The Generator Maker That Landed an $8 Billion Amazon Contract Is Trading Back Near Where It Started

For a decade, this was the most boring trade in industrials: buy before hurricane season, cross your fingers, sell after.

Then AI showed up and turned a generator company into an infrastructure play, because hyperscale data centers need massive backup power and even a brief outage can idle millions in hardware.

The company just signed a deal with Amazon worth up to $8 billion; the stock surged 33% overnight, gave almost all of it back within a week, and is now sitting close to where it was before the announcement.

A Q2 beat that was real, and then Amazon made it look small

Generac Holdings (NYSE: GNRC) reported Q2 adjusted EPS of $2.91 against a consensus of $1.95, a beat of nearly 50%. Revenue grew 11% to $1.17 billion.

Gross margin jumped from under 40% to above 44%. The commercial and industrial segment, the part serving data centers, grew 29% while the residential side softened slightly.

One caveat worth knowing upfront: about $71 million in tariff refunds flowed through Q2 and padded the result. The C&I momentum is genuine. Q2 just had an assist. You will see why that matters when late October arrives.

Management raised C&I growth guidance to the low 30% range for the full year and guided data center revenue to nearly $450 million this year, with the bulk landing in the back half. Q3 is expected to grow in the high teens. Q4 should accelerate further as new production capacity comes online.

Then the Amazon deal arrived

In September, Generac disclosed a long-term supply agreement to provide backup generators for Amazon data centers. Initial committed deliveries total $2.4 billion in 2027 and 2028.

Amazon also received a warrant for up to 1.69 million Generac shares at just over $200 per share, with the full warrant ceiling of $8 billion in purchases running through 2033.

The stock hit $233 in premarket the day of the announcement, then profit takers showed up and knocked it back toward $200. Late October earnings are the next moment the market has to revisit the model.

A year ago, this was a storm play. Now it has a $1.6 billion data center backlog.

The speed of this transition is genuinely unusual. Six months ago, Generac’s C&I segment was a respectable business. Today it has roughly $1.6 billion in data center backlog, built from about $1 billion in new orders over 90 days.

Before the Amazon announcement, the company had already signed two hyperscale agreements. The second, signed in June, is expected to be at least as large as the first and covers 2027 and 2028.

Management was essentially telling you on the Q2 call that they expected to “crush” their goal of doubling C&I revenue within three years, possibly as early as 2027.

None of that included Amazon.

The Amazon deal stacks on top of an already-full order book

The $2.4 billion in initial Amazon commitments is additive to the $1.6 billion backlog that was already there. The $8 billion ceiling represents the total potential relationship if Amazon’s purchases scale to that level over seven years.

Here is what changes everything about the setup: FY2027 earnings estimates of roughly $11.92 per share have already been revised up about 9% over the past few months, and they still do not fully reflect Amazon volumes.

The estimate is moving in the right direction, and it has further to go once analysts start modeling the ramp.

You can’t announce a deal that size and not face the obvious question

When a mid-cap industrial company drops an $8 billion supply agreement, the first reaction is enthusiasm. The second reaction, which tends to arrive about 48 hours later, is skepticism: can they actually pull this off?

Generac is stepping into a market where Cummins has spent decades building megawatt-scale manufacturing credibility. The home standby units that made this company famous are a completely different product from the industrial generators hyperscalers need.

Ramping to deliver $2.4 billion in committed capacity over two years, on top of the rest of the backlog, while building out new production capacity is operationally demanding. One stumble and the hyperscalers have other options.

That uncertainty is why the stock gave back the gains. The market is essentially saying: prove it.

The residential segment is a drag, not a footnote

Outages this year ran more than 30% below the long-term baseline through Q2.

The residential business softened as a result, and management trimmed the residential growth outlook. That matters because residential is still a meaningful piece of the installed base and generates ongoing service revenue.

C&I is doing all the heavy lifting right now. If C&I stumbles for any reason, there is no residential cushion to catch the fall. The market took one look at that dynamic and decided the post-Amazon surge was getting ahead of what execution data supports.

