarter's orders, of which Amazon's $2.4B is most (C15).
- 2026-10-28editedQ3 call split the backlog by year and customer (C65); the release had not.
The 10-Q names cancellation rights in some data-centre contracts (C16), so backlog is not revenue yet.
- 2026-10-28editedThe July backlog of $1.6B is a closed fact; the Q3 release puts the backlog at about $4.5B (C62).
No volume was disclosed on the call; product schedules were still being agreed.
The 8-K says deliveries are expected, not committed, and the agreement's terms are not filed in full.
The results
The numbers
Sales grew 6.1% over the twelve months to Sep 2026, faster than the −2.0% in the year to Dec 2025.
Operating margin is 10%, about where it was (9.6%) in the year to Dec 2023.
Running costs take 30¢ of each sales dollar, up from 17¢ in the year to Dec 2021.
Each dollar put into the business earns 8.7% a year after tax, down from 21% in the year to Dec 2021.
It turned 9.0% of sales into free cash, $414M over the last twelve months.
It owes $875M more than the cash it holds, about 1.5 years of the cash its operations bring in.
The share count fell 7.3% in 4.8 years, and pay in shares costs 1.1% of sales.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-11-10. Fiscal years end in December.
| Dec 2021 | Dec 2022 | Dec 2023 | Dec 2024 | Dec 2025 | 12 mo to Sep 2026 | |
|---|---|---|---|---|---|---|
| Size and growth | ||||||
| Sales | $3.74B | $4.56B | $4.02B | $4.30B | $4.21B | $4.62B |
| Sales growth | +50% | +22% | −12% | +6.8% | −2.0% | +6.1% |
| Profit | ||||||
| Gross margin | 36% | 33% | 34% | 39% | 38% | 40% |
| Operating margin | 19% | 12% | 9.6% | 12% | 6.9% | 10% |
| Net margin | 15% | 8.8% | 5.3% | 7.4% | 3.8% | 6.5% |
| What growth costs, per sales dollar | ||||||
| Research and development | 2.8% | 3.5% | 4.3% | 5.1% | 5.8% | 5.5% |
| Sales and marketing | 8.5% | 11% | 11% | 12% | 13% | 13% |
| General and admin | 3.9% | 4.3% | 6.3% | 6.6% | 10% | 8.9% |
| Pay in shares | 0.6% | 0.6% | 0.9% | 1.1% | 1.2% | 1.1% |
| Capital spending | 2.9% | 1.9% | 3.2% | 3.2% | 4.0% | 4.0% |
| Cash | ||||||
| Cash from operations | $411M | $58.5M | $522M | $741M | $438M | $601M |
| Free cash flow | $301M | −$27.7M | $393M | $605M | $268M | $414M |
| Balance sheet | ||||||
| Cash and short-term investments | $147M | $133M | $201M | $281M | $341M | $394M |
| Debt | $956M | $1.43B | $1.56B | $1.33B | $1.26B | $1.27B |
| Net cash | −$808M | −$1.30B | −$1.36B | −$1.05B | −$916M | −$875M |
| Returns and shares | ||||||
| Return on capital | 21% | 13% | 7.6% | 11% | 6.1% | 8.7% |
| Shares (diluted) | 64.3M | 64.7M | 62.1M | 60.4M | 59.3M | 59.6M |
What growth costs
Growth has been getting more expensive. Since 2021 spending on research, selling and overheads has risen much faster than sales, squeezing profit. The new growth comes from data centres, which earn about half the home margin and need new plants first.
The evidence under this - 4 claims
Management itself said tripling is worth nothing if the supply chain cannot keep up.
The last report (28 Oct)
Quarter ended September 2026, filed figures.
The quarter to September 2026 was a record: sales rose 16% to $1.29B. Commercial and industrial sales rose 28% on data-centre generators, and home sales rose 9%, a return to growth after a 2% fall in the spring. Profit before interest, tax and amortisation reached 19.8% of sales, and the release names no tariff refund this time.
