AHNIM STUDIO · Company research · BE
Business-source review: October 11, 2026. Valuation and analyst-comparison update: October 12, 2026 (Asia/Seoul). Latest financial period reviewed: June 30, 2026. All dollar amounts are US dollars.
Bloom Energy gives investors a direct way to examine a practical constraint on AI infrastructure: how a data center gets power. The investment question extends beyond demand for equipment. It includes the margin on each installation, the cash required to deliver it and the number of shares that ultimately divide the proceeds.
Bloom sells stationary fuel-cell systems that generate electricity at a customer’s site. For data centers, the appeal is a source of on-site power; the economics still depend on fuel, equipment, service and operating conditions. [10]
The operating business has improved. Our October 12 valuation review sets conditional end-2027 Bear / Base / Bull targets of $71 / $216 / $377 using 2028 adjusted EBITDA. Our $216 Base is 24.1% below the analyst mean of $284.74, triggering our 20% divergence review. Section 5 explains the forecast, capital assumptions, horizon mismatch and dated changes.
1. The operating improvement is measurable
Quarterly revenue rose from $401.2 million in Q2 2025 to $1,065.4 million in Q2 2026, while GAAP gross margin increased from 26.7% to 33.4%. The intervening quarters show both growth and variation. [1] [2]
| Quarter | Revenue (US$m) | GAAP gross margin |
|---|---|---|
| Q2 2025 | 401.2 | 26.7% |
| Q3 2025 | 519.0 | 29.2% |
| Q4 2025 | 777.7 | 30.8% |
| Q1 2026 | 751.1 | 30.0% |
| Q2 2026 | 1,065.4 | 33.4% |
A higher margin alongside greater volume is encouraging because it improves the potential return on the manufacturing base. Sustaining it will matter more than any single quarter. Product mix, installation timing and service economics can all change the relationship between revenue and profit.
On July 28, management guided to 2026 revenue of $3.9–4.2 billion and roughly 34% non-GAAP gross margin. These are company forecasts. The adjusted margin forecast is not directly interchangeable with the GAAP history above. [1]
2. Read the Oracle numbers by date and scope
Several widely discussed capacity figures describe different aspects of the same relationship. Adding them would inflate the apparent opportunity.
| Disclosure | Figure | What it describes |
|---|---|---|
| April 13, 2026 | Up to 2.8 GW; initial 1.2 GW contracted | Expanded agreement and the initially contracted portion on that date. |
| April 27, 2026 | Up to 2.45 GW | Project Jupiter site power design within the expanded relationship. |
| September 24, 2026 | 2.4 GW | Bloom’s statement reaffirming Oracle’s contract commitment, as quoted in Stocktwits reporting. |
The commercial test is delivery against a schedule that customers can actually use. A capacity agreement still has to become manufactured equipment, a permitted and supplied site, customer acceptance and cash collection. Public statements alone do not establish that every project milestone has been met.
Reuters reported on September 24 that Oracle issued a force-majeure notice citing possible power delays, while Oracle said the project remained on schedule. The notice alone does not establish a delay or a cancelled Bloom contract. [12]
Concentration raises the stakes. The corrected filing identifies one contractual counterparty as 73% of Q2 revenue. “Counterparty” does not necessarily mean the ultimate user, so that percentage should not automatically be labelled Oracle exposure. [4]
3. Cash improved, with meaningful help from customer funding
First-half operating cash flow moved into positive territory, as the table shows. Deferred revenue and customer deposits contributed a positive $301.2 million change in the H1 2026 cash-flow statement. [3]
| US$ millions | H1 2025 | H1 2026 |
|---|---|---|
| Operating cash flow (CFO) | −323.8 | 300.0 |
| Cash PP&E purchases | 21.5 | 77.8 |
| CFO less cash PP&E | −345.3 | 222.2 |
Advance customer funding can make expansion easier to finance. Its timing also makes a single period a weak guide to recurring cash generation. We do not simply subtract the deposit change and call the result “normalized cash flow”: working-capital movements are connected to deliveries, inventories and payment terms.
