S&P 500 membership did not make Bloom Energy rally on a record quarter. From Thursday’s $217.77 close, the path through 31 December 2026 is a $393.27 bull, a $192.48 base, and a $119.53 bear. That replaces September’s $400 and $145, set off the $252.87 close of 4 September. The new pair is 90 daily log-returns over 48 sessions, and the mean is negative. The cash session was already down 8.70% when the 8-K was accepted at 4:09 p.m. Eastern. The filing, not the guide, is the missing piece: two customers were 43% and 26% of revenue, and operating cash flow was a $27.1 million use against $225.4 million of operating income.
The base sits 11.6% under the close even if revenue lands at $4.4 billion to $4.6 billion. The bull clears that close only with an extra 1.5 standard deviations, and it is not a street target. The closes used here are Thursday’s completed prints.
Key facts
- Thursday 22 October 2026 close: $217.77, down 8.70% from $238.53. The 20-day average is $262.07 and the 50-day is $249.16. Nasdaq daily history.
- Third-quarter revenue $1,235.8 million, up 138.1%. Operating income $225.4 million. Non-GAAP EPS $0.94. Exhibit 99.1, 22 October 2026.
- 2026 guidance raised to revenue of $4.4 billion to $4.6 billion and non-GAAP EPS of $3.10 to $3.40. Same exhibit, the third raise.
- Two customers were 43% and 26% of third-quarter revenue. The 10-Q for the quarter ended 30 September 2026 says the second is a related party, and related-party revenue was $321.6 million.
- Operating cash flow was a $27.1 million use. Equipment purchases were $153.5 million. Cash was still $2,486.7 million. Exhibit 99.1, 22 October 2026.
- Contract assets were $792.8 million, against $178.9 million at year-end 2025. The quarter added $430.3 million. Exhibit 99.1, 22 October 2026.
- The prior five sessions, 15 to 21 October 2026, traded from $213.09 to $241.37. Thursday’s low was $210.74. Nasdaq daily history.
- AEP, 14 October 2026: about $4 billion for 702 megawatts near Cheyenne, Bloom cells, a 20-year Google contract, full operation at the end of 2028.
What just happened, and why the obvious reading is wrong
The obvious reading is that holders sold a bad quarter. They did not have the quarter in the cash session. Nasdaq shows the regular-session close at $217.77, stamped 4:00 p.m. Eastern on 22 October. The SEC accepted the 8-K, accession 0001628280-26-066890, at 16:09:25 Eastern. Exhibit 99.1 sets the call for 5:00 p.m. Eastern. The down day was over before the beat was on EDGAR.
The exhibit is a beat against Bloom’s own history. Revenue was $1,235.8 million, against $519.0 million a year earlier and $1,065.4 million in the second quarter. Product revenue was $1,073.6 million, up 179.4%. GAAP gross margin was 33.8% and non-GAAP gross margin was 34.5%. GAAP diluted EPS was $0.78. Non-GAAP diluted EPS was $0.94, against $0.15. Non-GAAP operating margin was 23.1%, against 8.9%.
Edwards called it the third raise of 2026. The exhibit sets revenue at $4.4 billion to $4.6 billion and non-GAAP EPS at $3.10 to $3.40. It does not restate the old range. The Inner Circle Trading Group, at 20:07 UTC, posted the prior guide as EPS of $2.55 to $2.85 and revenue of $3.9 billion to $4.2 billion. Nine-month revenue in the 10-Q is $3,052.3 million, so the fourth quarter must land near $1.35 billion to $1.55 billion. The midpoint is 17% above the quarter just printed.
After-hours was not the verdict either. TrendSpider at 20:09 UTC posted a 6.88% extended-session decline on the beat, with EPS of $0.94 against $0.68 and sales of $1.236 billion against $1.053 billion. That claim is clocked, and it is not a close. A Nasdaq info fetch at 6:46 a.m. Eastern on 23 October showed a last sale of $212.66, off 2.35% from the Thursday close, with status still premarket. Most of that hole had filled. The sale is not the spot.
