Residential standby was the past. Hyperscale backup power is the repricing engine — and Amazon just validated it at $2.4B firm.
The call: BUY pullbacks to $185–$205, target $300–$310, invalidation on a daily close below $165 — SPECULATIVE, 1–3 month trading / 6–12 month thesis.
Why now: Amazon signed up to $8B of Generac gensets with $2.4B firm for 2027–28 — about 29% of Generac's entire 2025 revenue ($4.2B) locked in two years.
The disagreement: Consensus still prices GNRC as a housing-linked generator stock; the backlog ($1.6B + $2.4B Amazon + a second hyperscaler pending) says it is becoming a data-center equipment compounder.
The level that changes everything: Hold above $185 and the re-rating sticks; lose $165 and the deal math breaks.
Takeaway: $2.4B firm in two years plus warrants is a strategic endorsement, not a one-off order.
After the close on Wednesday, September 16, 2026, Generac disclosed a long-term supply agreement with Amazon: up to $8B of backup generators for Amazon data centers, with $2.4B of initial deliveries in 2027–2028. Amazon's subsidiary received a warrant for up to 1.69M shares (~2.9% of shares out) at $200.93, exercisable through September 2033, vesting as purchases scale — about 308,000 shares vested immediately.
Shares spiked as much as 40–45% after hours, then traded ~$209 midday Thursday (+19.6%, high $231.89, low $202.26) vs. a $175.11 prior close, while Amazon (AMZN ~$252.37, +2.6%) barely moved. That asymmetry is the tell: for Amazon this is procurement; for Generac it is transformation.
Scale it: $1.2B/year from one customer equals ~25% of guided 2026 revenue (~$4.8–4.9B on mid-to-high-teens growth off $4.2B in 2025). Q2 2026 data-center revenue was just over $100M. Amazon alone is a ~6x run-rate step-up for that line.
Takeaway: two hyperscalers + $4B of visible backlog converts lumpy C&I sales into multi-year contracted growth.
Generac's July 29, 2026 Q2 print already showed the pivot: C&I sales +29% to $556M on data-center ramp, while Residential slipped 2% to $617M. Backlog for data-center products stood at ~$1.6B before Amazon, including ~$700M committed for 2027 from the first hyperscaler (signed June 2) plus ~$1B of add-on orders — and that excluded a second hyperscaler deal signed June 24 whose volumes were still being negotiated.
Add Amazon's $2.4B and visible data-center backlog is effectively ~$4B. That is why Baird called it a "top idea" ($305 target) and William Blair called it a "massive win" — "for frame of reference, Generac did $4.2B total revenue in 2025." Canaccord lifted its target by $100 to $375 on the multi-year durability argument.
Opinion: the market treated Generac as a storm-and-housing stock with a data-center kicker. After Amazon, the kicker is the company. Every incremental hyperscaler headline now prices as earnings power, not hope.
Takeaway: the deal alone can add ~30% to earnings power — which is why the gap is half-right and half-froth.
Run Citi analyst Vikram Bagri's math, because he is the skeptic and he is roughly right: $1.2B annual revenue at ~20% EBITDA = ~$240M EBITDA/year. Guided 2026 adjusted EBITDA is ~20–21% on ~$4.85B, or ~$1B. So Amazon adds ~24% EBITDA before any second hyperscaler, non-hyperscale colo, or residential recovery.
After tax that is roughly +$150M net (~$2.50/share on ~59.6M diluted shares) — taking a clean ~$7.50–8.00 2026 EPS run-rate toward ~$10–10.50 in 2028. At $209 that is ~20x 2028 vs. ~27x 2026. Caterpillar and Cummins sit ~15–20x with slower growth. A 25–30x multiple on $10.50 = $260–315 — exactly where Baird ($305), Citi neutral ($300), and Canaccord ($375 bull) cluster.
The warrant matters: 1.69M shares at $200.93 caps near-term dilution at ~2.9%, but $200.93 is now a magnet — Amazon is economically incentivized to see the stock hold above it. That is support and overhang at once.
Takeaway: Generac wins on packaging speed and natural-gas positioning, but execution at 3x volume is the entire risk.
Above 1MW, Caterpillar and Cummins together own the majority of global diesel genset deployments in data centers. Generac was historically sub-megawatt and residential. What changed: the Enercon acquisition (closed Q2) + Belvidere, Illinois packaging plant tripled large-megawatt packaging capacity, and hyperscalers are diversifying suppliers because Cat/Cummins lead times stretched as the data-center generator market (~$7.9B in 2026, ~4.5% CAGR) tightened.
Generac's edge is packaging velocity and gas gensets (emissions-friendlier for permitted sites facing community pushback — the exact issue William Blair flagged as a cloud-capex overhang). Its weakness is service density: Cat/Cummins have decades of global mission-critical service. A single failed acceptance test at an AWS region would travel fast.
