BE — Enhanced Equity Research

Executive Summary

The operating case is now proven. The price still assumes the next four years.

Bloom Energy (BE) has crossed from story to substance. Q2 FY26 revenue reached $1,065.4M, up 165.5% year over year and the first billion-dollar quarter, with 33.4% GAAP gross margin, $182.2M of operating income and $0.62 GAAP diluted EPS from a company that lost $0.37 per share across all of FY2025. Management raised FY2026 guidance to $3.9B–$4.2B with non-GAAP EPS of $2.55–$2.85, and the cash statement corroborates it: 1H26 operating cash flow of +$300.0M against $77.8M of capex, versus a $323.8M operating burn a year earlier. Yet the stock sits 41.1% below its June high, and the reasons are in the filings rather than the narrative. One non-related-party customer was approximately 73% of Q2 revenue — a figure the company had to file a 10-Q/A to state correctly — while audited remaining performance obligations are $494.1M against the ~$20B backlog management promotes. The gap is defensible under ASC 606, which excludes framework agreements without firm orders, but very little of the story is contractually enforceable today. Add an unresolved securities class action over Chinese scandium sourcing, with a Sep 28 lead-plaintiff deadline, and 76.6× guided FY26 earnings and 19.5× EV/TTM revenue leaves no room for error. The debate is no longer whether the demand is real — it is whether it diversifies before the turbine shortage that created it closes after 2028.

Q2 FY26 SCORECARD

Hypergrowth, newly profitable — at a hypergrowth price — Quarterly figures taken directly from the as-filed Q2 FY2026 10-Q and 10-Q/A (filed Jul 28–29, 2026) and the Jul 28 earnings release. Price, technicals and market cap per Yahoo Finance, Sep 1, 2026.

Q2 FY26 Revenue

$1.065B

+165.5% y/y vs $401.2M — first billion-dollar quarter

FY2026 Guidance

$3.9B–$4.2B

Raised Jul 28 · ~100% y/y growth at the midpoint

Q2 GAAP Diluted EPS

$0.62

vs −$0.18 a year ago · non-GAAP $0.78 vs $0.39 est

Q2 Operating Margin

17.1%

$182.2M operating income vs a $3.5M loss a year ago

Q2 Customer Concentration

73%

One non-related-party customer — restated via 10-Q/A

Remaining Perf. Obligations

$494.1M

$442.4M product + $51.7M service — vs the ~$20B promoted backlog

EV / TTM Revenue

19.5×

$60.7B EV on $3.11B TTM revenue · 76.6× guided FY26 EPS

Net Cash Position

+$189M

$2.67B cash less $2.48B debt · $222M of 1H26 free cash flow

Quarterly Revenue — Four Straight Records (SEC EDGAR, as filed)

$519M

Q3'25

$778M

Q4'25

$751M

Q1'26

$1.07B

Q2'26

GROWTH ENGINE

Backlog is outrunning revenue — but only $494.1M of it is contractually binding

$20B

Promoted total backlog

~$6B product, up 140% y/y · ~$14B annuity-like service

2.8 GW

Oracle master services agreement

Signed Apr 13, 2026 · initial 1.2 GW contracted and deploying

$25B

Brookfield financing framework

Expanded 5× from $5B on Jul 9, 2026 via its AI Infrastructure Fund

SEP 28, 2026

Securities class action lead-plaintiff deadline

Class period Feb 27, 2025 – Jul 8, 2026, over allegations that Bloom understated reliance on Chinese scandium intermediaries. The nearest dated risk event and the one sitting inside the current quarter.

~OCT 27, 2026

Q3 FY26 results

First read on whether $1B quarters are the new base. The $3.9B–$4.2B guide implies roughly $2.1B–$2.4B across H2. The customer-concentration line in the filing matters more than the revenue number.

Q4 2026

Fremont doubles from 1 GW to 2 GW

Roughly $100M of expansion targeted for completion by year end. This is the single physical gate on how fast product backlog converts into revenue.

2027

Oracle 1.2 GW tranche deploys; Brookfield and AEP draw down

Conversion of the remaining ~1.6 GW ceiling into firm orders is the key bookings catalyst, alongside AEP's up-to-1 GW agreement. Watch whether RPO closes any of the gap to the promoted backlog.

2028–2029

The scarcity window starts to close

GE Vernova and Caterpillar turbines are sold out through 2028 with under 10% of 2029 uncommitted. As that queue clears and GEV's own fuel cell arrives, Bloom's speed-to-power premium compresses.

