Why This Moment Is Different

Every prior data-center headline on this name was directional. The February 2026 Enercon acquisition agreement, the March 2026 EPC Power partnership, and the June 2026 global supply agreement with an unnamed "leading hyperscale data center operator" all described an intent to serve the segment. The September 16 filing is the first one that names the counterparty, states a dollar ceiling, and attaches a vesting mechanic that only pays out if the money actually moves.

That mechanic is the part worth reading twice. The warrant tranches vest against "aggregate gross payments, net of certain offsets" received from Amazon for backup power generators for Amazon data centers. Amazon does not own the equity upside unless it writes the checks. For an outside reader this converts an announcement into a scoreboard: future filings that show tranches vesting are direct, audited evidence that revenue under the agreement is being recognized. Very few supply announcements come with a built-in progress meter.

What the filing does not say is equally important. It does not disclose pricing, margin, volumes, exclusivity, or the offsets netted out of "gross payments"; portions of both the warrant and the Transaction Agreement were redacted under Item 601(b)(10)(iv). It does not commit Amazon to buy anything beyond the initial deliveries. And it does not restate guidance — the numbers land in the 2027 and 2028 income statements, not this year's.

What Broke It Over the Quarter

The honest answer is that no filing, release or news report explains the decline. The filed quarter does, however, explain something narrower: why the headline profit overstated the run-rate.

Start with the reported lines. Revenue was $1.17 billion versus $1.06 billion a year earlier, operating income $210.4 million versus $111.8 million, diluted EPS $2.40 versus $1.25. SEC filing: “Gross profit margin was 44.5% as compared to 39.3% in the prior year second quarter.” A quarter that nearly doubled operating income does not by itself explain a 25.8% three-month drawdown.

Now read the filing's own explanation of that margin. SEC filing: “The increase was primarily driven by tariff refunds which contributed approximately 6% to gross margin growth during the quarter.” The refund is a consequence of the Supreme Court decision invalidating tariffs imposed under the International Emergency Economic Powers Act, which allowed recovery of amounts previously paid. SEC filing: “we received tariff refunds totaling approximately $61 million” SEC filing: “Approximately $71 million of this tariff recovery reduced our second quarter 2026 cost of goods sold with the remainder reflected in inventory.” A further refund receivable of roughly $28 million was recorded as probable, and the balance of the recovery sits in inventory rather than in the income statement. The same paragraph says that "unfavorable sales mix and higher input costs were partially offset by favorable price realization" — that is, the underlying margin worked against the company.

Take the recovery out and the quarter looks ordinary rather than exceptional: operating income of $139.3 million instead of $210.4 million, up 24.6% year over year rather than 88.2%, on a gross margin of 38.4% instead of 44.5% — roughly a point below the prior-year quarter. Those figures are computed from the filed lines on a pre-tax basis, because the recovery was booked as a reduction of cost of goods sold. This is not a cause of the decline; it is a receipted reason the print was a weaker signal than it looked.

What else the record shows is three things that happened around the same window.

The cohort de-rated together. On settled prices, every peer in the desk's data-center power lists is down over the same three months: Cummins −26.9%, Vertiv −27.5%, Caterpillar −19.0%, GE Vernova −16.7%, against Generac's −25.8%. When five names in one supply chain fall by roughly the same amount over the same window, the decline is a lane re-pricing rather than a company-specific break. That does not name a cause, but it does rule out "the market learned something bad about Generac specifically."

The decline started from an enthusiasm peak. The three-month window opens on June 18. On June 13 the company announced a global supply agreement with an unnamed hyperscale operator; on June 15 UBS raised its target to $335 from $305 on that agreement. The measurement window therefore starts at the top of a news-driven re-rate, which mechanically makes the drawdown look worse than the underlying business trend.

Positioning pressed it. Short interest rose from 1.93 million shares at the June 30 settlement to 2.66 million at the August 31 settlement, a 37.9% build, with days to cover rising from 1.86 to 2.49. Shorts were adding into the fall. That is also the most likely mechanical amplifier of an 8.2-times-volume, 18% up day.

