Article published Oct 9, 2026. Prices below use latest available snapshots.
BE Price Structure
Loading timeframes…
Bloom Energy is doing well on its long trend and moving sideways on its short one: up 223.7% in a year and 32% above its 200-day average at the October 9 close, it has traded in a range since its S&P 500 inclusion rally in early September and sits about 20% under its 52-week high. The run has a filed cause. AI data centers bought enough onsite fuel-cell power to lift June-quarter revenue to $1.065 billion from $401 million a year earlier and turn a GAAP operating loss into $182 million of operating income (SEC 10-Q). The same filing carries the weakness: one buyer accounted for 73% of that quarter's revenue. The call is to buy a retest of $256–$273 that holds, with the October 27 report as the next test.
Data as of 2026-10-09 close: BE $280.50, RSI 54, strong-up, +1.3% vs SMA20, +13.8% vs SMA50, +32.0% vs SMA200, 1D +2.8% on half its average volume, 7D −3.0%, 30D +4.2%, 1Y +223.7%, −20.2% from the 52-week high.
Quick Snapshot (as of 2026-10-09)
| Question | Answer |
|---|---|
| The call | 🟡 Wait for the retest, then buy a close that holds it. Entry $256–$273 · stop $240 · first target $329 · medium conviction |
| What it is | A maker of solid-oxide fuel cells that turn natural gas into electricity on site without combustion. The 10-K says the systems are "designed to operate at a constant output 24x7": main power for a site, not a standby generator. |
| Why it ran | Filed demand and profit. June-quarter revenue $1.065B, +165.5% year over year; product revenue +215.4%; GAAP operating margin 17.1% against −0.9%. The 2026 revenue guide went from $3.1–$3.3B in February to $3.9–$4.2B in July. |
| Proof beyond the income statement | Customer deposits rose to $360.6M at June 30 from $78.2M at December 31 on "recently executed customer agreements"; first-half operating cash flow was $300.0M (SEC 10-Q). |
| Objection | One counterparty took 73% of June-quarter revenue, and the filing says a "customer" may be a project-finance vehicle rather than the end user. Oracle's Project Jupiter, the largest named site, has a force-majeure notice on file and a ribbon-cutting reported pushed to February 2027. The stock trades near 20 times the 2026 revenue guide. |
| Stock setup | Intact long trend in a month-long range. The $256–$273 zone holds the $265.16 support shelf and absorbed the September 28 test. |
| Next date | Third-quarter report scheduled for October 27, after the close. |
Action Matrix (as of 2026-10-09)
| Signal | If | Then |
|---|---|---|
| 🟢 Buy | A pullback into $256–$273 that holds on a closing basis. The $265.16 shelf sits inside it; the September 28 close of $262.87 recovered the next session. | Entry: the flag in the bottom → pop → flag sequence that began at the July 28 low. |
| 🟡 Wait | Price above $273, including a push through the month's high near $302 before October 27 | No chase. A breakout on the report with volume is a new plan, not this entry. |
| 🔴 Out | After an entry, a close below $240 | That sits under the 50-day average, the $247.94 value-area high and the month's low near $249: the range has failed. |
| 🟠 Target | $329, the June 18 close | Trim or trail; the 52-week high near $351 sits above. |
| 🔴 Thesis breaks | The October 27 report cuts the $3.9–$4.2B revenue guide, or Project Jupiter's fuel-cell supply is cancelled rather than delayed | The growth the stock priced is not converting. |
The Story Right Now (as of 2026-10-09)
Bloom Energy is doing well on its long trend and moving sideways on its short one: up 223.7% in a year and 32% above its 200-day average at the October 9 close, it has traded in a range since its S&P 500 inclusion rally in early September and sits about 20% under its 52-week high.
Why it ran: the filings, not the story. Bloom spent most of its public life losing money. The June 2026 quarter shows the change in four lines of the 10-Q. Product revenue was $935.4 million against $296.6 million a year earlier and supplied 96% of the year-over-year revenue increase. US revenue rose to 90% of the total from 59%. GAAP operating income was $182.2 million against a $3.5 million loss, and diluted earnings were $0.62 a share. Management set its 2026 revenue guide at $3.1–$3.3 billion on February 5 and raised it twice, to $3.4–$3.8 billion on April 28 and $3.9–$4.2 billion on July 28 (company releases filed with the SEC). It had already delivered $2.02 billion of 2025 revenue against the $1.65–$1.85 billion guide it reaffirmed in July 2025.