Earnings & Signals

Forward Estimates, Surprises & Insider Activity

Forward Earnings Estimates

FY+1 EPS Consensus$2.71

FY+2 EPS Consensus$4.92

PEG Ratio~1.3 (on ~40% FY27 growth)

Forward P/E76.6× FY26 · 42.1× FY27

EPS Revisions (90d)↑n/a — per-analyst counts not disclosed by sources checked ↓n/a (Sharply upward — Zacks consensus for FY2026 moved +25% and FY2027 +12.5% over the trailing 30 days, with no material downward revisions found)

Guidance AccuracyStrong — guidance raised at both Q1 and Q2 FY26; FY26 now $3.9B–$4.2B revenue and $2.55–$2.85 non-GAAP EPS, up from $3.4B–$3.8B and $1.85–$2.25

Earnings Surprise Track Record

Q3 FY25 Est: $0.08 Act: $0.15 +87.5%

Q4 FY25 Est: $0.25 Act: $0.45 +80.0%

Q1 FY26 Est: $0.12 Act: $0.44 +255.1%

Q2 FY26 Est: $0.39 Act: $0.78 +100.0%

Beat Rate4/4 — note these are non-GAAP figures against non-GAAP consensus; Q2 GAAP diluted EPS was $0.62

Insider Activity (90 Days)

Net Buying/Selling−$22.1M · roughly 89,464 shares sold, zero shares bought on the open market

Sell/Buy RatioNo open-market purchases in the trailing 90 days

Uniformly negative. Every Form 4 disposition between June and September 2026 was a sale or a routine director option award, with not one open-market purchase. Some is mechanical — 10b5-1 plans and RSU tax withholding — but the complete absence of buying at a 41% drawdown from the high is the signal worth weighing, particularly against insider ownership of only 1.80%.

August was dense: John T. Chambers (Director) sold 15,000 shares on Aug 3 at $205.58 and another 15,000 on Aug 13 at $250.00; Jeffrey R. Immelt (Director) sold 30,000 on Aug 17 at $238.91 under a 10b5-1 plan adopted May 1, 2026. Aman Joshi (CCO) sold 4,677 at $241.75, Satish Chitoori (COO) 2,053 at $241.64 and Shawn Marie Soderberg (CLO) 2,895 at $233.60, all on Aug 14. Earlier sales printed at $288–$300 in June and July.

Peer Comparison

Relative Valuation vs. Competitors

Company	P/E (fwd)	EV/Rev	Rev Growth	Gross Margin
Bloom Energy	58.8×	19.5×	+165.5%	33.4%
GE Vernova	42.6×	5.4×	+21.9%	20.6%
Vertiv Holdings	31.0×	8.1×	+28.9%	38.0%
Cummins	17.2×	2.4×	+9.4%	25.8%
Plug Power	n/a	5.2×	+10.6%	−24.8%

Comps chosen for genuine overlap rather than label. GEV sells the gas turbines Bloom displaces and is sold out through 2028, which is precisely why Bloom wins on speed-to-power today and why the moat is dated. VRT sells the rest of the data-center electrical stack and is the cleanest read on what the market pays for AI-infrastructure hardware. CMI is the reciprocating-engine and onsite-generation alternative. PLUG shows what the fuel-cell category looks like without the data-center pivot. Bloom trades at roughly 3.6× GEV's EV/revenue and 8× Cummins', and the premium is entirely a function of the growth line. Forward P/E is next-twelve-months; against Bloom's own FY26 guidance midpoint of $2.70 the current-year multiple is 76.6×. Peer multiples per stockanalysis.com, Sep 1, 2026.

Valuation

Price Targets & Scenarios

Scenario	Price Target	Assumptions	Probability
Concentration breaks open	$330	The Q3 and Q4 prints show the top customer falling below 50% of revenue as MiTAC, neocloud and additional hyperscaler orders convert — management already claims every major US hyperscaler has validated the platform. Fremont reaches 2 GW on schedule and framework gigawatts convert into disclosed RPO, closing the credibility gap the short report exploited. The scandium allegations resolve without a restatement. FY27 revenue clears $6.3B with gross margin holding near 34%, non-GAAP EPS reaches roughly $5.50, and the multiple holds near 60× on demonstrated durability.	28%
Guidance met, concentration persists	$230	FY26 lands inside the $3.9B–$4.2B guide and FY27 grows about 40% to roughly $5.7B, but the anchor customer stays above half of revenue and RPO remains far below the promoted backlog. We model FY27 non-GAAP EPS near $4.50 against $4.92 consensus, and apply a 52× multiple — a premium for 40% growth and expanding operating leverage, discounted for concentration and the legal overhang. Litigation settles or fades without a restatement; the Fremont ramp lands roughly on time.	42%
The one customer slips	$120	Anchor-customer shipments push right or terms are renegotiated, and with 73% of a quarter riding on one account the revenue line breaks visibly. Alternatively the scandium allegations force a disclosure restatement, or scandium supply itself caps the ramp — independent work argues 2.3–2.5 GW of annual throughput needs 35–100 metric tons of scandium oxide that may not exist before 2029. Growth decelerates below 30%, FY27 non-GAAP EPS resets toward $4.00 and the multiple collapses to 25–30× as the story loses its scarcity premium.	30%