Management said unfilled data-centre orders have grown to about $4.5B, kept its sales forecast and raised the low end of its margin forecast. Sales came in about 2% under expectations, though adjusted earnings per share beat by 7%. On the call, the chief executive said over half of those orders ship in 2027.
What it did to the thesisIt supported the data-centre half: orders grew sharply, most of them ship in 2027 and home sales returned to growth. Nothing is booked yet for 2028 from the first large customer, so our view holds.
The quarter in detail
Sales growth from the previous quarter, filed figures.
Earnings per share $2.54 against $2.40 expected.
The evidence under this - 4 claims
The 10-Q names cancellation rights in some data-centre contracts (C16), so backlog is not revenue yet, and $2.9B of the rise comes from one quarter's orders, of which Amazon's $2.4B is most (C15).
- 2026-10-28editedQ3 call split the backlog by year and customer (C65); the release had not.
Residential fell 2% in Q2 (C06), so the home guide needs about 16% growth in the second half (C51).
- 2026-10-28editedQ3 2026 release maintained the sales guide and raised the low end of the adjusted EBITDA margin range; Q4 arithmetic added.
The split rests on management's word on the call; the Q3 10-Q is not filed yet and the last 10-Q names cancellation rights in some data-centre contracts (C16).
What management did with the money
Insiders in 2026Q3: 11 purchases and 23 sales, from their Form 4 filings.
The price, and where we disagree
In short. Fair. Cheap on the adjusted profit the market uses, but only about fair once amortisation and pay in shares are counted, even with 2027's orders shipping.
The shares fell by a third from June's peak to September, jumped 18% when the Amazon deal was announced, then fell 23% between 9 October and 5 November, to the lowest close since the deal, with no company news behind the latest drop.
The next test is the fourth-quarter report, expected around 10 February with 2027 guidance: the forecast implies sales near $1.4B, and investors want orders for 2028 beyond Amazon's.
The evidence under this - 5 claims
- 2026-11-06editedRe-struck at the 2026-11-05 close after a further 11.0% fall with no company news; route P1 fired
The 8-K says deliveries are expected, not committed, and the agreement's terms are not filed in full.
Residential fell 2% in Q2 (C06), so the home guide needs about 16% growth in the second half (C51).
- 2026-10-28editedQ3 2026 release maintained the sales guide and raised the low end of the adjusted EBITDA margin range; Q4 arithmetic added.
Q4 2025 residential sales fell 23% (C09), so the comparison is easy, and one active storm season could lift portables and standby sharply.
How we value it
- Method
- Cash the business will produce, in three stages
- Required return
- 8%
- Long-run growth
- 4%
- Moat
- An average moat
| Bear | Base | Bull | |
|---|---|---|---|
| Sales growth, years 1 to 3 | 16%, 10%, 0% | 16%, 30%, 12% | 16%, 38%, 20% |
| Normal operating margin | 11% | 14% | 16% |
| Return on new capital after year ten | 8% | 12% | 15% |
| Value a share | $74.62 | $189 | $307 |
- Price
- $177
- Priced on
- 6 Nov
- Price as a share of the base value
- 93%
- Starting growth the price pays for
- 20.6%
- Yearly return the price offers on the base case
- 8.2%
- Risk check: valued at 10%, the rate for a cyclical or indebted business
- $115
Where each figure comes from
| Figure | Source | Where exactly |
|---|---|---|
| Sales, the last full year | Filed report | Annual report (10-K) for the period to 31 Dec 2025, filed 18 Feb 2026; the full year. |
| Operating margin today | Filed report | Quarterly report (10-Q) for the period to 30 Sep 2026, filed 3 Nov 2026; the four quarters to that date. |
| Tax rate | Filed report | Quarterly report (10-Q) for the period to 30 Sep 2026, filed 3 Nov 2026; the four quarters to that date. |
| Sales per dollar of new capital | Filed report | Annual report (10-K) for the period to 31 Dec 2025, filed 18 Feb 2026: the year's sales over its operating capital. |