Who paid for it. The 10-K credits 2025's product gross-profit gain "largely" to "a major hyperscaler project facilitated through the joint venture with Brookfield." The warrant footnotes point to Oracle. The 10-K cut 2025 product revenue for "share-based consideration payable to a key hyperscaler customer," and the 10-Q identifies that warrant as Oracle's, booked as "consideration payable to a customer's customer": Oracle uses the power while another party, Bloom's customer, buys the servers. Oracle exercised it on May 1 for 2,154,231 shares worth $324.4 million. The concentration followed the money: sales to Brookfield joint ventures in which Bloom holds stakes were $373 million of first-quarter revenue, and in the June quarter one unnamed, unrelated counterparty took 73%. The filing defines that counterparty as possibly "a project-finance affiliate rather than the ultimate end user" and points to the Brookfield structure.
The market paid for more than the revenue. Revenue for the four quarters through June rose 91% on the prior four, while the stock rose 224% in a year, so the valuation multiple expanded as well: the stock trades at about 20 times the 2026 guide midpoint at an $82.6 billion market value. September added index buyers. S&P Dow Jones put Bloom in the S&P 500 effective September 21 (Bloomberg, September 4).
The last month: range, not damage. Since the July 28 close of $166.84, 52.5% off the high, the stock has run the bottom → pop sequence and spent four weeks between roughly $249 and $302. The September 28 drop of 8.95% has no cause on file. News pulled October 9 holds only a law-firm class-action notice that day, stock-event files hold nothing, and that day's captures hold an options post. The next session rose 10.8% after RBC reiterated an outperform rating and a $335 target, citing Bloom's purchase of more Fremont manufacturing space (The Motley Fool, September 29). The past week's 3.0% slip overlapped the AI-hardware selloff after OpenAI's lower revenue figure (Financial Times and CNBC, October 8); no Bloom-specific news is on file for it. Friday's 2.8% rise came on half-normal volume, and news pulled October 9 holds only quote pages for that day: cause not on file.
The setup. The trend is intact: 32% above the 200-day and first in its 19-name data-center backup-power group on both one-year return and distance above that line. The group's median one-year change is +22.7%. The open $256–$273 entry zone still fits. It brackets today's $265.16 support shelf, sits above the 50-day average, the $247.94 value-area high and the month's low, and it held the September 28 test. These levels agree with it, so the zone, the $240 stop and the $329 first target stand. A stop at $240 risks about 9% from the zone's midpoint and sits under every support the range has used.
The call: 🟡 Wait for the retest, then buy a close that holds $256–$273. Medium conviction, because the weak link is conversion. The 2026 guide needs second-half revenue of $2.08–$2.38 billion against $1.82 billion in the first half, while one buyer supplied 73% of the June quarter. At Oracle's Jupiter site, the project's ribbon-cutting was reported pushed from August 2026 to February 2027 on power-supply and permit delays (The Motley Fool, September 24). Oracle sent the site's developer a force-majeure notice on September 24 (Reuters).
What would change it. A report on October 27 that holds the guide and names a second large buyer or project upgrades the call; a guide cut, or a Jupiter cancellation rather than a delay, breaks it. A close under $240 ends the trade regardless of the story.
Extended analysis
Summary
Bloom makes the Bloom Energy Server, a solid-oxide fuel cell that converts natural gas, biogas or hydrogen to electricity electrochemically at the customer's site. Its 10-K pitches "time to power": US generator-interconnection queues totaled 2,300 gigawatts at the end of 2024, turbine makers quote multi-year delivery, and skid-mounted fuel cells install "in weeks and months rather than years, subject to permitting and site conditions," in some cases within about 90 days.