Macro context, attributed rather than filed: the Federal Reserve raised its main rate by a quarter point on September 16, citing stubborn inflation. A same-day market recap adds that it was the first hike since July 2023, that the dot plot points to one further hike in 2026 and two in 2027, and that US homebuilder confidence fell to 32, its weakest since late 2022, with the 30-year mortgage above 7%. A company that sells consumer-financed home standby generators through dealers is exposed to exactly that. This is a third-party summary, not a filing, and nothing published connects it to Generac's share price.

Two further receipts frame the standing bear case, both older than this quarter: in November 2025 the company cut its full-year sales forecast citing weaker power-outage-driven residential demand, and in March 2026 it missed on the December quarter and cut 2026 guidance on the same line. The December quarter is visible in the filed series as an operating loss of $9.3 million and diluted EPS of −$0.41.

Cause verdict: partially unknown. The peer de-rating, the June starting point and the short build are documented, and the tariff recovery explains why the headline quarter flattered the run-rate. None of that is a cause: no filing, release or news report states why Generac in particular fell 25.8% over the quarter, and the quarterly report — now read in full — offers no explanation of the share price. The one remaining document that might is the Q2 earnings call transcript, which is not available.

The June Agreement — It Exists, and It Is Not This One

The June hyperscale supply agreement exists in the press record, and only there. Two syndicated stories dated June 13 describe a global supply agreement with a "leading hyperscale data center operator" for backup power, and UBS raised its target two days later on it. No filing discloses it: the current report the company filed on June 12 reported annual-meeting voting results, and the September 16 Amazon agreement is the only entry into a material definitive agreement in the company's recent filing index.

It should not be merged with the September agreement. The June coverage does not name the counterparty; the September 16 filing names Amazon and describes a supply agreement executed "on the date of issuance" of the warrant — September 16, 2026. Nothing published establishes whether the two are the same commercial relationship, an expansion of it, or separate customers. Treat them as two announcements until a filing says otherwise.

Neither is broken out in filed revenue. The June quarter's report attributes the segment's growth to "ramping revenue from products sold into the global data center market" without naming a customer or a contract, and the September agreement's deliveries are dated 2027 and 2028.

How the Business Makes Money

From the 2025 annual filing: Generac designs and manufactures power generation equipment, energy storage systems and energy-management devices for residential, commercial, data center, telecom, rental and industrial markets. It sells through what it calls an omni-channel network — independent residential dealers and contractors, industrial distributors, national and regional retailers, e-commerce partners, electrical/HVAC/solar wholesalers, equipment rental companies — plus direct sales to national accounts and end users. It employed 9,400 people at year-end 2025, roughly 4,200 of them in manufacturing.

The revenue mix is the structural fact that matters most here. Residential products were 53.9% of 2025 net sales (56.6% in 2024, 51.3% in 2023); commercial and industrial products 34.6% (32.3%, 37.2%); other products and services 11.5% (11.1%, 11.5%). The data-center business is inside that C&I bucket. An Amazon contract is a genuine change to the second-largest line, not to the largest one.

The June quarter is where those two halves are visibly moving in opposite directions. SEC filing: “Commercial & Industrial segment total sales increased approximately 29% to $556.5 million from $430.6 million in the prior year quarter, including an approximate 6% net favorable impact to sales growth from the combination of acquisitions, divestitures, and foreign currency. The core total sales growth for the segment was primarily driven by ramping revenue from products sold into the global data center market.” SEC filing: “Residential segment total sales decreased approximately 2% to $621.3 million as compared to $634.7 million in the prior year quarter. This modest sales decrease was primarily driven by lower energy storage system and portable generator shipments compared to the prior year, mostly offset by growth in home standby generator sales.” Commercial and industrial is growing on data-center product revenue; residential is flat to down, with weaker energy storage and portable generator shipments offsetting growth in home standby. That is the filed context for the Amazon agreement: it lands on the segment that is already accelerating, and it does nothing for the segment that is not.

