On Wednesday, September 16, 2026, Generac Holdings disclosed in an SEC filing that it had signed a long-term supply agreement with Amazon to provide backup power generators for Amazon data centers, with initial deliveries expected to total approximately $2.4 billion across 2027 and 2028. Alongside it, Generac issued Amazon.com NV Investment Holdings LLC — a wholly owned Amazon subsidiary — a warrant to acquire up to 1,693,745 shares at $200.9266 per share, most vesting only as Amazon actually pays for generators. The stock jumped sharply in after-hours trading.
Read it as a structure story first and an order story second, as this series read Helix earlier this week. The interesting part is not just the quantity of generators. It is that the contract fuses an equipment sale to an equity mechanism — a warrant vesting in step with cumulative purchases — so Amazon’s stake grows with its spending. Like Vistra’s milestone-linked Helix commitment, this is alignment engineered into the paperwork.
Set against Generac’s size, the scale makes this a company-defining event. Generac reported 2025 net sales of $4.21 billion: commercial and industrial products $1.46 billion, residential products $2.27 billion. The expected initial Amazon deliveries of $2.4 billion across two years equal roughly 57 percent of one full recent year’s revenue — about $1.2 billion a year before additional purchases. At the September 16 close of $175.11, with roughly 59 million shares outstanding, Generac’s implied market value stood near $10.3 billion. One customer, over two years, is expected to deliver revenue equal to nearly a quarter of that market value.
A pivot already underway
The Amazon deal did not create Generac’s data-center business. It crowned a pivot already visible in quarterly results. In Q2 2026, the commercial and industrial segment reported external net sales of $556 million, up 29 percent year over year, driven by data-center demand alongside mobile product strength and international growth. Management said data-center revenue exceeded $100 million in the quarter, the data-center backlog had reached $1.6 billion as of July 2026 — including roughly $1 billion of new orders in the prior 90 days, excluding committed volumes from a second hyperscale customer since confirmed as Amazon — and raised its 2026 data-center revenue expectation to roughly $450 million. Generac had secured two multi-year hyperscale supply agreements by the second quarter.
That context disciplines the story. Generac is not a residential-generator company that woke up one morning and decided to sell to data centers. It had already spent several quarters building a hyperscale backlog and expanding megawatt-generator capacity before the Amazon agreement made the transformation legible to the market. The Amazon deal is the moment the pivot reached a scale no one could ignore, not the moment it began. Chief executive Aaron Jagdfeld said the agreement establishes Generac as a “long-term partner for the supply of industrial backup generators to Amazon,” adding that it “provides visibility to our multi-year growth expectations and our ongoing investments in vertically integrated large megawatt generator manufacturing capacity.” The Amazon volumes justify factory investment already underway; manufacturing capacity is the moat, not just the contract.
What the deal actually is
Precision matters, because the headline numbers invite overreading. The arrangement has two principal documents: a long-term supply agreement for backup generators, and a transaction agreement under which the warrant was issued. The supply agreement covers backup generators for Amazon data centers, with initial deliveries expected to total approximately $2.4 billion in 2027 and 2028. Note the word “expected.” The filing describes an expectation for initial deliveries, not a guaranteed order book with a disclosed take-or-pay minimum. Treat the $2.4 billion as Generac’s disclosed delivery expectation, not cash already contracted beyond dispute.
The $8 billion figure requires even more care. It is the ceiling on the warrant’s vesting schedule: the remaining warrant shares vest in tranches as aggregate gross payments, net of certain offsets, received by Generac and its affiliates from Amazon and its affiliates for backup generators reach cumulative thresholds extending up to $8 billion. It is not a committed order, not a revenue forecast, and not a number Amazon has promised to spend. It defines how much Amazon would have to buy for the full warrant to vest. $2.4 billion is the expectation; $8 billion is the vesting ceiling.
The warrant, decoded
A warrant is a right to buy shares at a set price, not ownership. Amazon’s warrant covers up to 1,693,745 Generac shares at $200.9266 per share. Of those, 307,954 vested immediately upon signing. The remainder vest in multiple tranches as Amazon’s cumulative qualifying payments for backup generators climb toward the $8 billion ceiling. The warrant is exercisable through September 16, 2033 — a seven-year window — in whole or in part, by cash or cashless exercise.
The exercise price deserves a pause. Generac closed the regular session on September 16 at $175.11, so the $200.9266 strike was set roughly 15 percent above the pre-announcement market price — a premium strike, not a discount. Amazon’s warrant only has intrinsic value if Generac’s shares trade above roughly $201. A strike below the market would have handed Amazon an immediate paper gain; a strike above it means Amazon profits only if Generac’s shares rise — most likely if Amazon itself keeps ordering generators.
The dilution arithmetic
Fully exercised, the warrant would give Amazon up to 1,693,745 shares — about 2.9 percent of Generac’s roughly 59 million shares outstanding, just under 3 percent on a fully diluted basis. The immediately vested tranche of 307,954 shares represents about 0.5 percent. If Amazon exercised the entire warrant for cash at the strike, Generac would receive approximately $340 million — 1,693,745 shares multiplied by $200.9266 — in exchange for issuing shares equal to just under 3 percent of the company. In a cashless exercise, Amazon would surrender part of the warrant’s embedded value to cover the strike, receiving fewer shares and paying no cash, so the route Amazon chooses will determine how many shares actually get issued and how much cash Generac receives.