AI data centers turned that pitch into revenue. Quarterly revenue rose from $326.0 million in the March 2025 quarter to $1.065 billion in the June 2026 quarter, and GAAP operating margin went from −5.8% to 17.1%. Most buyers pay for power rather than equipment, so the server is usually sold to a financier or strategic partner (10-K). That is why Brookfield, AEP and other project-finance buyers matter as much as the end users.
The decisive uncertainty is conversion through very few hands. One counterparty took 73% of June-quarter revenue, the largest named project is delayed on power supply and permits, and the guide requires the June pace to continue for two more quarters.
Why It Ran: The Dated Record
Every row carries a filing or a dated report. Single-day moves without a source are marked.
| Date | Event | Receipt |
|---|---|---|
| Aug 2025 | Brookfield framework of "up to $5.0 billion over five years" for Bloom projects, housed in a Brookfield AI fund | 10-K |
| Oct 28, 2025 | Oracle partnership; Bloom agrees to issue Oracle a warrant for up to 3,531,073 shares at $113.28, that day's close | 10-Q, Note 3 |
| Nov 4, 2025 | $2.5 billion of 0% convertible notes due November 2030 (conversion price $194.97), with a secured revolver in December; cash rises to $2.45 billion at December 31 from $595 million at September 30 | 10-K; SEC XBRL |
| Jan 2026 | AEP order reported at $2.65 billion; the stock rose 74.2% in January | The Motley Fool, February 3 |
| Feb 5, 2026 | 2025 revenue $2.02 billion against a $1.65–$1.85 billion guide; first 2026 guide $3.1–$3.3 billion | Company release (8-K) |
| Apr 9 – May 1, 2026 | Oracle warrant issued and exercised: 2,154,231 shares worth $324.4 million | 10-Q, Note 3 |
| Apr 27, 2026 | Oracle, BorderPlex and Bloom announce fuel cells for Project Jupiter, up to 2.45 GW | Oracle announcement |
| Apr 28, 2026 | March-quarter revenue +130.4%, product revenue +208%; guide raised to $3.4–$3.8 billion | Company release (8-K) |
| Jul 28, 2026 | June-quarter revenue $1.065 billion, +165.5%; guide raised to $3.9–$4.2 billion. The stock fell 11.34% that day before the release and rose 26.5% on July 30 | Company release; 10-Q |
| Sep 4–21, 2026 | Added to the S&P 500, effective September 21 | Bloomberg, September 4 |
| Sep 24, 2026 | Oracle force-majeure notice to the Project Jupiter developer over possible power-supply delay; Oracle said the project remained on schedule | Reuters, September 24 |
| Sep 28, 2026 | Down 8.95% | Cause not on file (news, stock events and captures checked) |
| Sep 29, 2026 | Up 10.8%; RBC reiterates outperform, $335 target, citing Fremont manufacturing space | The Motley Fool, September 29 |
| Oct 9, 2026 | Up 2.8% on half-normal volume | Cause not on file (news pulled October 9) |
How the Business Makes Money
Product first, service tail second. In the June quarter, product revenue was $935.4 million of $1.065 billion (87.8%); installation $51.0 million, service $69.0 million and electricity $10.0 million (10-Q). Product gross margin was 36.5% and service 18.7%; installation ran at a small loss. Service contracts run 5 to 20 years but customers can end them annually (10-K).
Sold through financiers. "To date, the majority of our end user customers prefer to pay for the power they will consume," so Bloom sells the server to "a financier or other strategic partner" that owns it (10-K). Brookfield's AI fund holds joint ventures in which Bloom owns 9.9% (two entities) and 15% (a third); Bloom defers its share of profit on sales to them, and its maximum exposure was $68.8 million at June 30. Bloom is also using its balance sheet to close projects: in May it paid $50.0 million for option rights on one project, assigned them to a Brookfield vehicle in July, and holds a path to recover the money from Brookfield or, failing that, from Oracle (10-Q, Note 16).