The company's own stated growth vector is explicit about it: the filing says its "recent introduction of our large-megawatt diesel generator line-up has substantially increased our served addressable market, allowing us to participate in the supply-constrained data center market which is expected to grow significantly over the coming years due to the mass adoption of artificial intelligence." Alongside that it describes a longer-horizon strategy — natural-gas gensets, microgrids, storage, the ecobee smart-home layer, and Mobile Link subscription monitoring — aimed at a "decarbonized, digitized, and decentralized" grid.

The top filed risks read directly onto this quarter. The first risk factor in the annual filing is that "demand for the majority of our products is significantly affected by unpredictable power outage activity that can lead to substantial variations in, and uncertainties regarding, our financial results from period to period." Others name durable-goods spending by consumers, raw-material and component cost and availability, single-sourced parts, US trade policy and tariffs, foreign-sourced components, and indebtedness. A company whose own filing leads with weather-dependence is a company whose revenue line is hard to forecast, in both directions.

What the Street and the Public Conversation Say

Thin, and worth saying so plainly.

The fresh public-news pull returned five items, three of which are portal pages rather than reporting. The substantive ones: a Motley Fool and a Zacks item on the Amazon deal, a syndicated piece headlined "Generac Soars 45% on Amazon's $8B Deal," and a TipRanks summary of the June quarter describing an EPS beat with a revenue miss, growing data-center backlog, and — this is the line worth flagging — margin expansion helped by tariff refunds. An Alphastreet transcript item dated August 3 characterizes the same call as an EPS beat with a revenue shortfall, strong data-center backlog and capacity-expansion plans. Both are third-party carriers, and the filing owns the actuals — but on the tariff point the carriers were right, and the quarterly report says it more bluntly than they did: the refund contributed roughly 6 points to gross-margin growth, which is more than the entire 5.2-point jump. The non-recurring share of that quarter is not a suspicion; it is disclosed.

The social sweep was poor. Thirteen raw results collapsed to one clustered item — a YouTube trading channel explaining the after-hours breakout, which describes its own prior August call on the name as "a very low win rate or low chance of a potential bounce, very high risk versus reward." That is a promotional after-the-fact framing, not evidence, and it is the only social item the sweep returned.

The more interesting public signal is in the retail chatter itself, and it points the wrong way. In August a Discord poster asked about Generac "with all the tornados in the Midwest"; in early September another asked whether a strong El Niño winter made it an outage play. The retail conversation was trading the weather. Nobody in that record was positioned for a hyperscaler contract. The morning after the filing the same channel produced one line: "That $GNRC move is wild."

Insider Filings

The pattern in the filings is a steady, roughly monthly selling cadence through the decline — and it did not stop.

The desk's filing feed shows a three-insider sell cluster between August 4 and August 10: 6,136 shares for about $1.22 million across the CEO, Aaron Jagdfeld, the president of Home Power Generation, Kyle Raabe, and the president of Generac Home, Norman Taffe. September added more. A Form 4 filed September 2 reports Jagdfeld selling 5,000 shares for $911,050 on September 1. One filed September 3 reports a 3,932-share disposition by Taffe coded as tax withholding, not an open-market sale. One filed September 9 reports Taffe selling 400 shares for $75,481.

The Form 144 cadence is the clearer tell. Notices of intent to sell were filed on June 1, June 5, July 1, July 6, August 3, August 5, August 7, September 1 and September 8 — roughly monthly, each followed by Form 4s within days. Two more Form 144s were filed on September 17, the day of the pop.

Read this carefully. A Form 144 is a notice of intent, not a completed sale, and a monthly filing rhythm is the shape a pre-arranged trading plan produces. The filings on hand do not state that a Rule 10b5-1 plan exists, so the honest reading is: officers including the CEO have been selling steadily on a regular cadence through a 25% drawdown, in modest size, and two fresh notices arrived on the day the stock jumped 18%. No insider purchase appears anywhere in the record. The aggregate insider panel reads 0 buys against 33 sells for −$38.9 million, which is a count of transactions and not a measure of discretion.