The derived measure above excludes other uses of cash, including joint-venture investment and financing repayments. It should be read as a transparent calculation with a limited scope. The next useful evidence is whether cash generation remains positive as funded projects are delivered.
4. Expansion announcements need an economic bridge
Fremont’s September newsletter describes Bloom as having “acquired” a 158,000-square-foot facility. The September 30 Silicon Valley Business Journal headline and accessible excerpt describe a 158,500-square-foot lease. The full article is restricted and was not reviewed. Because the public descriptions differ, we treat this as additional facility space secured for expansion, not a verified land purchase. The 2025 annual filing had described a plan to reach a 2 GW annual manufacturing run rate by the end of 2026. Neither the facility report nor that plan establishes the new building’s commissioned output today. [15] [9] [10]
Brookfield’s expanded financing framework of up to $25 billion could support eligible projects. That headline amount is not Bloom product revenue or a funded order book. [8]
We will judge expansion through usable capacity, equipment utilization, capital spending and margins. A larger footprint earns its value when it can serve profitable demand. Financing access helps, but the terms and allocation of project economics still matter.
Fuel choice and permits also belong in the operating analysis. Oracle’s September announcement concerns carbon-capture research; it does not demonstrate commercial capture at these sites. Gas-fuelled generation should not be described as carbon-free on that basis. [11]
5. Valuation update and analyst comparison
Updated October 12, 2026: Bear $71 / Base $216 / Bull $377. Model AHNIM-BE-M3-forward-EBITDA values FY2028 earnings at December 31, 2027. These are conditional market-valuation scenarios, not present intrinsic values, guaranteed prices or 12-month targets. The company has not guided to our 2027–2028 forecasts. No probabilities are assigned to the cases.
20% divergence flag: our Base target is 24.1% below the analyst mean. The mean is $284.74, median $303 and published range $97–390. Stock Analysis identifies 29 analysts; its displayed update date is October 7, checked October 12, 2026. Investing.com also shows $284.74. These are generally rolling 12-month targets, whereas ours is end-2027; the comparison is a screening flag, not a fully horizon-matched valuation. [V1] [V2]
| Scenario | AHNIM target | Vs. analyst mean $284.74 | Vs. median $303 |
|---|---|---|---|
| Bear | $71 | -75.1% | -76.6% |
| Base | $216 | -24.1% | -28.7% |
| Bull | $377 | +32.4% | +24.4% |
Forecasts, formula and reproducible targets
| AHNIM assumption | Bear | Base | Bull |
|---|---|---|---|
| 2027 revenue, US$m | 4,500 | 6,500 | 8,000 |
| 2028 revenue, US$m | 5,250 | 8,500 | 11,250 |
| 2028 adjusted EBITDA margin | 20% | 27% | 31% |
| 2028 adjusted EBITDA, US$m | 1,050 | 2,295 | 3,487.5 |
| EV / adjusted EBITDA | 22× | 30× | 35× |
| Equivalent EV / 2028 sales | 4.40× | 8.10× | 10.85× |
| Enterprise value, US$m | 23,100 | 68,850 | 122,062.5 |
| End-2027 cash incl. option proceeds, US$m | 2,441.172997 | 3,041.172997 | 3,541.172997 |
| Remaining face debt, US$m | 2,500 | 0 | 0 |
| Other claims deducted, US$m | 239.899 | 239.899 | 239.899 |
| Economic diluted claims, millions | 319.636729 | 332.459229 | 332.459229 |
| Calculated value per share | $71.334962 | $215.518980 | $377.080144 |
| Published target | $71 | $216 | $377 |
Target = (2028 revenue × adjusted EBITDA margin × EV/EBITDA + end-2027 cash − remaining debt − other claims) / economic diluted claims.
Bear: (5,250 × 20% × 22 + 2,441.172997 − 2,500 − 239.899) / 319.636729 = $71.33.