| Shares | Filed report | Quarterly report (10-Q) for the period to 30 Sep 2026, filed 3 Nov 2026; the diluted share count. |
| Cash and investments | Filed report | Quarterly report (10-Q) for the period to 30 Sep 2026, filed 3 Nov 2026; the balance sheet on that date. |
| Debt | Filed report | Annual report (10-K) for the period to 31 Dec 2024, filed 19 Feb 2025; the balance sheet on that date. |
| Sales growth, years 1 to 3 | Our judgement | Our reading of the evidence, set against the analysts' forecasts; the base case below says where it follows them and where it departs. FMP's consensus of 10 to 18 analysts read 37.1%, 22.3%, 9.1%. |
| Normal operating margin | Analysts' forecasts | FMP's consensus operating margin for the year to 31 Dec 2028, the average of 18 analysts' estimates, 17.6%, less stock pay of 1.1% of sales and less the 3.4% the analysts add back on their adjusted basis: 13.1%, which the base case takes as 14%. FMP publishes how many analysts, not who they are. |
| Return on new capital after year ten | Our method's rule | Set by our reading of the moat: the 8% rate with none, 12% with an average one, 15% with a strong one; the bear case earns the rate. |
| The bear and bull cases | Our judgement | Our reading of the evidence; each case below names it. |
| Price | Market price | FMP's closing price on 6 Nov 2026. |
What each case rests on
Bear caseData-centre customers use their cancellation rights or defer (C16): 2027 grows only 10% and 2028 is flat, the lower-margin commercial mix holds the margin at 11% (C07), and new capital earns only its 9% cost.
Base case2026 grows about 16%, analysts' $4.94B (C42) after Q3 sales rose 16%; 2027 grows 30% to their $6.52B as the $4.5B data-centre backlog and Amazon's deliveries ship (C62, C15); 2028 slows to 12%; the GAAP operating margin, after stock pay and about $120M of acquisition amortisation, rises from 10.4% to 14%.
Bull caseThe backlog ships on time and more hyperscale orders follow the Amazon deal (C15, C62): 2027 grows 38% and 2028 20%, and scale lifts the GAAP margin to 16%, near 2024's 12.5% plus the refunds-free home margin.
What moves the base value
| Lower | Higher | |
|---|---|---|
| Growth in years 1 to 3, a quarter lower or higher | $162 | $220 |
| The normal margin, 15% lower or higher | $156 | $222 |
| Capital per dollar of growth, a quarter more or less | $180 | $195 |
| Return on new capital after year ten, at the rate or one step up | $145 | $207 |
| The rate, a point higher or lower | $144 | $264 |
| Long-run growth a point lower | $168 | - |
Value a share as growth and margin change
| Normal margin ↓ · growth → | 50% | 75% | 100% | 125% | 150% |
|---|---|---|---|---|---|
| 11.2% | $107 | $125 | $145 | $168 | $194 |
| 12.6% | $123 | $144 | $167 | $194 | $225 |
| 14% | $139 | $162 | $189 | $220 | $255 |
| 15.4% | $154 | $181 | $211 | $246 | $285 |
| 16.8% | $170 | $199 | $233 | $272 | $316 |
Growth is shown as a share of the base case's. Green: the price is at or under 80% of that value. Red: at or over 120%.
How we read the multiple
A buyer pays about 12 times the profit before interest, tax and amortisation that management guided for 2026, leaving out this year's one-off refunds. It also leaves out about $120M a year of acquisition charges.
- The whole company, debt included and cash taken off, is worth about $11.5B.
- Analysts expect 2027 sales of $6.52B, which needs the data-centre orders to ship.
- Valued on profit after amortisation and pay in shares, the shares are worth about $189.
Against its own history. On 9 October the shares sat well above their own six-year normal, because the market now values a data-centre supplier rather than a home-generator maker.
The evidence under this - 5 claims
Adjusted EBITDA leaves out stock pay, guided at $54M to $58M, and $118M to $122M of acquisition amortisation; on trailing GAAP earnings the vendor's price-to-earnings is 51.