Late October is the first update.

The trailing P/E is the wrong number. Look forward.

The trailing multiple looks stretched because trailing earnings do not capture where this business is heading. The forward multiple on 2026 estimates of roughly $9.74 per share, at around $200, is about 20 times.

For a company with C&I growing 30% and a $1.6 billion backlog, that is not expensive if the ramp executes.

The FY2027 consensus near $11.92 per share suggests roughly 22% earnings growth next year. That estimate does not yet fully reflect the Amazon agreement. Analyst targets cluster between $285 and $340, with JP Morgan maintaining Overweight after the deal and UBS out at $340.

The question you have to answer is simple: do you believe the execution story? If yes, you are buying a $12 billion market cap company with multi-year revenue visibility that is not in estimates yet. If no, you are buying a high-beta industrial that has already moved a lot on narrative alone.

The backlog makes this more than pure narrative. That is the difference from six months ago.

Q2 gross margin of 44.5% was the headline. The question is whether it holds without the tailwind.

About $71 million in tariff refunds flowed through Q2 and helped drive the margin expansion from under 40% to above 44%. That is a one-time benefit. Q3 will not have it.

If gross margin holds near 44% in Q3 without the tariff refund, that is genuine structural improvement. The C&I mix shift toward higher-margin industrial units is working.

If margins slip back toward 40% as the refund rolls off, some of the profitability case gets complicated. You will know the answer in late October. That is your first signal on whether the margin expansion thesis is structural or just Q2 timing.

Watch the gross margin line before anything else on the Q3 print.

Execution is the whole game. If lead times slip, if the new production capacity is not ready, or if quality issues show up at megawatt scale, Amazon and other hyperscalers will go elsewhere. They are not captive customers. The backlog is a commitment, not a guarantee, and the moment execution wobbles, the multiple compresses fast in a name with a beta near 2.

The residential drag could deepen. Outages are running well below historical averages this year. If that continues and storm season stays quiet, the residential business keeps softening with no C&I offset large enough to absorb the narrative hit. Any C&I softness on top of that is a double negative.

The warrant creates dilution optionality. Amazon holds a warrant for up to 1.69 million shares at just over $200. If the stock rises significantly and Amazon exercises, that is real dilution. Not the main risk, but worth tracking for long holders.

The backlog is concentrated in a few large relationships. If one hyperscale partner delays delivery windows or renegotiates, the revenue timeline shifts in a way that hits estimates hard. A $1.6 billion backlog built mostly on two or three agreements is not the same as a diversified order book.

The honest version of this setup is that the best entry was before the Amazon deal. You missed that. So did most people.

What you have now is a stock that ran 33% on the news and gave it all back, sitting on a $1.6 billion backlog with a $2.4 billion Amazon commitment layered on top, and FY2027 estimates that have not caught up to the reality yet.

The pullback happened because execution risk is real, not because the thesis broke.

If late October earnings show that C&I volumes are on track, gross margins are holding near 44% without the tariff refund assist, and management updates the delivery timeline for the Amazon agreement without pushback, this stock moves toward analyst targets between $285 and $340.

If the ramp slips or margins compress in Q3, get out fast. Beta near 2 means it punishes quickly.

Accumulate around current levels with a tight plan. The late October print is not a passive hold event. It is the first real test of whether this story survives contact with execution.

Looking to buy? Around current levels is a reasonable entry after the post-deal surge and pullback. Size conservatively because late October is a binary event. Do not add ahead of the print without knowing your exit plan if margins disappoint.

Already own it? Hold through late October and watch exactly two things: gross margin direction without the tariff refund, and any update on the Amazon delivery timeline. Both need to be on track for the thesis to stay intact.

Main risk to respect: Execution. The ramp from home standby units to megawatt-class hyperscale generators at $2.4 billion per year in committed deliveries is a different manufacturing challenge entirely. One stumble reprices the stock in a hurry.

That's our coverage for today; thanks for reading! Reply to this email with feedback or any value stocks you want me to check out.

Best Regards,
—Noah Zelvis
Undervalued Edge