Inclusion is not Thursday’s story. Bloom Energy’s account posted at 20:08 UTC that the S&P 500 addition was already a recorded milestone. Volume of 23.6 million shares was about 1.9 times Wednesday. The decline finished before the filing.
Who is exposed
Holders of the 295,537,217 shares outstanding at 30 September carry the multiple. At $217.77 that equity value is $64.36 billion, against $2.49 billion of cash. On the quarter’s diluted share count of 324.3 million the same close is about $70.6 billion. Either way the price is a claim on future acceptances.
Two customers were about 43% and 26% of third-quarter revenue, and 43% and 23% for the nine months. The second is the related party, matching $321.6 million of related-party revenue. The 43% name is absent. That share is about $531 million of one quarter. A delay there moves the print more than a point of margin.
The 10-Q defines customer as the contractual counterparty, which may be a project-finance affiliate rather than the end user, and it points to Brookfield Asset Management joint ventures in Brookfield’s AI fund. The pages opened here do not restate a dollar cap on that framework. A large slice of revenue can sit with the financing vehicle, not the user of the power.
Oracle is named as a warrant, not as a revenue percentage. On 28 October 2025 Bloom agreed to issue Oracle up to 3,531,073 shares at $113.28 for an on-site power partnership. The warrant was issued on 9 April 2026, struck far below Thursday’s close. The exhibit and the 10-Q do not restate an older gigawatt figure, so this piece does not either. The buyer’s financing is the live question in Oracle’s backlog and the debt in front of it, and in Jupiter loans trading away from par.
Google sits behind a utility. AEP’s 14 October release puts about $4 billion into a 702-megawatt Cheyenne plant, Google buying power for 20 years, full operation at the end of 2028. The cells are Bloom’s and the dollars are AEP’s, so 2028 is not the 2026 bridge. Vistra’s contracts with Amazon and Meta take years. Bloom sells months, and the cost is concentration plus cash after acceptance.
What the tape and the primary documents show
The Nasdaq file runs from 2 June 2025 through 22 October 2026. From the 31 December 2025 close of $86.89, Thursday is up 150.6%. The 25 June high was $351.28. Nasdaq lists a 52-week range of $75.70 to $351.28.
Thursday is 16.9% under the 20-day average of $262.07 and 12.6% under the 50-day average of $249.16. The prior five sessions ran from a high of $241.37 to a low of $213.09. The close is 9.8% under that high and 2.2% above that low. The session low of $210.74 broke the prior low. The open was $228.15 and the high was $229.91, so the day finished near its low, before the news.
Two sessions stay in the sample because they dominate it. On 30 July the close rose from $163.75 to $207.12, up 26.5%, a log-return of 0.235. On 26 June it fell from $309.18 to $252.02, down 18.5%, a log-return of -0.204. Together they are about 23% of the sum of squared log-returns. Dropping them would fake a calmer sigma.
| Item | What the file or the filing says | What it is not |
|---|---|---|
| Thursday cash close | $217.77, down 8.70%, 4:00 p.m. Eastern, Nasdaq | Not the reaction to the 8-K |
| 8-K acceptance | 22 October 2026, 16:09:25 Eastern, SEC | Not a price |
| After-hours claim | Down 6.88%, TrendSpider at 20:09 UTC | Not an official close |
| Friday premarket sale | Clocked in the Nasdaq info fetch above | Not the formula spot |
| Q3 revenue and non-GAAP EPS | $1,235.8 million and $0.94, Exhibit 99.1 | Not cash earnings |
| Q3 operating cash flow | Use of $27.1 million, Exhibit 99.1 | Not a liquidity crisis |
| Customer split | 43% and 26% of Q3 revenue, 10-Q | Not a named end-user table |
| September framework | $400 bull and $145 bear on a $252.87 close | Not this piece’s title |
Product revenue was $1,073.6 million, about 87% of the quarter. Service was $67.6 million. Stock-based pay of $57.2 million is most of the gap between GAAP operating income of $225.4 million and non-GAAP operating income of $285.7 million. Neither is cash flow.