- 2026-10-28editedRe-struck at the 2026-10-27 close of $193.71 (was $224.94) on the 30 September balance sheet: company value $12.37B and 13.3 times guided EBITDA before refunds (was $14.34B and 15.4 times).
- 2026-11-06editedRe-struck at the 2026-11-05 close of $172.33 (was $193.71): company value $11.11B and 11.9 times guided EBITDA before refunds (was $12.37B and 13.3 times).
- 2026-11-08editedRe-struck under the valuation rules of 2026-11-08
If 2027 sales stall near $5.5B, the same arithmetic gives about 10.6 times and the case for the price rests on 2028 orders not yet booked.
- 2026-10-28editedRe-struck with C45: company value $12.37B puts the price near 10.0 times the 2027 EBITDA analysts' sales imply (was 11.6 times).
- 2026-11-06editedRe-struck with C45: company value $11.11B puts the price near 9.0 times the 2027 EBITDA analysts' sales imply (was 10.0 times).
- 2026-10-28editedThe vendor multiples were struck at the 9 October close; the price has since fallen 14% (C64).
The adjusted EBITDA the market prices on leaves out about $120M of amortisation and $50M of stock pay; on that measure a 19% margin would give a higher value. If the bull case's 2027 and 2028 deliveries arrive, the value is about $307, above the price.
- 2026-11-09editedRe-struck on the upgraded method: every view read at 9%, the return on new capital set by the moat, and capital per dollar of sales read from the filings.
- 2026-11-09editedRe-struck at the 8% required return the owner chose after the roll-out screen: the market priced the median name on the roster at about 7.7%, so 9% on top of a 20% margin read most of it Unattractive.
- 2026-11-09editedRe-struck on the decaying fade the owner chose on 9 November: after the stated path each year closes 30% of the gap to 4%, and an analysts' margin is cut by their gap to the filed operating profit.
- 2026-11-10editedre-valued on the owner's rules of 10 November: the return on new capital glides to its terminal level over years six to ten; a US company's normal tax is the federal 21% plus state, 25%
Priced against its rivals
The competitors its own annual report names, each confirmed against the rival's filing. Measured on company value for each $1 of cash profit, cheapest first:
| Ticker | Company | company value for each $1 of cash profit |
|---|---|---|
| CMI | Cummins | 15.5× |
| GNRC | Generac Holdings | 22.8× |
| CAT | Caterpillar | 26.3× |
The middle of the other 2 is 20.9. Generac Holdings Inc. is 2nd cheapest of 3.
Rivals' figures come from a data provider, so read the spread of the group rather than any single pair.
Between Cummins and Caterpillar on value against profit, so the market prices it like the big generator makers, not above them.
Private or foreign rivals that cannot be priced: Rehlko (formerly Kohler Energy), Briggs & Stratton, Champion Power Equipment, MTU (Rolls-Royce Power Systems), Atlas Copco, Himoinsa, AKSA, IGSA, Doosan.
The bet: what it pays if it works
This bet is on one small line: utilities paying ecobee, Generac's thermostat maker, to turn enrolled thermostats down for a few hours when the grid is short of power. Generac does not report what it earns from this.
Everything else Generac makes is counted separately, at the price its expected profit already supports, and on its own it covers today's price. So the thermostat line comes free, but even a good outcome adds very little, which makes it an expensive bet. A rival platform already enrols every brand.
| ecobee grid services: utility payments for enrolled thermostats revenue in two to three years | Enrolled thermostats needed | At the stated multiple |
|---|---|---|
| $10.00M | about 297.6K at 30% kept | 0.9× |
| $25.00M (the big win) | about 744.0K at 30% kept | 0.9× |
| $50.00M | about 1.5M at 30% kept | 0.9× |
To pay 2× (a Fair bet) it would need about $5.02B of ecobee grid services: utility payments for enrolled thermostats revenue (about 149.5M enrolled thermostats at 30% kept); to pay 5× (a Cheap bet), about $18.20B of ecobee grid services: utility payments for enrolled thermostats revenue (about 541.6M enrolled thermostats at 30% kept), at the same multiple.