Primary power, not backup. The new data-center backup-power group includes Bloom because onsite generation can carry a site until the grid arrives. The filing describes something else as the core product. The 10-K says "we primarily compete against alternative sources of electricity generation which provide firm, always on power," calls grid-connected primary power "historically our core offering," and says the systems "are designed to operate at a constant output 24x7." It lists "traditional backup equipment, such as diesel generators" as something its grid-independent configurations can make unnecessary. The 10-K does not use "bridge to grid" for its own product; the word appears only for natural gas as a policy "bridge resource." Bloom is a baseload, behind-the-meter generator that competes with reciprocating engines, small turbines and the grid. Generators, UPS and batteries in the same group sell standby.
Concentrated by design. Three buyers made up 43%, 13% and 12% of 2025 revenue, the largest a related party (10-K). The first half of 2026 shows two at 44% and 21%; the June quarter alone shows one at 73% (10-Q/A, filed July 29 to correct which period each figure described). The systems run at about 1,100 sites in nine countries; Korea, through SK ecoplant, is the second market at about 682 MW.
Capacity. Cells are printed in Fremont, California, and systems assembled in Newark, Delaware. The company plans "to double our factory capacity from 1 gigawatt to 2 gigawatts by the end of 2026" (10-K risk factors).
What the Company Says It Is Becoming
"Our long-term objective is to establish our solid oxide fuel cell technology as the standard architecture for onsite power across data centers, advanced manufacturing, critical infrastructure, and other sectors globally" (10-K). It sells into data centers across "hyperscalers, colocation operators, neocloud providers, developers, and infrastructure investors—each anchored by lighthouse customers." It lists its edges as time to power, emissions, reliability (about 99.9% fleet availability for non-redundant installations since 2020), modularity, and load-following for AI racks, with native DC output for 800-volt data-center designs.
Two filed statements matter for the stock. The 10-K says "an increasing number of transactions move from a booking to revenue in less than twelve months," which is why guidance can rise mid-year. It also says "our ability to deploy our backlog is directly tied to our ability to secure project financing, which is often an unpredictable process."
Tax support is real and conditional. The 2025 tax law made fuel cells eligible for the 30% Section 48E investment credit for projects that begin construction from 2026 through 2033, but excluded them from the domestic-content and energy-community bonuses (10-K). The 10-K says it does not expect the foreign-entity-of-concern limits on that credit to restrict its products "given the location of our supply chain," and states "our supply chain is not dependent on China," while naming China as a source of rare earth metals in tier 2 and tier 3 sub-assemblies. The securities class actions allege undisclosed Chinese scandium sourcing (GlobeNewswire, September 21–28, citing a July 8 Hunterbrook report). The allegations are unproven; buyers' credit eligibility is what they would touch.
What the Street and Public Conversation Say
Bull case. RBC reiterated outperform with a $335 target on September 29 (The Motley Fool). Zacks ranked the stock Strong Buy on September 23 and 29 on beats, the raised guide and Brookfield's framework growing to $25 billion from $5 billion.
Valuation skeptics. Zacks moved to Hold by October 7, citing a forward P/E of 109 against an industry 18, and suggested waiting for a better entry. The Motley Fool (October 2) put the trailing P/E at 380 and (September 25) questioned how fast a $20 billion backlog disclosed in February, $6 billion product and $14 billion service, can be built. That backlog is management's measure. The 10-Q's narrower remaining performance obligations were $442.4 million for product and installation plus $51.7 million for service at June 30. The accounting figure excludes much of what management counts, including financier commitments and service contracts customers can cancel annually.
Delivery skeptics. SemiAnalysis has questioned Project Jupiter's timing since May. Oracle's own risk is now public: the force-majeure notice (Reuters, September 24) and the reported slip of its ribbon-cutting to February 2027 (The Motley Fool, September 24).
Positioning. Short interest was 19.0 million shares on September 15, 6.4% of shares outstanding and 1.33 days of volume (Massive). Insiders sell and do not buy. Five officers and directors sold 54,625 shares (about $13.2 million) in Form 4s filed August 14–18. Four officers sold 11,636 shares (about $3.1 million) on September 16. Chief Commercial Officer Aman Joshi sold 8,437 shares (about $2.3 million) on October 1. The 10-Q discloses trading plans for chief executive KR Sridhar to sell up to 200,000 shares and director Jeffrey Immelt up to 60,000. No open-market purchase appears in the Form 4s captured since August.