Watch Eaton (ETN $410.83, +3.3% today), Constellation Energy (CEG $269.16, +3.7%) and Vistra (VST $145.60, +3.7%) — all green today. The market is pricing power-equipment scarcity, not just GNRC. That corroborates the thesis but also says: if grid-connection delays slow AWS builds, the whole chain sags together.
Takeaway: base case pays; bull case needs the second hyperscaler to convert.
| Scenario | Trigger | Price | Weight* |
|---|---|---|---|
| Bull: $8B paces ahead + 2nd hyperscaler | 2027 volumes confirmed + 2nd hyperscaler product terms with >$500M/yr; C&I margins >16% | $340–$375 | 25% |
| Base: $2.4B delivers, $8B paces to ~7 yrs | Q4'26/Q1'27 backlog confirms $2.4B cadence; no cancellations | $290–$310 | 50% |
| Bear: execution or capex-air-pocket | Delivery pushouts, margin <13% on mix/costs, or AWS build deferrals; tariff snap-back | $140–$165 | 25% |
*Weights are opinion, not a model output. Expected value of the midpoints ≈ $275 — above $209 but below target; hence scale-in, not all-in, on the gap.
Takeaway: cancellation rights and insider selling are the falsification checklist.
1. Contract risk is real. Generac's own Q2 filing warns of "cancellation rights, delivery requirements, and potential liability exposure" on data-center contracts. The $8B is a ceiling, not a purchase order — only $2.4B is initial-delivery volume. If AWS can walk or defer with limited penalty, today's 15x-sales intraday pricing (Citi's complaint) collapses.
2. Insiders sold the ramp. CEO Aaron Jagdfeld sold 5,000 shares at $194.89 on August 3; officers Raabe and Taffe sold into $213–256 in July–August. Routine diversification, but no open-market buys ahead of the biggest deal in company history is a yellow flag — management didn't front-run its own repricing.
3. Q2 quality was flattered. $71M of tariff refunds padded gross margin by ~6 points and EBITDA by ~2 points. Strip it and C&I EBITDA margin is ~12–13%, not 14.6%. If large gensets price at lower margin to win AWS, the $240M EBITDA assumption is high by a third.
Falsification: daily close below $165, or any 8-K disclosing Amazon volume deferral/cancellation, or C&I margin below 12% for two quarters — any one kills the thesis and I will close it as a loss.
Takeaway: the gap high ($231.89) is resistance; $200 (warrant strike) then $185 are the entries that pay.
Contents: A. Data snapshot · B. Model & load-bearing assumption · C. Sources
| Item | Level |
|---|---|
| GNRC | $209.41 (+19.6%), open $231.56, high $231.89, low $202.26, prev close $175.11 (financial-data handler, 16:00 UTC) |
| AMZN / ETN / CEG / VST | $252.37 (+2.6%) / $410.83 (+3.3%) / $269.16 (+3.7%) / $145.60 (+3.7%) |
| Market backdrop | SPY $762.35 (+1.1%), QQQ $716.55 (+1.7%) — post-Fed-hike (3.75–4.00%) rebound; XLE $64.04 flat |
| GNRC fundamentals | P/E TTM ~40x, fwd ~65x (screen; inflated by tariff-refund noise), 52-wk $134.80–$296.44, beta 1.90, mkt cap ~$12.4B at $209 on ~59.6M diluted shares |
| Q2 2026 (Jul 29) | Sales $1.174B (+11%); C&I $556M (+29%); Residential $617M (-2%); adj EPS $2.91 vs $2.01 cons (+45%); adj EBITDA $291M / 24.8% incl $71M tariff refund; backlog ~$1.6B data-center |
| Deal terms | Up to $8B total; $2.4B initial 2027–28; warrant 1,693,745 shares @ $200.9266 thru Sep 2033, ~308k vested at signing |
| Street | Baird outperform $305 top idea; Canaccord +$100 to $375; William Blair outperform "massive win"; Citi neutral $300 (valuation); Stephens/Stifel/Jefferies/Wells/Cantor buys; Barclays/Guggenheim neutrals |
| Insiders (30–60d) | CEO sale 5k @ $194.89 (Aug 3); Raabe/Taffe sales $213–256 (Jul–Aug); director share awards @ $196.17 (Jul 31); no open-market buys |
2026 base: $4.85B sales × 20.5% EBITDA = ~$995M EBITDA; net margin ~9.5% = ~$460M net ≈ $7.70/share. Amazon layer: $1.2B × 20% = $240M EBITDA, ~$150M net after tax/interest ≈ +$2.50/share → 2028 ~$10.20/share. At 28–30x = $285–305. Load-bearing: 20% incremental margin. If large-genset pricing/mix delivers only 12–13% (Q2 ex-refund C&I level), Amazon adds ~$1.50/share, fair value drops to ~$230–250 and the gap was the top. Watch Q4 C&I margin print.