Washout Turn or Bounce in a Downtrend

This is the reader's actual question, and the desk has a mechanical answer for it.

The washout-turn screen exists to catch exactly this shape — long-term broken, short-term turning up — and it takes four tests:

Three of four. The name fails on the 30-day, and it is not on the screen's board tonight. That is the precise, unglamorous answer: not yet a washout turn — one test short, and the test it fails is the one that measures whether the turn has lasted more than a few sessions. Two more constructive weeks would flip it. A give-back would not.

The desk's price-state screen tells the same story from the other side. Its states run collapse → basing → breakout → uptrend, where basing requires a flat 50-day, a compressed range and no fresh lows. Its stamped read on Generac is the deepest bucket — collapse, on a drawdown past 30% with price below a falling 50-day and 200-day — and that stamp is from the prior session; it has not been rewritten on tonight's close. Nothing in the settled numbers moves the name to basing: the 50-day is still drifting down and September 16 was a fresh low, four sessions ago.

The parabolic-momentum screens do not carry Generac in any tier either. It is not a monster; it is a broken name that had one very loud day.

Grading It Against the Written Rules

The desk's own archetype doc is unambiguous about what this is, and then about what to do.

Which archetype. Archetype #4, broken downtrend, takes three tests: a strong-down trend, a death cross, and price below a falling 200-day. On September 16 Generac hit the first two and half of the third: price was 16.5% below the 200-day, but that average has been rising, not falling — it has gained about four dollars over the past twenty sessions while the 50-day fell more than thirty — which is what a one-year gain of 14.3% followed by a hard quarter looks like. On tonight's settled close it hits one and a half: the death cross is intact and price is still 1.3% below the 200-day, but the trend classifier has flipped to weak-up and price closed 1.6% above the 50-day. One session does not re-classify an archetype — the doc's own re-classification triggers are the crosses and the volume base, and neither has changed. It remains a #4. The written instruction for #4 is "avoid (until it becomes #3)," with two lines that speak directly to today: "Low RSI is NOT a buy — oversold can stay oversold. A broken trend bottoms by basing, not bouncing," and "Trap: buying the dead-cat bounce. A relief rally in a death-crossed name is to be sold/avoided, not bought."

Where it sits on the written path back. The doc lays out four steps from #4 to #3. Walking them on the settled numbers:

It jumped to step four without steps one, two or three. That is the single most useful sentence in this dive. The written sequence exists because a reclaim built on a base tends to hold and a reclaim built on a headline tends to fill back in. This one is built on a headline — a good headline, a filed one, but a headline.

Where it sits in "bottom → pop → flag → breakout." The trend-hold doctrine's entry rule is "No pre-confirmation buys. Wait for bottom → pop → flag → breakout. Buying before the breakout confirms is gambling." Generac has had a candidate bottom and it has had the pop. The flag — a quiet, low-volume consolidation that holds most of the pop — has not happened, and the breakout by definition has not. Two of four steps.

Which written box applies. Not trend-hold: that box trades "durable secular uptrend names. NOT deep-value / falling-knife / bargain-bin / value traps. A 'dip' only counts if there's an existing uptrend to dip within." There is no existing uptrend here. The applicable box is base-buy, and its entry rule is one sentence long: "A coiled base is research. Set the range-high alert; there is no trade yet." That is the call.

The levels a rules-follower alerts on.

• $210.05, the 200-day. The doc's re-classification test for #4 → #3 is a golden cross restored, price above a rising 200-day, and a positive three-month return. Reclaiming this level is the first of the three.
• $203.97, the 50-day, turning up rather than down. That is written step three, and it is the one that converts a pop into a base.
• $198.78, the nearest high-volume shelf. Losing it on a close voids the reclaim and returns the stock to the low-volume void between there and the $165.11 point of control, which the desk's own profile read calls "catching the knife."
• The 30-day change turning positive — the one washout test currently failing.
• $229.09, the top of the value area, as the first place old supply is dense enough to matter.