Social. A captured October 7 post from @CKCapitalxx calls the July lows "generational entries," retail enthusiasm rather than evidence. A social-media search on October 9 returned two unrelated Hacker News items, and its YouTube and X legs timed out, so it says nothing either way.
The Conviction Chain
| Link | Evidence | Confirms | Breaks |
|---|---|---|---|
| 1. AI sites need power faster than the grid connects it | 10-K: a long grid-interconnection queue and multi-year turbine lead times; bookings converting to sales inside a year | More behind-the-meter awards; utilities such as AEP using Bloom as a channel (10-K) | Interconnection or turbine supply catches up |
| 2. Bloom wins the onsite slot | Product sales tripled in the June quarter, and customer deposits rose on newly signed agreements (see Financial Proof) | A second named hyperscaler or a large non-Oracle project on October 27 | Engines and turbines take the megawatts; deposits stall |
| 3. Orders convert on schedule | Guide raised twice since February; the June quarter cleared $1 billion | Third-quarter revenue near the June pace, guide held | Jupiter cancelled; more permit stops; guide trimmed |
| 4. Conversion is profitable | A GAAP operating profit and a widening product gross margin (see Financial Proof) | Margin holds as capacity doubles and the Oracle warrant amortizes | Oracle warrant amortization still ahead (see Margin quality); warranty costs rise |
| 5. Financing holds | Cash covers debt principal, mostly 0% converts due 2030 (see Balance sheet); Brookfield's framework reported larger | Financier purchases continue through high rates | Financiers slow; share issuance rises further |
Weakest link: 3, through concentration. Demand and margin are filed. Whether a handful of financed projects keep shipping at the June pace is not. Conviction stays medium until the October 27 report shows a second source of volume.
Financial Proof
Quarterly results, GAAP, US dollars in millions. December quarters are full year less nine months. Margins and growth are computed from the reported lines.
| Quarter (period end) | Revenue | YoY | Gross margin | Operating margin | Net income | Operating cash flow | Source |
|---|---|---|---|---|---|---|---|
| Q2 2026 (Jun 30) | 1,065.4 | +165.5% | 33.4% | 17.1% | 198.9 | 226.4 | SEC 10-Q/A |
| Q1 2026 (Mar 31) | 751.1 | +130.4% | 30.0% | 9.6% | 73.7 | 73.6 | SEC XBRL |
| Q4 2025 (Dec 31) | 777.7 | +35.9% | 30.8% | 11.3% | 1.4 | 418.1 | SEC XBRL |
| Q3 2025 (Sep 30) | 519.0 | +57.1% | 29.2% | 1.5% | −23.0 | 19.7 | SEC XBRL |
| Q2 2025 (Jun 30) | 401.2 | +19.5% | 26.7% | −0.9% | −42.2 | −213.1 | SEC XBRL |
| Q1 2025 (Mar 31) | 326.0 | +38.6% | 27.2% | −5.8% | −23.4 | −110.7 | SEC XBRL |
| Q4 2024 (Dec 31) | 572.4 | n/a | 38.3% | 18.3% | 105.2 | 484.2 | SEC XBRL |
| Q3 2024 (Sep 30) | 330.4 | n/a | 23.8% | −2.9% | −14.6 | −69.5 | SEC XBRL |
| Q2 2024 (Jun 30) | 335.8 | n/a | 20.4% | −6.9% | −61.2 | −175.5 | SEC XBRL |
What the series says. Revenue has been lumpy before. The December 2024 quarter, at $572.4 million and an 18.3% operating margin, was followed by $326.0 million and a loss. What changed in 2026 is that two consecutive quarters held above $750 million and margins widened instead of resetting. Non-GAAP earnings of $0.78 a share in the June quarter beat a $0.41 consensus (Yahoo Finance); the 2026 guide calls for about 34% non-GAAP gross margin and $2.55–$2.85 of adjusted earnings per share (company release, July 28). Those are adjusted measures, not GAAP.