What the company and its counterparties have said
KR Sridhar, founder, chairman, and chief executive of Bloom Energy, said in the exhibit: “Electricity is the foundation of economic growth, human well-being and national competitiveness. Today, the world is moving from a century-old model of centralized power to a new architecture that includes firm, distributed onsite generation. Bloom is leading that transformation.” The paragraph names no 43% customer and says nothing about cash.
Simon Edwards, chief financial officer, stayed with the guide in that same exhibit: “Our year-to-date performance and customer demand give us confidence to raise our 2026 outlook for the third time this year.” The sentences before that cite revenue growth, wider operating margins, and investing to build capacity. “With discipline” sits next to a quarterly cash use.
Neither document names the 43% counterparty or pairs the EPS range with a cash-flow target. Nine-month operating cash flow was a $272.9 million source and equipment purchases were $231.3 million. The year is not a cash loss. The record quarter was.
Bill Fehrman, chairman, president, and chief executive of American Electric Power, said in the 14 October release: “As power demand continues to grow, AEP is bringing together new technologies and strong partnerships to deliver reliable, innovative energy solutions quickly and at scale.” AEP is promising speed to Google. Bloom supplies the cells, not the $4 billion. MarketBeat reported on 15 October that Robert W. Baird raised its target from $310 to $351 and kept outperform. That note predates the exhibit. It is not this bull. $351 is 61% above Thursday and still under $393.27.
The financing tension is the cash, not the cash balance
Cash of $2,486.7 million against recourse debt of $2,473.5 million is not a payroll problem. The tension is working capital. Contract assets went from $178.9 million at year-end to $792.8 million. The quarter added $430.3 million of contract assets, $175.2 million of receivables, and $170.4 million of inventory. Deferred revenue supplied $309.3 million and did not cover those uses, or the capital spending.
An operating use of $27.1 million plus $153.5 million of equipment purchases is a $180.6 million use, matching the $180.4 million drop in cash and restricted cash. Financing added only $5.4 million. The 10-Q says product revenue rose on acceptances, driven by time-to-power demand. Acceptance is the accounting event. Collection is later. At 67 times the $3.25 EPS midpoint, the stock assumes that lag stays short.
The 10-Q says AI-related debt has been enormous, spending has run ahead of earnings, and markets are worrying about leverage and negative free cash flow. Bloom can recognize revenue while cash slows. No server has to be cancelled for the base case to sit under the spot.
Bull, base, and bear through 31 December
The title numbers are this arithmetic only. Spot is Thursday’s close, $217.77. Mu, the mean of the last 90 log-returns from 15 June through 22 October, is -0.002572. Sigma, the sample standard deviation, is 0.06876, about 6.88% a day. N is 48 weekdays from 23 October through 31 December. October adds six sessions, 23 and 26 through 30. November adds 20, Thanksgiving on 26 November out and 27 November in. December adds 22, Christmas on 25 December out and 24 December in.
Base compounds the drift and nothing else. The exponent is 48 times -0.002572, or -0.12347. e to that power is 0.8838. Times $217.77 is $192.48, 11.6% under the close. Bull adds 1.5 standard deviations. The extra term is 1.5 times 0.06876 times the square root of 48. That square root is 6.9282, so the extra term is 0.7145. The combined exponent is 0.5911. e to that power is 1.8059. Times the spot is $393.27, 80.6% above the close. Bear subtracts one standard deviation. The exponent is -0.5998. e to that power is 0.5489. Times the spot is $119.53, 45.1% under the close. Bull is above the spot, so the no-drift fallback was not used.