Base: (8,500 × 27% × 30 + 3,041.172997 − 0 − 239.899) / 332.459229 = $215.52.
Bull: (11,250 × 31% × 35 + 3,541.172997 − 0 − 239.899) / 332.459229 = $377.08. Calculations use unrounded inputs.
The 2026 company revenue guide is $3.9–4.2 billion. Our 2027 growth over its $4.05 billion midpoint is 11.1% / 60.5% / 97.5%; 2028 growth is 16.7% / 30.8% / 40.6%. Bear assumes slower deployment and margin pressure, not insolvency or a guaranteed downside floor. Base assumes major orders become shipments and customers are ready to accept equipment. Bull additionally requires faster production, more customers and stronger economics. The forecasts are top-down AHNIM scenarios, not a customer-by-customer contracted revenue schedule. [V7]
The margin assumptions imply adjusted gross margins of 32% / 36% / 39%, adjusted operating costs of 14% / 11% / 10% of sales, and depreciation and amortization of 2% of sales. Thus adjusted EBIT margins of 18% / 25% / 29% become EBITDA margins of 20% / 27% / 31%. This is a consistency bridge, not independent proof of the forecasts. Adjusted EBITDA excludes stock compensation and is not GAAP profit or free cash flow.
We no longer treat the end-2026 2 GW plan as a permanent ceiling: the April investor presentation describes existing facilities as scalable to 5 GW as needed. That is not 5 GW commissioned today or a dated output forecast. The Fremont space announcement provides no independently quantified shipment uplift. Higher forecasts reflect our reassessment of the growth path and execution assumptions, not an extra order manufactured by adding overlapping Oracle figures. [V6]
Why our Base differs from brokers
Public reports of April 29 valuation examples describe Evercore using about $2.8 billion of 2028 EBITDA at 27.5× and TD Cowen using about $2.9 billion at 26×. Our Base EBITDA of $2.295 billion is lower, while our 30× multiple is higher. The earnings forecast therefore explains part of the difference; it would be incorrect to blame only a low multiple. These are historical examples reported by a secondary source, not today’s full earnings consensus or independently reproduced broker models. We do not have their complete cash, debt and share bridges. [V3] [V4]
Our 22× / 30× / 35× multiples are explicit judgments about growth, execution and cash economics, not measured peer medians. Holding the Base earnings forecast and capital rules fixed, 25×–35× produces roughly $180–250. Bull at 30×–40× produces roughly $325–430. These are sensitivities, not confidence intervals. The 24.1% divergence remains disclosed; we have not increased the target simply to fall inside the consensus band.
Cash, convertible debt and dilution
The capital snapshot begins at June 30, 2026. Cash = $2,666.859m reported unrestricted cash − $3.370m assumed repayment ($0.787m residual 2028 notes plus a $2.583m Korean joint-venture loan) + $77.683997m option exercise proceeds + cumulative H2 2026–2027 net cash generation of −$300m / +$300m / +$800m. The cash-generation cases include net investment and working-capital effects as a top-down assumption. They are retained from the previous model for comparability and have not been rebuilt from a detailed customer and capex schedule despite higher revenue forecasts; this limits confidence. [V5]
| Share component | Millions of shares | Treatment |
|---|---|---|
| June 30 common shares | 293.354001 | Starting count |
| Unsettled conversion commitments | 0.877687 | Add once |
| Outstanding RSU / PSU awards | 10.101219 | Count in full as economic claims |
| Outstanding options | 5.009767 | Add gross shares and associated exercise cash |
| 2029-note conversion shares | 1.294055 | Convert economically in all three cases; remove $26.971m principal |
| Future grant / ESPP allowance | 9.000000 | Through end-2028; prior allowance was 5.4m through end-2027 |
| Subtotal before 2030 conversion | 319.636729 | Bear denominator |
| 2030-note conversion shares | 12.822500 | Base and Bull only, paired with removal of $2,500m debt |
| Base / Bull denominator | 332.459229 | Both include 2030 conversion |