Cash came from deposits. Operating cash flow for the six months to June 30, 2026 was $300.0 million, and deferred revenue and customer deposits rose $301.2 million in the same period (10-Q). The cash is customers paying ahead on new agreements: good evidence of orders, but not yet steady-state free cash flow. Capital spending rose to $77.8 million in the half from $21.5 million.
Margin quality. First-half 2026 product costs include a $58.3 million specific warranty reserve; the notes do not say what it covers, and the 10-Q's management discussion is not in the filing text on file. The half also includes a $32.4 million receivable for recoverable import tariffs previously expensed. The Oracle warrant is charged against sales as Oracle's systems deliver: $17.9 million so far, with $306.5 million still on the balance sheet.
Balance sheet and dilution. Cash was $2.67 billion on June 30 against $2.53 billion of debt principal, almost all $2.5 billion of 0% convertible notes due November 2030, convertible at $194.97, about 12.8 million shares. Shares outstanding rose 25.5% in a year to 293.4 million; diluted weighted shares were 323.3 million in the June quarter. Equity rose $692.7 million in that quarter: net income plus $311.5 million for the Oracle warrant, $126.5 million of note conversions and stock compensation. The December 2025 credit agreement is secured by "a lien on substantially all" tangible and intangible personal property other than intellectual property (10-K).
Stock Setup and What to Watch
The data-center backup-power group, ranked by one-year change at the settled close:
| Stock | Makes | As of | Price ($) | 7D (%) | 1Y (%) | vs SMA200 (%) | From 52-wk high (%) |
|---|---|---|---|---|---|---|---|
| BE | Fuel cells (primary power) | 2026-10-09 | 280.5 | -2.99 | 223.72 | 32 | -20.15 |
| FCEL | Fuel cells | 2026-10-09 | 18.03 | -2.38 | 82.49 | 23.8 | -52.4 |
| SEI | Mobile gas turbines | 2026-10-09 | 76.4 | 1.96 | 60.2 | 23 | -11.36 |
| GEV | Gas turbines, grid | 2026-10-09 | 1004.73 | 1.62 | 58.41 | 8.5 | -15.99 |
| CAT | Generator sets, engines | 2026-10-09 | 800.47 | -5.32 | 59.98 | -0.1 | -25.43 |
| GNRC | Standby generators | 2026-10-09 | 224.94 | 3.75 | 30.84 | 4.8 | -24.12 |
| CMI | Engines, generator sets | 2026-10-09 | 523.39 | -0.93 | 22.66 | -13.3 | -29.06 |
| VRT | UPS, rack power | 2026-10-09 | 242.78 | -3.73 | 43.76 | -9 | -36.1 |
| PLUG | Hydrogen fuel cells | 2026-10-09 | 1.68 | -11.58 | -55.56 | -31.6 | -61.2 |
At the October 9 close Bloom led the 19 names on one-year change and on distance above the 200-day; the group median one-year change was +22.7%. Bloom is the fuel-cell name the market paid for: FuelCell Energy was up 82.5% on the year but 52.4% under its high, and Plug Power and Ballard were in collapses.
Levels. Support sits at the $265.16 shelf, the 50-day average, the $247.94 value-area high (the top of the heaviest-traded price range) and the 200-day average. The past month's intraday range runs from about $249 to about $302. Overhead: the June 18 close of $328.91 and the 52-week high near $351. The stock is in the Momentum Board's 30–40%-above-the-200-day tier; it is not among the multi-year outperformers in pullback or the steady climbers, and it is not on the Turnaround Watch.
Dated items.
- October 27: third-quarter report, scheduled after the close. The test is revenue against the June quarter's $1.065 billion (a Zacks consensus of $1.07 billion and $0.72 of adjusted EPS as of October 7), whether the $3.9–$4.2 billion guide holds, Jupiter's status, and whether a second large counterparty appears.
- February 2027: reported ribbon-cutting for Project Jupiter, moved from August 2026 on power-supply and permit delays (The Motley Fool, September 24).