Where the warrant strike sits. $200.9266 — the settled close sits 3.1% above it, and it sits just above the $198.78 shelf. That is a coincidence worth noticing rather than a signal: it means the contractual counterparty's own cost basis sits inside the same congestion zone the chart has to hold.

The honest nuance. Archetype #4's caricature is a dying business bouncing on nothing. Generac is not that. It is up 14.3% over one year, 44.2% over two and 88.4% over three — the five-year number is −52.4%, a hangover from the 2021 backup-power bubble, and that is the window where it has badly trailed the index. It sits in the desk's AI-power lane with a filed hyperscaler contract behind it. The archetype doc itself carves out a sixth bucket for this: "special-situation / event-driven — M&A, spinoff, binary catalyst," which "trades off the event/date, not the chart." Read Generac as a #4 chart carrying a #6 event. The rules for the chart still apply; the event is the reason to keep the alert set rather than drop the name.

The Cohort Read

A double-digit move on a tracked power name obliges a same-session read of the lane. It does not confirm the move.

The lane had an ordinary up day and Generac had an extraordinary one: the closest read-through names went nowhere. Cummins, the most direct gen-set comparison, fell 0.69% and closed at RSI 26.2, down 15.6% over 30 days; Caterpillar rose 2.0%, GE Vernova was flat at −0.02%, and Vertiv rose 0.87%. Not one of them was bid on the read-through.

That is informative in both directions. It says the market priced this as a Generac-specific contract win rather than evidence of a broader step-up in data-center backup-power demand — so nobody should buy Cummins because of this filing. It also says the lane has not been rescued: those same peers are down 17% to 27% over three months, and the pattern that broke Generac over the quarter is still intact for everyone else in it.

Financial Proof

Eight quarters of filed results. The June 2026 quarter is the best in the series on every profitability line.

Source: Massive financials (quarterly income and cash-flow statements). Margins are computed from the filed revenue, gross-profit and operating-income lines. The quarter ended December 2024 is absent from the provider's eight-quarter window — a gap in the series, not a zero. The December 2025 basic share count in that file is flagged unusable and is not printed here.

Full-year and balance-sheet figures from SEC XBRL: 2025 revenue $4.21 billion, operating income $289.2 million, net income $161.4 million, diluted EPS $2.69; operating cash flow $438.0 million and capital expenditure $169.8 million for 2025; cash and equivalents $264.9 million at June 30, 2026; inventories $1.24 billion and receivables $669.2 million at the same date.

What the series says:

• The headline operating leverage is mostly a refund. June-quarter revenue grew 10.6% year over year while reported operating income grew 88.2% and diluted EPS 92.0%. The quarterly report attributes the margin gain to a tariff recovery of roughly $71 million booked against cost of goods sold. Removing it leaves operating income of $139.3 million, up 24.6% year over year, on a 38.4% gross margin — below the prior-year quarter's 39.3%. Computed on a pre-tax basis from the filed lines; the after-tax effect on EPS is not computable from the disclosure, because no tax rate is given for the recovery.
• Underneath the refund, the mix is the story. Commercial and industrial sales rose about 29% to $556.5 million on ramping data-center product revenue, while residential fell about 2% to $621.3 million on weaker energy storage and portable generator shipments. Growth is real and it is in the segment the Amazon agreement serves; it is just not yet large enough to carry the company's margin.
• The December quarter is the scar. An operating loss of $9.3 million and −$0.41 diluted EPS, on the residential guidance cut. It is four quarters of ugly comparison that the last two quarters have walked back.
• Cash generation is steady but not large relative to the opportunity. Operating cash flow has run between $58 million and $212 million a quarter; the 2025 full-year figure was $438.0 million against $169.8 million of capital expenditure. A multi-year, multi-billion-dollar delivery commitment implies capacity spend, and the filing describes capacity expansion without sizing it.

Positioning: short interest 2.66 million shares at the August 31 settlement, 2.49 days to cover — up 37.9% from 1.93 million at the June 30 settlement. That is the most recent settlement reported and it is stale relative to tonight's tape.