The 9.0m future allowance is H1 new RSU/PSU grants of 1.142215m plus ESPP issuance of 0.644651m, multiplied by five half-years to end-2028: 8.934330m, rounded to 9.0m. This adds 3.6m claims versus the prior allowance, with no additional assumed ESPP proceeds. Option cash is 1.672079m × $25.58 + 3.337688m × $10.46 = $77.683997m. The 2.154231m Oracle warrant shares are already inside June common shares and are not added again. The Q2 weighted-average EPS denominator is not the valuation denominator. [V5]
The 2030 notes convert economically above approximately $194.97 in this model. Base and Bull clear that level; Bear retains $2.5bn face debt. This pairs debt removal with extra shares and does not claim legal conversion has already occurred. Other claims are $206.480m financing obligations + $4.746m finance leases + $28.673m noncontrolling interests = $239.899m. We deduct the full financing-obligation book value conservatively; ordinary operating leases remain within operating expenses. Restricted cash, capped-call value and equity-method investment value receive no credit. Legal settlement choices, option time value and future financing could change the result. [V5]
Dated changes and the earlier upside discussion
| Published scenario | October 11 model | October 12 model | Change |
|---|---|---|---|
| Bear | $66 | $71 | +7.6% |
| Base | $140 | $216 | +54.3% |
| Bull | $265 | $377 | +42.3% |
The earlier archive recorded $66 / $205 / $390 with an insufficiently specified target date. The October 11 model changed these to $66 / $140 / $265 using 2027 sales multiples of 5× / 8× / 12×, mainly a valuation-policy reset. This October 12 revision retains the end-2027 target date, raises 2027 sales from $4.2 / $5.5 / $7.0bn to $4.5 / $6.5 / $8.0bn, looks forward to 2028 earnings and replaces sales multiples with explicit EBITDA margins and multiples. The older figures remain historical, not current targets.
| Controlled model bridge, US$/share | Bear | Base | Bull |
|---|---|---|---|
| Prior 2027 sales × 5 / 8 / 12, with prior capital rules | 65.50 | 140.18 | 265.47 |
| Revise 2027 revenue only | 70.25 | 165.49 | 301.96 |
| Use 2028 revenue at the old sales multiples | 82.11 | 215.29 | 420.55 |
| Use new EBITDA margins × 22 / 30 / 35 | 72.15 | 217.88 | 381.21 |
| Add 3.6m further 2028 economic claims | 71.33 | 215.52 | 377.08 |
A previously discussed $360 target could not be tied to a verified original formula and is not inserted into the official revision history as fact. It can be explained under the current Bull earnings assumptions at about 33.4× EBITDA. The archived $600 illustration used EPS $10 × P/E 60 and was an extreme-upside case. With current Bull cash and share assumptions held fixed, $600 would require about 56.3× Bull EBITDA, or about $18.08bn revenue at a 31% margin and 35× multiple. This reverse calculation is not a forecast; investment needs, cash and dilution would need rebuilding at that scale.
DCF as a supporting check
We use forward operating results and explicit market multiples for the headline targets. The separate DCF check produces end-2027 values of about $14.75 / $51.34 / $104.24 at a 13% assumed WACC. It is not a price ceiling and does not validate the higher market-multiple targets. We do not average the methods. Their gap highlights the importance of persistent growth, the price assigned to that growth, reinvestment and stock compensation.
That DCF starts with the same 2028 sales and EBITDA scenarios, projects cash flows through 2035, fades growth to 3%, assumes 25% tax, 2%-of-sales depreciation, and initial capex of 5% / 6% / 7% of sales. It charges post-2027 new compensation at 4% of revenue, uses 10% of incremental sales for working capital and terminal returns on invested capital of 15% / 20% / 25%. Its share denominator includes awards through 2027 and does not also add the market model’s 2028 allowance. The 2030 debt stays outstanding because the DCF outputs are below its conversion threshold. These are uncertain assumptions; 13% WACC is not automatically the same as an investor’s required return.