Price Data (live)
Live values from the desk's settled-close summaries; the dated sections above keep their own date's numbers.
| Metric | BE |
|---|---|
| As of (close) | 2026-10-16 |
| Price ($) | 221.31 |
| RSI (14) | 36.5 |
| vs SMA20 (%) | -18.4 |
| vs SMA50 (%) | -11 |
| vs SMA200 (%) | 2.3 |
| SMA20 ($) | 271.38 |
| SMA50 ($) | 248.62 |
| SMA200 ($) | 216.3 |
| 1D (%) | -6.34 |
| 7D (%) | -21.1 |
| 30D (%) | -18.04 |
| 3M (%) | 2.95 |
| 1Y (%) | 92.29 |
| 2Y (%) | 2025.94 |
| 1Y vs S&P 500 (pts) | 76.76 |
| From 52-week high (%) | -37 |
Risks
Concentration. One counterparty, 73% of the June quarter. If that vehicle's projects pause, the quarter has no second engine on the filed record.
Gas and permits. The systems "need a constant source of fuel such as natural gas, biogas, or hydrogen," usually from local utilities, and "delays in obtaining these approvals and permits could stall the installation process of our products and adversely affect our revenue" (10-K). Jupiter's reported power-supply and permit delays are the live example. A report relayed on X on September 27 said Texas's governor directed the state environmental agency to pause permits pending an audit; it is unconfirmed by a primary source on file.
Project size against capacity. Jupiter alone is planned larger than a full year of the factory capacity Bloom expects to have at the end of 2026.
Financing and rates. Buyers finance servers; the 10-year Treasury yield has been at or above 5% since mid-September (FRED, as cited in the October 7 market brief). Data-center financing is under strain: Firmus Grid is expected to postpone its Australian IPO (Bloomberg, October 8).
Dilution. Shares outstanding up 25.5% in a year, $2.5 billion of converts in the money at $194.97, and executive selling plans in place.
Litigation and supply chain. Securities class actions over scandium sourcing, against a 10-K that says the supply chain "is not dependent on China" but names China for rare earth metals in tier 2 and tier 3 sub-assemblies.
Crowding. A 52.5% drawdown from the high to July 28 shows how this holder base trades a disappointment.
Cross-references
- Watchlists: data-center backup power, AI power, AI bottlenecks, electricity, energy, EV and clean energy, AI scan, fuel cells and alternative data-center power, natural-gas turbines for data-center power, war sensitivity, SpaceX S-1 supply chain, tech insider buys, insider universe.
- The October 9 AI compute power vertical, "AI Money Bought Running Power Plants and Sold the Build-Out," lists Bloom as an intact trend in the backup-power group and puts its October 27 report inside the late-October capex test.
- Neocloud lane: Nebius is a named Bloom customer at Vineland, New Jersey (Nebius second-quarter shareholder letter).
- Prior work: the September 28 company profile and the April 10 deep dive (rows below).
Refresh History
| Date | Price | RSI | vs SMA20 | 30D | What changed |
|---|---|---|---|---|---|
| 2026-04-10 | $168.85 | 60.8 | +15.8% | +9.6% | Legacy dive: entry $145–$155, stop $135, target $181; cause of the run unsourced at the time. |
| 2026-09-28 | $262.87 | 52.7 | +1.5% | +24.7% | Profile: 🟡 wait for event review after the zone was reached on an 8.95% down day; zone $256–$273, stop $240, target $329. |
| 2026-10-09 | $280.50 | 54.0 | +1.3% | +4.2% | This dive: zone, stop and target kept. New from the June-quarter 10-Q/A: one counterparty at 73% of revenue, deposits up to $360.6M, cash from deposits, Oracle warrant economics, $50M project option, $58.3M warranty reserve. |
Sources
- Tape (price, RSI, averages, returns, support shelf, value-area high, market value, group ranking and median): the desk's settled-close summaries, BE row dated 2026-10-09 — data-center-backup-power (same BE row in
ai-power.json,ai-bottlenecks.json,war-sensitivity.json,thebuildout-fuel-cells-alternative-dc-power.json); 30-day intraday range from the desk's own tooling (2026-09-08 to 2026-10-09). Momentum Board tier: 2026-10-09-monster-scan (Strong tier); not on 2026-10-09-washout-scan. Prior closes (July 28, August 28, September 25, September 28, October 2): 2026-09-28-BE Research Log and 2026-09-25-be-target-entry-zone. - June-quarter filing: SEC 10-Q/A, filed 2026-07-29 — 10Q-2026-07-29.html (statements, Notes 1, 3, 6, 7, 8, 11, 16; Item 1 only, no MD&A); original 10-Q, filed 2026-07-28; 8-K results, July 28.