Rounded, $393.27 and $119.53 are not within a dollar of the September pair of $400 and $145, or of the August pair of $360 and $130. The old street range of $286 and $70 is not reused. The pair moved because the spot moved, from $252.87 to $217.77, and because this window’s mean is negative. It did not move because a target was copied.
| Case | Price | Versus $217.77 | What has to be true |
|---|---|---|---|
| Bull | $393.27 | +80.6% | The negative daily drift keeps compounding and the path also finishes 1.5 standard deviations above it. One calm quarter does not get there. |
| Base | $192.48 | -11.6% | Only the mean. The 2026 guide can be met and the stock can still finish here, because the last 90 sessions already include the rally and still average a small daily loss. |
| Bear | $119.53 | -45.1% | The mean, minus one standard deviation over 48 sessions. Revenue does not have to fall. The multiple has to follow cash, concentration, or one session like 26 June. |
Baird’s $351, as reported by MarketBeat on 15 October, sits between the base and the bull and predates the exhibit. Thursday’s $64.4 billion is 14.3 times the $4.5 billion revenue midpoint. The bull is about 18 times that midpoint and the bear about 7.8 times. Those ratios translate the distribution. They are not Baird’s number.
What happens next
Friday 23 October is the first official close that can contain the 8-K. The filing hit at 4:09 p.m. Thursday, and the premarket sale above had already retraced most of the 6.88% after-hours claim. A Friday close still under both moving averages leaves the drift behind $192.48 intact. A close through $241.37 would not, by itself, rewrite the 90-session mean.
By 31 December the prediction is the range, not a point. Base is the drift alone. Bull needs the drift plus 1.5 sigma. Bear is the drift minus one sigma. Daily sigma of 6.88% means a session like 30 July or 26 June moves the stock by more than the entire gap from $217.77 to the base. The guide raise did not put the center of the recent tape above the close.
The test inside the window is the fourth quarter, not Cheyenne. Nine-month revenue of $3,052.3 million leaves about $1.35 billion to $1.55 billion still to book. AEP’s plant is an end-of-2028 asset, so it is not that bridge. If acceptances land in range and contract assets rise by another few hundred million, after a $430 million rise in the third quarter, the bear case needs no cancelled order. It needs the multiple to follow cash. That can show up before the 31 December close.
This is not financial advice.
Frequently asked questions
Where do the sessions through 31 December 2026 put a fresh Bloom Energy bull and bear?
From Thursday’s $217.77 official close, the 48-session distribution is a bull of $393.27, a base of $192.48, and a bear of $119.53. The September framework of $400 and $145 was struck off a $252.87 close and is not reused. The new pair is the mean and the standard deviation of the last 90 daily log-returns, not an analyst target. The base is under the close because that mean is negative.
Did the stock fall on Thursday because the quarter missed?
No. The official close of $217.77 was set at 4:00 p.m. Eastern, and the SEC accepted the 8-K at 4:09 p.m. Revenue was $1,235.8 million, up 138.1%, and non-GAAP EPS was $0.94. The guide went to $4.4 billion to $4.6 billion of revenue, the third raise this year. Thursday’s decline happened into the report. It is not a verdict on the numbers in the exhibit.
Why is the base under the close if guidance was raised?
The base ignores the press release and compounds the average daily log-return. That average, over the last 90 sessions, is -0.002572. Over 48 sessions the compound is $192.48, or 11.6% under $217.77. A higher full-year range can be achieved and the stock can still follow a drift that was already negative while the company was raising guidance. The raise changes the comparison. It does not edit the tape.
How concentrated is the revenue?
The 10-Q says two customers were about 43% and 26% of third-quarter revenue, and that the second is a related party. Related-party revenue was $321.6 million. The filing defines customer as the contractual counterparty, which may be a project-finance affiliate rather than the end user, and it points at the Brookfield structure. It does not name the 43% customer. That unnamed share is the exposure.
Is the $4 billion Google project inside the 2026 guide?
AEP said on 14 October that it plans to spend about $4 billion on a 702-megawatt Bloom fuel-cell plant near Cheyenne. Google buys the power on a 20-year contract, and full operation is targeted for the end of 2028. The $4 billion is AEP’s project capital, not a Bloom revenue line. A plant that starts full service in 2028 is not the fourth-quarter bridge inside this year’s guide.
How does this bull differ from Baird’s target?
MarketBeat reported on 15 October that Robert W. Baird raised its target from $310 to $351 and kept an outperform rating. That note predates this exhibit. The bull of $393.27 is 1.5 standard deviations above the recent daily drift, compounded to 31 December. Baird’s figure is a separate research opinion. This headline does not copy that street number.