Probability-Weighted Target: $225 (+8.7% vs. $206.92)

$225

Weighted

Bull $33028%

Base $23042%

Bear $12030%

Analyst Consensus

Evercore ISI

$350

Reiterated Outperform · Sep 1, 2026

J.P. Morgan

$314

Target adjusted to $314 · Aug 3, 2026

Mizuho Securities

$242

Upgraded to Outperform post-Q2 FY26

Jefferies

$229

Raised from $188, maintained Hold · Aug 13, 2026

15 Buy · 12 Hold · 2 Sell — 29 analysts carry a consensus Buy with an average target of $275.08 (stockanalysis.com, Sep 1, 2026), implying roughly 33% upside from $206.92. The dispersion is the real signal: a $97 low against a $390 high is a four-fold spread, meaning the Street agrees on direction and not at all on magnitude.

Technical

Key Levels & Options Intelligence

S/RSupport & Resistance

52-Week High

$351.28

Upper Bollinger — August rejection zone

$248.69

50-DMA — first resistance to reclaim

$231.38

Current Price

$206.92

Round-number support

$200.00

200-DMA — the bull/bear line

$189.30

Post-Q2 washout low — tested and held Jul 29

$163.75

52-Week Low

$48.87

OptOptions & Sentiment

30-Day Implied Volatility: 77.9% mean — near-dated Sep 4 strikes quoted 91.5%–103.9%
30-Day Realized Volatility: 72.6% close-to-close · 79.1% Parkinson (as of Aug 31, 2026)
Put/Call Ratio — Open Interest: 1.31 — persistently put-skewed
Put/Call Ratio — Volume: 1.06
IV Skew (put minus call): −0.036 — puts 78.7% vs calls 77.2%, close to flat
Short Interest: 18.26M shares — 6.32% of float (Aug 14 settlement)
Days to Cover: 1.09 days — very little squeeze fuel remains
Beta (5-year): 3.83 — downside is long liquidation, not a short squeeze

Conviction Analysis

Systematic Conviction Score: 53/100 (Medium)

Analyst Alignment

30%

FCF Visibility

25%

Catalyst Clarity

20%

Valuation Safety

15%

Mgmt Quality

10%

Free cash flow visibility earns a genuine mark at 55: 1H26 produced roughly $222M of free cash flow against a $323.8M operating burn a year earlier, which is a real inflection rather than an adjusted-metric artifact. But $301.2M of that operating cash came from deferred revenue and customer deposits, and contracted RPO is only $494.1M against a ~$20B promoted backlog, so the cash is real while the visibility is mostly non-binding. Catalyst clarity at 66 is two-sided: the Q3 print and the Fremont 2 GW completion cut both ways, and the Sep 28 lead-plaintiff deadline sits directly inside the window. What caps the score is valuation safety at 18 — at 76.6× guided FY26 earnings and 19.5× EV/TTM revenue there is no floor beneath this price other than the 200-DMA at $189.30 — and management quality at 55, where excellent operational delivery and two guidance raises are offset by zero insider buying against roughly 89,464 shares sold in 90 days and the promotion of a $20B backlog against $494.1M of audited obligations. Analyst alignment at 60 is deliberately below the Street: consensus is $275.08 and we are at $225.

Risk & Ownership

Risk Assessment & Insider Signals

OOwnership & Insider Signals

Institutions own 84.07% of a 289.03M-share float: 1,007 institutional holders against 294.53M shares outstanding. Ownership that concentrated in professional hands cuts both ways: it validated the re-rating on the way up, but positioning unwinds are fast and correlated when the AI-power trade rotates.
Top holders — BlackRock, Ameriprise, Vanguard: At Mar 31, 2026 BlackRock held 25.06M shares (8.81%), Ameriprise Financial 22.89M (8.05%) and Vanguard entities 19.53M combined (8.72%). BlackRock's later Jun 30, 2026 13G/A shows 17,011,609 shares (6.0% of class) — a reduction of roughly 8M shares across the quarter that ran into the short-report drawdown.
SK ecoplant is the notable strategic holder: SK ecoplant Co., Ltd. discloses 10,883,701 shares, 4.7% of the class, on Schedule 13D/A. A commercial partner rather than a financial holder, which makes the stake stickier than index money but ties a block of the register to the Korean deployment relationship — and to the related-party revenue line.
Russell 1000 Growth inclusion, June 2026: BE graduated from the Russell 2000 into the Russell 1000 and Russell 1000 Growth benchmarks at the June 2026 reconstitution. That converted a chunk of the float into passive growth-index money, which buys mechanically on inclusion and does not defend the stock on the way down.
A crowded long, not a crowded short: Short interest is 18.26M shares, 6.32% of float, with only 1.09 days to cover, after collapsing roughly 35% between mid and late June 2026. Against a 3.83 beta that argues downside comes from long liquidation rather than a squeeze — there is very little short base left to fuel one.
Insiders own 1.80%, and the register is momentum-driven: Insider ownership is small relative to a $60.9B market cap, so management's economic alignment is limited. Separately, an Aug 24 congressional disclosure of a Pelosi-household purchase moved the stock 5–7% on sentiment alone — a sign of how narrative-sensitive the current holder base is.