Each row is ecobee grid services: utility payments for enrolled thermostats alone, plus the rest of the company (Generac without ecobee grid services: generators, the rest of ecobee, batteries and data centres) counted at about $12.40B.
The steps it has to take
- Done ecobee bought with over two million connected homes
- Done Utilities paying ecobee for direct connections to enrolled thermostats
- Done Connected homes past 5.25 million, growing at a mid-teens rate
- Not started Generac discloses grid-services revenue or enrolled thermostats due 2027-02
- Not started Connected homes past seven million due 2028-06
The detail behind the label
The evidence under this - 4 claims
If 2027 sales stall near $5.5B, the same arithmetic gives about 10.6 times and the case for the price rests on 2028 orders not yet booked.
- 2026-10-28editedRe-struck with C45: company value $12.37B puts the price near 10.0 times the 2027 EBITDA analysts' sales imply (was 11.6 times).
- 2026-11-06editedRe-struck with C45: company value $11.11B puts the price near 9.0 times the 2027 EBITDA analysts' sales imply (was 10.0 times).
Where we disagree with the market
A data-centre winner with a temporarily soft home business, worth paying up for.
The data-centre orders are real, but growth is moving to the lower-margin half while the richer half stalls.
Bulls count the orders and bears point to the margin. The segment figures show both at once: sales shift toward data centres, which keep about half as much of each dollar.
The evidence against us: if commercial margins rise as volumes grow and 2028 orders arrive, profit could beat what analysts expect.
What analysts say, and the evidence
The evidence under this - 5 claims
- 2026-10-28editedThe 9 October price record is closed; the 27 October close is C64.
What could go wrong, and how we would know
The risks
- Orders cancelled or delayed. Some data-centre contracts carry cancellation rights, and a large operator delaying its sites would leave 2027 short.
- Quiet weather. Home sales depend on outages; a run of calm years leaves the higher-margin half stalling.
- Rivals catch up. Cummins and Caterpillar are adding capacity; if their waits shorten, Generac's speed stops winning orders.
- One engine supplier. Its large generators rely on one engine maker, exclusive only in the US.
- Lawsuits and regulators. Injury claims, a home-generator defect class action and an emissions inquiry into portable generators cost $142.3M of provisions in 2025.
- Amazon's warrant. Amazon can earn nearly 3% of the company as it buys, so part of the gain goes to the customer.
The evidence under this - 7 claims
The company does not publish its outage index; the EIA's annual series runs only to 2024 (C25).
Neither Cummins nor Caterpillar states its lead times; the rival figure is Generac's.
The supplier is not named in the filings, so the term of the exclusivity cannot be checked.
The 8-K says deliveries are expected, not committed, and the agreement's terms are not filed in full.
What we still do not know
- Whether orders beyond Amazon's reach $1B for 2028, and what margin the large data-centre contracts earn; the call said more than half of the $4.5B ships in 2027.
- What margin the large data-centre contracts earn, and how much a customer can cancel.
- How much new data-centre load will use diesel backup rather than batteries or on-site power.
The evidence under this - 6 claims
Generac has not named either hyperscale customer, and the 8-K does not say whether earlier orders count toward the $2.4B.
- 2026-10-28editedTested against the Q3 release: the rise in backlog fits the reading, but the release gives no split.
- 2026-10-28editedQ3 call: almost $600M of orders in the last 90 days came from non-hyperscale customers, matching the rise less Amazon's $2.4B.
The 8-K says deliveries are expected, not committed, and the agreement's terms are not filed in full.
The 10-Q names cancellation rights in some data-centre contracts (C16), so backlog is not revenue yet, and $2.9B of the rise comes from one quarter's orders, of which Amazon's $2.4B is most (C15).
- 2026-10-28editedQ3 call split the backlog by year and customer (C65); the release had not.