The Conviction Chain

Conviction: medium. The weakest link is not the event — it is whether the chart earns an entry before the event stops being news.

1. Amazon signed a contract with a payment-linked equity kicker. Evidence: the September 16 8-K, Items 1.01 and 3.02, read from the primary document. Confirms it: tranches vesting in future filings. Breaks it: nothing short of termination; the agreement is executed.
2. The contract is material relative to the company's size. Evidence: initial deliveries of $2.4 billion across 2027–2028 against $4.21 billion of 2025 revenue. Confirms it: 2027 revenue recognition. Breaks it: delivery slippage, or pricing so thin that revenue arrives without margin — the filing discloses neither price nor margin, and portions are redacted.
3. The business was already inflecting before the contract — in the segment, not in the margin. Evidence: commercial and industrial sales up about 29% to $556.5 million, which the quarterly report attributes to ramping data-center product revenue. The margin half of the case is already broken: the filing says a tariff recovery drove the gross-margin gain, and ex-recovery margin was below the prior year. Confirms the segment claim: a September quarter that grows commercial and industrial again without a refund. Breaks it: that growth stalling once the comparison is clean.
4. The majority of revenue is still weather-and-consumer dependent. Evidence: residential at 53.9% of 2025 sales, and the annual filing's lead risk factor on unpredictable outage activity. This is the structural drag no data-center contract removes on this timescale.
5. The chart has not earned an entry. Evidence: death cross intact, price still below its 200-day, a 50-day that is still falling, three of four washout tests passed, and a reclaim that skipped the three preceding steps. Confirms a change: the 200-day reclaimed and held with the 50-day turning up. Breaks the whole setup: a close back under $198.78.

The chain is strong at the top and weak at the bottom, which is the correct shape for a watch-class name rather than a position.

Risks and Kill Tests

• The gap fills. Less than half the after-hours move survived the first full session. Losing the $198.78 shelf on a close puts Generac into the thin zone above the $165.11 point of control, where the profile read says there is no support until far lower.
• Residential demand stays weak. Two guidance cuts in ten months came from this line, and it is still the majority of sales. The Amazon contract does not deliver a dollar until 2027.
• The margin was borrowed, and the filing says so. The June quarter's gross-margin gain came from a tariff recovery of roughly $71 million against cost of goods sold. The September quarter has no such item unless further refunds land, so several points of margin come back out and the headline "operating leverage" read goes with them.
• The company filed a data-center contract-risk warning before the Amazon deal. The June-quarter report's forward-looking statements name an "increase in contract risk related to terms with certain data center customers, including cancellation rights, delivery requirements, and potential liability exposure tied to our performance obligations or other claimed damages," alongside "uncertainty regarding the growth of the data center market." That language was written six weeks before the Amazon agreement, and it describes the shape of the risk the agreement carries.
• Capacity costs money. Multi-billion-dollar deliveries require capacity the filing describes but does not size. Watch capital expenditure and inventories for the build ahead of 2027 revenue.
• Concentration cuts both ways. A named hyperscaler that can be one-fifth of a segment can also pause. The vesting mechanic that proves revenue is flowing also broadcasts when it stops.
• Insiders keep selling. No purchases appear in the filings; a monthly notice cadence that continued through the drawdown and produced two more notices on the day of the pop.

Kill tests. Closing below the $198.78 shelf kills the reclaim. A September quarter in which commercial and industrial sales stop growing — now that the tariff recovery is out of the comparison — kills the data-center demand claim. A 2027 filing showing no warrant tranches vested kills the delivery timetable. Any of the three retires the watch.

Cross-references

• Watchlists: the desk tracks Generac in its AI-infrastructure and electricity lists and in all three data-center buildout lists — gas turbines, power distribution, and UPS and storage.
• Screens on the settled close: not on the washout-turn board, not in any parabolic-momentum tier, and not in the market-wide discovery pass.
• Cohort: Cummins, Caterpillar, GE Vernova, Vertiv and Eaton are the read-through names, all down 17% to 27% over three months.
• No prior deep dive or company profile exists for this name; this is the first.