Reference standard and future revisions
Our fundamental analyses, news updates and evergreen company research share a common structure: dated evidence, investment implications, scenario targets or a maintain/change/under-review decision, the reasoning and calculation, analyst comparison, risks and original links. Article length and order vary by topic. General educational explainers use clearly labeled examples where useful rather than inventing a target. Korean and English versions of the same analysis use the same dated inputs and values.
For this and future AHNIM investment articles, we show the target date, earnings year, primary method, assumptions, shares, original sources and an investment-risk notice. DCF is a supporting test, selected according to business stage and forecast visibility. We compare published targets with a dated analyst mean, median and range where available. Difference = (AHNIM target / analyst mean − 1) × 100; an absolute gap of at least 20% triggers explicit explanation and review. We calculate the flag before rounding the percentage, focus on Base and label Bear/Bull comparisons separately. Unavailable coverage is disclosed; it is not treated as agreement. This check occurs when preparing or updating an article, not as continuous monitoring.
Our Base currently triggers that review. Differences in earnings forecasts, multiples, dilution and target horizons are disclosed above. Future updates will show previous and new values, the changed input and evidence, or explicitly maintain the target after review. Firm non-overlapping orders, customer readiness, commissioned output, margins, service costs, working capital, capex, permits and financing can alter the model. Missing material evidence results in an “under review” label, not an unsupported new target.
Investment information: AHNIM STUDIO is a personal research team. This article provides investment analysis for information and education, not personalized investment advice or an offer to buy or sell securities. Estimates can be wrong, returns are not guaranteed and loss of capital is possible. Readers must independently verify original filings, current facts and risks before investing and make their own decisions. We do not guarantee accuracy, completeness or investment outcomes. To the extent permitted by applicable law, we accept no liability for losses from reliance on this material; this notice does not exclude obligations that cannot lawfully be excluded.
Sources & original links
Sources 1–11 and 13–14 link to primary documents, company statements or the SEC filing inventory. Source 7 is retained as the original statement link but was not directly retrievable. Sources 12, 15 and 16 are secondary reporting; source 15 was reviewed only through its headline and accessible excerpt. Valuation assumptions and calculations are AHNIM STUDIO’s own. Accessed October 11, 2026.
[1] Bloom Energy — Second-quarter 2026 results and full-year guidance
2026-07-28
https://investor.bloomenergy.com/press-releases/press-release-details/2026/Bloom-Energy-Reports-Record-Second-Quarter-2026-Financial-Results-and-Raises-Full-Year-2026-Guidance/default.aspx
[2] Bloom Energy — Fourth-quarter and full-year 2025 results
2026-02-05
https://investor.bloomenergy.com/press-releases/press-release-details/2026/Bloom-Energy-Reports-Fourth-Quarter-and-Full-Year-2025-Financial-Results-with-Record-Full-Year-Revenues/
[3] Bloom Energy — Form 10-Q, quarter ended June 30, 2026 (financial statements)
2026-07-28
https://www.sec.gov/Archives/edgar/data/1664703/000162828026050247/be-20260630.htm
[4] Bloom Energy — Form 10-Q/A (corrected customer-concentration disclosure)
2026-07-29
https://www.sec.gov/Archives/edgar/data/1664703/000162828026050325/be-20260630.htm
[5] Bloom Energy and Oracle — Expanded partnership, up to 2.8 GW
2026-04-13
https://investor.bloomenergy.com/press-releases/press-release-details/2026/Bloom-Energy-and-Oracle-Expand-Strategic-Partnership-to-Deploy-up-to-2-8-GW-to-Accelerate-AI-Infrastructure-Build-Out/default.aspx
[6] Oracle — Project Jupiter fuel-cell power announcement
2026-04-27
https://www.oracle.com/news/announcement/oracle-borderplex-and-bloom-energy-to-power-project-jupiter-with-fuel-cell-technology-2026-04-27/
[7] Bloom Energy — Original statement link on Oracle commitment (X)
2026-09-24; direct retrieval unavailable during this review. The statement is attributed through the accessible secondary report in source 16.