Neither Cummins nor Caterpillar states its lead times; the rival figure is Generac's.
Facilities are built for peak, not average, load, so installed backup may exceed this; and some new sites may use batteries, fuel cells or on-site turbines instead of diesel standby.
The split rests on management's word on the call; the Q3 10-Q is not filed yet and the last 10-Q names cancellation rights in some data-centre contracts (C16).
What would change our view
- Neutral→UnattractiveThe price runs 20% above fair value
It needs:
- ○The share price rises to about $227, 20% above what we think the business is worth.about ≈$223.01 today, -1% above the last close
- Neutral→AttractiveThe price falls 20% below fair value
It needs:
- ○The share price falls to about $151, 20% below what we think the business is worth.about ≈$153.66 today, 32% below the last close
We would also be wrong if
- Data-centre deliveries lift 2027 sales above analysts' $6.52B with the commercial margin holding.
What is coming
- 10 February 2027 Q4 2026 results and 2027 guidance: Whether sales reach the roughly $1.4B the 2026 guide implies, whether 2028 orders beyond Amazon's reach $1B, and what margin the data-centre mix gives.
Every dated event ahead
- 10 February 2027 Q4 2026 results and 2027 guidance: Whether sales reach the roughly $1.4B the 2026 guide implies, whether 2028 orders beyond Amazon's reach $1B, and what margin the data-centre mix gives.
- 31 March 2027 Belvidere plant starts production: Whether capacity arrives in time for the 2027 deliveries.
Since we wrote this
- 9 NovThesis reviewNeutral: about fair against a value of about $195Re-valued in three stages at an 8% return, the method now drawn on every Attractive thesis: the price sits within 20% of its base-case value.
- 6 NovPrice movePrice fell to 12 times guided profit; view moved to AttractiveThe route set on 28 October fired at the 5 November close. No company news moved; orders remain cancellable and 2028 orders beyond Amazon's unproven. Conviction stays Medium.
- 28 OctEarningsCall split the backlog: over half for 2027; view holdsManagement said most of the $4.5B ships in 2027 and none is booked for the first hyperscale customer in 2028; the price is still 13 times guided profit.
- 28 OctEarningsCall split the backlog: over half for 2027; view holdsManagement said most of the $4.5B ships in 2027 and none is booked for the first hyperscale customer in 2028; route P2 is not yet met and the price is still 13 times guided profit.
- 11 OctThesis reviewecobee utility programmes lifted out as an early line, and bet onThe owner's rule of 2026-10-11: a Futures winner declares the line each theory names it for, and an early line owes a bet that cannot be declined.
- 10 OctThesis writtenNeutral, medium convictionA home-generator leader turning into a data-centre supplier, at a price that assumes the 2027 ramp lands.
Also checked 1 time with nothing new to act on.