- Annual filing: SEC 10-K, filed 2026-02-09 — 10K-2026-02-09.html; business and risk-factor text 10k-business,
10k-risk_factors.json. - Guidance: Q4 2025 release, 8-K EX-99.1, February 5; Q1 2026 release, 8-K EX-99.1, April 28; Q2 2025 release, 8-K EX-99.1, July 31, 2025 — extracts guidance-2026-02-05,
guidance-2026-04-28.json,guidance-2025-07-31.json; Business Wire, Q2 2026 release, July 28 and 2026-07-28-be-earnings (July guide, non-GAAP EPS vs consensus, July 28–30 moves). Scheduled report: 2026-10-27-be-earnings. - Quarterly series: SEC XBRL company facts company-facts (Q2 2024 – Q1 2026; December quarters derived as full year less nine months); June 2026 quarter from the 10-Q/A statements; frozen cross-check income-statements,
cash-flow-statements.json(period ends through 2026-06-30, retrieved 2026-09-28). Ratios computed by script from these lines. - News: latest (fetched 2026-10-09T20:48Z, ok, not stale; quote pages and the July release only, nothing dated to the September 28 or October 9 moves); news-latest and
news.jsonl(pulled 2026-10-09): The Motley Fool, September 29; The Motley Fool, September 24, markets; The Motley Fool, October 2; The Motley Fool, September 25, backlog; The Motley Fool, February 3; Zacks, October 7; Zacks, October 7, SMAs; Zacks, September 23; Zacks, September 29; GlobeNewswire, September 21; GlobeNewswire, September 22; GlobeNewswire, September 28. - Oracle and Jupiter: Oracle announcement, April 27; Reuters, September 24 — both as captured for 2026-09-28-BE (2026-09-28-raw).
- Index inclusion: Bloomberg, September 4.
- AI hardware selloff and financing: CNBC, October 8; Financial Times; Bloomberg, Firmus, October 8 — via 2026-10-08-market-brief; September 29–30 and October 7 moves and lane context: 2026-09-29-market-brief,
2026-09-30-market-brief.md,2026-10-07-market-brief.md; vertical 2026-10-09-vertical-ai-compute-power-running-plants-over-the-build-out. - Texas permit pause (unconfirmed): relayed on X, September 27 — 2026-09-27-thebuildout-global-ai-compute-capacity-projected-to-hit-115-gw-by-2028-a-75e3b9b3.
- Positioning: short interest short-interest (settlement 2026-09-15), shares outstanding
massive/company.json; insider filings filing-watch (August 14–18 cluster) and captures 2026-09-18-be-4-*,2026-10-02-be-4-*.md— Form 4, Joshi, October 1, Form 4, Joshi, September 16, Form 4, Chitoori, September 16, Form 4, Kurzymski, September 16, Form 4, Soderberg, September 16; trading plans in the 10-Q/A Item 5 text. - Social: @CKCapitalxx, October 7 (2026-10-07-CKCapitalxx-still-can-t-believe-we-saw-be-in-the-150s-and-nb-19875909); sweep 2026-10-09-social-search-be-bloom-energy (YouTube and X legs timed out); captures checked for September 28 and October 9: 2026-09-28-wallstreetbets-yolo-be-bloom-energy-10-2-exp-260-strike-calls-1wsmdva, 2026-10-09-michaelsikand-wow-no-wonder-be-stock-has-been-ripping-these-fu-57143580.
- Nebius: Nebius Q2 2026 shareholder letter, SEC EX-99.2, as cited in the September 28 profile.
- Stock events checked:
research/stock-events/holds only2026-07-28-be-earnings.mdand2026-10-27-be-earnings.mdfor BE.