Risk Heatmap

Quantified Risk Assessment

Severity Risk Factor Prob. PT Impact

High Valuation leaves no room for error

At $206.92 the stock trades at 76.6× the midpoint of its own guided FY26 non-GAAP EPS, 19.5× EV/TTM revenue and roughly 15× guided FY26 revenue, against GE Vernova at 5.4× and Cummins at 2.4×. The entire market value assumes deployments scale from roughly 1 GW in 2026 toward 5 GW by 2030, so any slip in the ramp compresses the multiple long before it touches earnings.

50% -35%

High Customer concentration is extreme and getting worse

One non-related-party customer was approximately 73% of Q2 FY26 revenue — a disclosure the company had to amend its 10-Q to state correctly. Across 1H26 two customers were 44% and 21%, the second a related party, against four customers at 30/18/15/11% a year earlier. US revenue also jumped to 90% of the total from 59%. A pause by the anchor account removes most of the growth at once.

35% -40%

High Scandium sourcing allegations and the securities class action

A July 8, 2026 short report alleged Bloom sources scandium from China through intermediaries in Thailand, Japan and South Korea, triggering a class action covering Feb 27, 2025 to Jul 8, 2026 with a Sep 28, 2026 lead-plaintiff deadline. Independent work argues 2.3–2.5 GW of annual throughput requires 35–100 metric tons of scandium oxide, and that supply for even half of the stated ambition is a 2029–2030 event — a physical ceiling on the ramp the backlog assumes.

45% -30%

Medium The speed-to-power moat narrows after 2028

Bloom's premium rests on delivering in months while GE Vernova's turbines are sold out through 2028 with under 10% of 2029 slots uncommitted — Oracle reportedly had power 55 days from first engagement against one-to-two-year interconnection queues. As turbine, reciprocating-engine and grid queues clear, and as GEV's own fuel cell arrives in two to three years, the scarcity rent being capitalised today compresses.

40% -25%

Medium Convertible dilution — the 2030 notes are now in the money

Issued shares were 294.5M at the Q2 cover date against 323.3M weighted diluted, a roughly 10% gap. The 3.00% 2029 notes convert near $20.84 and are deeply in the money, while the 0% 2030 notes convert near $194.97 — below the current $206.92, so that tranche is dilutive too. Oracle's 3.5M-share warrant was cashless-exercised on May 1, 2026 for 1,905,433 net shares, with its fixed fair value amortised as a reduction of revenue as Energy Servers ship.

70% -10%

Medium Policy dependence and the 48E domestic-content trap

From January 2026 the Energy Servers qualify for a 30% Section 48E investment tax credit, effectively a one-third discount on customer capex, but only if prevailing-wage and domestic-content tests are met — the same tests the Chinese-scandium allegations put in question. The legacy fuel-cell ITC expired at end-2024 with safe-harboured projects running only to Dec 31, 2028, so adverse rulemaking or tariff action hits demand economics directly rather than gradually.

30% -20%

Recommendation

Summary

Rating

HOLD

Conviction

Medium

Price Target

$225

Timeframe

12 mo

Upside

+8.7%

Position Size

0%–2%

Entry Strategy

Tranche 1 — 40%

~$190

At the 200-DMA ($189.30), the last well-defined trend support and the level that now defines the bull/bear line. Do not chase at $206.92 into the $231 50-DMA overhead — the risk/reward to our $225 target is roughly 1:1 there and does not justify a starter position.

Tranche 2 — 35%

~$165

At the post-Q2 washout low of $163.75, which was tested on Jul 29 and reversed hard the next session. A return here likely means the Q3 concentration disclosure disappointed — add only if the single-customer share is improving, not if it has widened.

Tranche 3 — 25%

~$135

Approaching the bear-case zone. Reserve this for genuine capitulation where the multiple compresses on sentiment rather than a broken order book. If the anchor customer's volumes are actually cut, or the scandium allegations force a restatement, do not average down — re-underwrite the thesis from scratch.

BE

1-Year Price & Key Levels