https://x.com/Bloom_Energy/status/2103192271202213975
[8] Brookfield and Bloom Energy — Expanded AI infrastructure financing partnership
2026-06-30
https://www.bloomenergy.com/news/brookfield-and-bloom-energy-expand-ai-infrastructure-partnership/
[9] City of Fremont — September 2026 economic-development newsletter
2026-09
https://content.govdelivery.com/accounts/CAFREMONT/bulletins/4293ade
[10] Bloom Energy — Form 10-K for the year ended December 31, 2025
2026-02-09
https://www.sec.gov/Archives/edgar/data/1664703/000162828026006516/be-20251231.htm
[11] Oracle — Carbon-capture research announcement
2026-09-08
https://www.oracle.com/news/announcement/oracle-announces-research-into-carbon-capture-2026-09-08/
[12] Reuters (secondary source) — Oracle force-majeure notice and project response
2026-09-24
https://www.reuters.com/business/oracle-cites-force-majeure-shield-itself-controversial-data-center-bloomberg-2026-09-24/
[13] SEC — Bloom Energy filing inventory
Checked 2026-10-11
https://data.sec.gov/submissions/CIK0001664703.json
[14] Bloom Energy — April 2026 investor presentation
2026-04; manufacturing capacity and scalability context
https://s29.q4cdn.com/452919417/files/doc_presentations/2026/04/Investor-Presentation-PDF-April-2026.pdf
[15] Silicon Valley Business Journal — Fremont manufacturing expansion
2026-09-30; headline and accessible excerpt only. Full article restricted and not reviewed.
https://www.bizjournals.com/sanjose/news/2026/09/30/bloom-energy-fremont-manufacturing-expansion.html
[16] Stocktwits — Report quoting Bloom’s Oracle commitment statement
2026-09-25; secondary reporting used because the original X post could not be directly retrieved. A reported company statement is not an independently reviewed contract.
https://stocktwits.com/news-articles/markets/equity/be-stock-oracle-reaffirms-fuel-cell-deal-amid-project-jupiter-jitters/cZMOGvmRBa0
Valuation sources added October 12, 2026
[V1] Analyst targets and consensus forecasts; S&P Global data via Stock Analysis, page update October 7; checked October 12, 2026
https://stockanalysis.com/stocks/be/forecast/
[V2] Second public check of the analyst mean; analyst-count labels differ on this page, so the count used here is from V1
https://www.investing.com/equities/bloom-energy-corp-consensus-estimates
[V3] Evercore historical valuation formula, April 29, 2026; secondary reporting, not the full broker report or its latest target
https://www.investing.com/news/analyst-ratings/evercore-isi-raises-bloom-energy-stock-price-target-on-record-results-93CH-4644756
[V4] TD Cowen historical valuation formula, April 29, 2026; secondary reporting, not the full broker report or its latest target
https://www.investing.com/news/analyst-ratings/td-cowen-raises-bloom-energy-stock-price-target-on-ai-demand-93CH-4645437
[V5] Corrected Q2 2026 Form 10-Q/A: shares, cash, convertibles and other claims
https://www.sec.gov/Archives/edgar/data/1664703/000162828026050325/be-20260630.htm
[V6] April 2026 investor presentation: facilities scalable to 5 GW as needed
https://s29.q4cdn.com/452919417/files/doc_presentations/2026/04/Investor-Presentation-PDF-April-2026.pdf
[V7] Q2 2026 results and full-year company guidance
https://investor.bloomenergy.com/press-releases/press-release-details/2026/Bloom-Energy-Reports-Record-Second-Quarter-2026-Financial-Results-and-Raises-Full-Year-2026-Guidance/default.aspx
Leave a Reply