The more important feature is the sequencing: the dilutive shares arrive only as the revenue arrives. The 307,954 immediately vested shares are the exception — the upfront piece Amazon earned by signing. Every share beyond that vests only because Amazon paid Generac for generators. If Amazon exercises the full warrant after paying $8 billion for equipment, Generac has booked $8 billion in revenue and issued shares equal to under 3 percent of the company. If Amazon buys little beyond the initial deliveries, most of the warrant never vests, and the dilution never materializes. Note also that the warrant’s value is leveraged to Generac’s share price: the after-hours surge carried the stock through the $200.93 strike within hours of signing, so the customer’s orders move the supplier’s stock while the customer holds an option on it.
Why Amazon would structure it this way
Amazon is not in the business of collecting generator-company warrants for their own sake. The warrant is a supply-assurance instrument dressed as equity. It gives Amazon a financial reason to keep buying — and to want Generac to succeed in serving it. A supplier in which you hold an appreciating option is a supplier you prefer over one in which you hold nothing. The pattern has precedent: Amazon has used supplier warrants before, and Oracle struck a similar equity-linked arrangement with Bloom Energy earlier in 2026, according to trade coverage. When the gating item for a data-center buildout is a physical input — generators, fuel cells, turbines — a warrant converts a vendor relationship into a partnership in which the supplier’s incentives point at your delivery schedule.
For Generac, the consideration runs the other way. The warrant cost nothing in cash, and the shares it may eventually issue are the price of a multi-year revenue stream that transforms its commercial segment. What Generac bought is visibility — a disclosed, quantified delivery expectation from the world’s largest cloud operator — and validation: Amazon’s willingness to tie equity to purchases signals to every other hyperscaler that Generac’s megawatt generators are credible.
Why standby generation is a data-center power story
It is worth being precise about what Generac actually sells into a data center, because the category is easy to misunderstand. These are backup — standby — generators. They sit idle, maintained and tested, until the grid supply fails or falters, and then they carry the facility. They do not supply day-to-day electricity or solve the interconnection problem that dominates so much data-center power coverage. They solve a different problem: continuity. For an AI training cluster, a brief outage is not an inconvenience; it can destroy a run, corrupt a checkpoint, and idle capital that costs millions of dollars a day to operate.
That is why backup power has become a data-center power story rather than a facilities footnote. As this series described in “Data-Center Power as an Asset Class,” the economics of behind-the-meter generation invert the usual logic: duplicating infrastructure for a single user is expensive per megawatt, but speed, certainty, and resilience justify the premium when the load is an AI cluster. Standby generation is the resilience leg of that logic. Generac’s product sits exactly at that intersection — on-site generation whose value grows as the grid gets slower and the workloads get more expensive to interrupt.
What Generac has to execute
The risks in this deal are execution risks, and they are substantial. The first is delivery. Roughly $2.4 billion of generators across 2027 and 2028 is an enormous manufacturing ramp for a company whose entire commercial and industrial segment sold $1.46 billion in 2025. Factory capacity, component supply, skilled labor, testing, and logistics all have to scale on a schedule Amazon’s data-center buildout will dictate, not Generac’s. The second is working capital. Equipment manufacturing consumes cash before it produces it: Generac buys components, builds machines, ships them, and collects payment on terms that may lag production by months. A multi-billion-dollar ramp means a multi-hundred-million-dollar working-capital swing, and 2025 free cash flow of $268 million, down from $605 million in 2024, shows how quickly cash conversion can tighten.
The third risk is concentration. If initial deliveries run at roughly $1.2 billion a year, a single customer would account for a large share of Generac’s total revenue for two years — and, within the commercial segment, a dominant one. Concentration cuts both ways: it is the source of the visibility the market celebrated, and the source of the vulnerability if Amazon’s data-center buildout slows, its generator specifications change, or its purchasing shifts to a competitor. The fourth is the margin question the filing does not answer. The disclosure says nothing about the pricing or profitability of the Amazon generators, and this analysis will not guess at it. The deliveries will be reported in revenue; whether they earn an acceptable return is a question for future quarters, not for the announcement.
What to watch
The warrant’s vesting schedule gives investors an unusual gift: a public, mechanical tracker of the relationship. Each tranche that vests is a disclosed signal that Amazon’s cumulative payments crossed another threshold. Watching the vesting disclosures in future filings will tell the market, quarter by quarter, whether the relationship is progressing from the $2.4 billion expectation toward the $8 billion ceiling — or stalling at the initial deliveries. The second watch item is capacity: Generac’s statements about expanding megawatt-generator manufacturing capacity will be tested against the delivery schedule. The third is the next hyperscaler: whether a third agreement follows will show whether the warrant was the start of a platform or a single-customer event.
The bottom line
Generac’s Amazon deal belongs in the same analytical frame as Helix: it is a structure story, not just a sales story. Helix fused the data center, the chip, and the power plant into one capitalized vehicle, with the generator’s equity conditioned on the power relationship materializing. Generac’s deal fuses the generator manufacturer and the data center into one aligned contract, with the customer’s equity conditioned on the equipment relationship materializing. In both cases, the old arm’s-length boundary — supplier on one side, buyer on the other, a purchase order between them — has been replaced by a structure in which each side owns a piece of the other’s outcome. For power investors, the lesson is one this series has tracked all year: the buildout is pulling every layer of the power stack into contracted, financed, aligned structures. First the PPA, then the platform stake, now the supplier warrant.
None of this is investment advice, and nothing in this analysis recommends buying or selling any security. What it does suggest is a structure worth watching: a warrant that dilutes only when revenue arrives, a delivery expectation equal to more than half a year’s sales, and a residential-generator company that has now, unambiguously, become a data-center power company. Whether that transformation earns its keep will be decided the way these things always are — delivery by delivery, megawatt by megawatt.