Refresh History

Sources

Primary filings and documents:

• Generac Holdings 8-K, filed 2026-09-16 — Items 1.01, 3.02, 9.01; the Amazon Transaction Agreement, warrant and supply agreement. Authenticated span receipt: GNRC-9fe9616ed430d3cd; local copy 8K-2026-09-16.html; desk capture 2026-09-16-gnrc-8-k-0001437749-26-030550.
• Generac Holdings 10-K for 2025, filed 2026-02-18 — business description, revenue mix, distribution, risk factors. Cached text: 10k-business, …/10k-risk_factors.json.
• Generac Holdings 10-Q for the quarter ended 2026-06-30, filed 2026-08-04 — segment results, the tariff-recovery note and management's discussion, plus the data-center contract-risk language. Authenticated span receipt: GNRC-67be69496c0e14ac.
• Form 4s: 2026-09-02 (Jagdfeld), 2026-09-03 (Taffe), 2026-09-09 (Taffe). Filing index and Form 144 cadence: submissions; August cluster from filing-watch.

Financial data:

• SEC XBRL company facts: company-facts, …/edgar/key-financials.json.
• Massive financials: income-statements, …/cash-flow-statements.json, …/balance-sheets.json, …/short-interest.json.

News and public conversation:

• Amazon obtains right to buy stock in Generac, boosting power company's share price by over 40% — CNBC, 2026-09-16. Capture: 2026-09-16-cnbc-amazon-obtains-right-to-buy-stock-in-generac-boosting-power--108364062.
• Generac (GNRC) Soars 45% on Amazon's $8B Deal — Vantage Markets, 2026-09-17.
• Generac Holdings (GNRC) Q2 2026 Earnings Report — TipRanks, 2026-09-17.
• Generac Holdings Q2 2026 Earnings Call Transcript — Alphastreet, 2026-08-03.
• Generac Signs Global Supply Agreement with Leading Hyperscale Data Center Operator — Oilprice via Yahoo Finance Canada, 2026-06-13 and Generac Lands Global Backup Power Deal With Major Data Center Operator — Yahoo Finance Canada, 2026-06-13.
• UBS Raises Generac Price Target to $335 from $305 — TheFly, 2026-06-15.
• Generac Cuts Full-Year Sales Forecast on Weak Residential Demand — MarketWatch, 2025-11-04 and Generac Holdings Falls After Earnings Miss and Lowered Guidance — Yahoo Finance, 2026-03-09.
• News receipt: latest and …/news/articles.jsonl.
• Why $GNRC Broke Out 40% After Hours — YouTube, 2026-09-17. Sweep capture: 2026-09-17-social-search-gnrc.
• Fed hikes main interest rate by quarter point, citing stubborn inflation — Yahoo Finance, 2026-09-16. Capture: 2026-09-16-yahoofinance-fed-hikes-main-interest-rate-by-quarter-point-citing-stubbor-e-135731525-html.
• Market recap referenced for macro context: @amitisinvesting, 2026-09-17; capture 2026-09-17-amitisinvesting-a-ton-of-things-happened-in-the-stock-market-tod-57334842.
• Desk Discord captures: 2026-08-12-dumb-money-trade-ideas-and-research-59571055-anyone-thoughts-on-generac-gnrc-with-all-the-torna, 2026-09-08-dumb-money-trade-ideas-and-research-98587915-super-nino-ravaging-the-south-this-winter-feel-lik, 2026-09-16-dumb-money-meme-stocks-03758734-that-gnrc-move-is-wild.

Price truth and screens:

• Settled 2026-09-17 summary rows: electricity, …/ai-infrastructure.json, …/thebuildout-natural-gas-turbines-dc-power.json, …/thebuildout-power-distribution-pdus-busways.json, …/thebuildout-ups-energy-storage-data-centers.json.
• Volume profile, washout-turn and price-state screens: washout-scan, regime, regimes (stamped to the 2026-09-16 bar).
• Written rules graded against: ARCHETYPES, trend-hold, base-buy.