Sources 35
Sources - 35
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S01primarycompany filing2026-02-18business, products, sales by product class, distribution, customers, competition, manufacturing, penetration, legal proceedings and executive officersbacksC01C02C03C04C21C22C23C24C31C37C48C49C54
- S02primarycompany filing2026-02-112025 results, the adjusted EBITDA and EPS reconciliations, buybacks and the first 2026 guidancebacksC09C11C35C44
- S03primarycompany filing2026-03-252025 sales and adjusted EBITDA for the Residential and Commercial and Industrial segmentsbacksC05C51
- S04primarycompany filing2026-04-29Q1 2026 results by segment, the Allmand acquisition and the raised guidancebacksC08C36C44C51
- S05primarycompany filing2026-07-29Q2 2026 results by segment, tariff refunds, segment adjusted EBITDA, cash flow and the July 2026 guidancebacksC06C07C11C12C18C34C42C44C45C50C51
- S06primarycompany filing2026-08-04the June 2026 balance sheet, debt, share count, acquisitions, tariff refund receivable, legal matters and the data-centre contract riskbacksC12C16C31C34C35C36C45
- S07primarytranscript2026-07-29Second quarter 2026 earnings call transcript (FMP), held 2026-07-29data/cache/transcripts/GNRC-2026Q2.jsonmanagement on data-centre revenue and backlog, the hyperscale agreements, capacity, margins, lead times, engine supply, outages, dealers and ecobee
our backlog for products sold to the data center market now stands at $1.6 billion, representing new orders of approximately $1 billion over the last 90 days
backsC10C11C13C14C17C18C22C24C28C29C30C46C47C54 - S08primarycompany filing2026-09-16the Amazon supply agreement, expected 2027 and 2028 deliveries and the warrant termsbacksC15C46C47
- S09primarycompany filing2026-09-16the warrant's vesting on payments for generators and its exercise pricebacksC15
- S10primarycompany filing2026-04-29beneficial ownership, the 2025 bonus payout and the performance share measures and vestingbacksC37C38
- S11primarycompany filing2026-02-10Cummins Inc. Form 10-K for 2025 Cummins Inc.a rival's generator range, its data-centre demand and its Power Systems sales
We are a global OEM offering standby and prime power generators ranging from 2 kilowatts to 3.5 megawatts, as well as controls, paralleling systems and transfer switches, for customers with data center, consumer, commercial, industrial, health care, prime rental fleet and defense applications.
backsC19 - S12primarycompany filing2026-02-13Caterpillar Inc. Form 10-K for 2025 Caterpillar Inc.a rival's power generation growth from data centres and its Power and Energy sales
Power Generation - Sales increased in large reciprocating engines, primarily data center applications.
backsC20 - S13primarygovernment data2025-12-01US Energy Information Administration, Today in Energy: electricity interruptions in 2024 US Energy Information Administrationaverage outage hours per customer in 2024 against the previous decade, and the share from major events
U.S. electricity customers experienced an average of 11 hours of electricity interruptions in 2024, or nearly twice as many as the annual average experienced in the decade before
backsC25 - S14primarygovernment data2025-01-15Lawrence Berkeley National Laboratory: report on electricity demand from data centres (news release) Lawrence Berkeley National Laboratory (US Department of Energy)US data-centre electricity use in 2014 and 2023 and the range for 2028
The total data center electricity usage climbed from 58 TWh in 2014 to 176 TWh in 2023 and is estimated to increase between 325 to 580 TWh by 2028.
backsC26C27C28 - S15primarycompany filing2026-10-10Generac XBRL company facts, 2021 to June 2026, frozen as content/GNRC/financials.jsoncontent/GNRC/financials.jsonannual and trailing sales, gross profit, operating costs, operating income, cash flows, net debt and diluted sharesbacksC32C33C35C40C52
- S16secondaryvendor data2026-10-10Research pack frozen beside this thesis, 2026-10-10content/GNRC/research-pack.jsonthe 9 October close, the 252-day return and its decomposition, estimates, the rating split and its history, price-target drift, earnings history, valuation history, insider rows and FINRA short interestbacksC39C40C42C43C44C45C46C52
- S17secondaryvendor data2026-10-10daily closes around each results date and the Amazon agreementbacksC41
- S18primarycompany filing2026-10-10each open-market sale, its Rule 10b5-1 flag and plan adoption datebacksC39
- S19primarycompany filing2026-02-10Enphase Energy Form 10-K for 2025 Enphase Energy, Inc.a rival's home energy and battery products
We design, develop, manufacture and sell home energy solutions that manage energy generation, energy storage, and control and communications on one intelligent platform.
backsC48 - S20primarycompany filing2026-01-29Tesla Form 10-K for 2025 Tesla, Inc.a rival selling home energy storage
We sell energy generation systems directly to customers and also through channel partners.
backsC48 - S21primarycompany filing2026-02-20Resideo Technologies Form 10-K for 2025 Resideo Technologies, Inc.a rival's home heating and cooling controls
We are a leading player in key product markets including home heating, ventilation, and air conditioning controls; smoke and carbon monoxide detection home safety and fire suppression; and security.
backsC49 - S22primarycompany filing2026-09-16the terms of the warrant issuance, the vesting conditions and the parties' representationsbacksC15
- S23primarygovernment data2025-01-15Lawrence Berkeley National Laboratory: report on electricity demand from data centres (news release), share of US electricity Lawrence Berkeley National Laboratory (US Department of Energy)the data-centre share of US electricity in 2023 and by 2028
Data centers consumed about 4.4% of total U.S. electricity in 2023 and, depending upon how much the rest of the economy grows, are expected to consume between 6.7 and 12% of total U.S. electricity by 2028.
backsC26C27 - S24primarygovernment data2025-12-01US Energy Information Administration, Today in Energy: interruptions from major events US Energy Information Administrationoutage hours from major events in 2024 against 2014 to 2023
Interruptions attributed to major events averaged nearly nine hours in 2024, compared with an average of nearly four hours per year in 2014 through 2023.
backsC25 - S25primarycompany filing2021-11-02the ecobee purchase price and its connected homes when boughtbacksC53
- S26primaryindustry data2021-06-01NRECA, Shifting Space Conditioning Load: A Smart Thermostat Demand Response Pilot (June 2021) National Rural Electric Cooperative Associationmeasured peak reduction per home from ecobee thermostats in a co-op demand response pilot
Results of this pilot resulted in an average per-home demand reduction of 0.94 kW during summer events and 1.9 kW during winter events.
backsC55C59 - S27primaryindustry data2026-07-14PJM: 2028/2029 Base Residual Auction results (news release) PJM Interconnectionthe capacity clearing price and the shortfall against the reliability requirement
The price came in at the FERC-approved cap, $325/MW-day (UCAP) for the entire PJM footprint
backsC56C59 - S28primarycompany release2025-12-02EnergyHub acquires Resideo Grid Services (Alarm.com's EnergyHub, news release) Alarm.com Holdings (EnergyHub)a rival platform's device count and utility customers
Together,we now manage over 2.5 million DERs on one platform, allowing any utility to orchestrate any DER to meet evolving grid needs.
backsC57 - S29primaryregulator2025-12-01FERC, 2025 Assessment of Demand Response and Advanced Metering Federal Energy Regulatory Commissiondemand response enrolled in the US wholesale markets and its growth
From 2023 to 2024, total demand response participation in the seven U.S. wholesale markets increased by approximately 217 MW or 0.7 %, to a total of 33,272 MW.
backsC58 - S30primarycompany release2023-09-05what two Ontario programmes pay homeowners who enrol an ecobee thermostat
Customers will receive a $75 prepaid Mastercard for participating, as well as an additional $20 for each year they stay enrolled in the program starting in 2024
backsC59C60 - S31primarycompany filing2026-10-28Q3 2026 results by segment, adjusted EBITDA, cash flow, the September 2026 balance sheet and share count, the October 2026 guidance and management's statement that the data-centre backlog is about $4.5BbacksC11C45C47C61C62C63C66
- S32secondaryvendor data2026-10-28FMP end-of-day prices, GNRC, close of 2026-10-27, retrieved 2026-10-28The 2026-10-27 close of $193.71, the 252-session change and the year's range, the session before the Q3 results.backsC64
- S33primarytranscript2026-10-28Q3 2026 earnings call transcript (FMP), held 2026-10-28data/cache/transcripts/GNRC-2026Q3.jsonManagement's answers on the timing and customer split of the $4.5B data-centre backlog, the capacity build, the executive order on foreign-made grid components, supply-chain readiness and margins; 8 analyst questions
more than half of that backlog is for 2027
backsC47C62C65 - S34primarycompany filing2026-11-03the September balance sheet, the Amazon warrant, remaining performance obligations of about $330M, the cancellation-rights risk language and the 2 November New Way Power purchasebacksC66
- S35secondaryvendor data2026-11-06FMP end-of-day prices, GNRC, close of 2026-11-05, retrieved 2026-11-06the 5 November close of $172.33backsC45C64C66