Bloom Is Growing Faster Than It Can Bill

One-page summary

1. A larger business, with more cash tied up in it

Bloom’s operating results were strong. Revenue of $1,235.8 million grew 138.1% year on year and product revenue grew 179.4%. Gross margin reached 33.8% reported and 34.5% adjusted, against 29.2% and 30.4% a year earlier, and operating margin went from 1.5% to 18.2%.^[1] Net income was $249.0 million against a $23.0 million loss a year earlier. Management raised full-year revenue guidance to $4.4–4.6 billion and non-GAAP earnings to $3.10–3.40 a share, the third increase this year.

The income statement shows how much more business Bloom is doing. The cash flow statement shows what financing that growth required.

Contract assets are the line that matters. The balance went from $178.9 million at the start of the year to $792.8 million, and $430.3 million of that increase came in this quarter alone. A contract asset is revenue the company has recognised but cannot yet invoice, because something in the contract — a milestone, an acceptance, a commissioning step — has not been satisfied. It is a measure of how far recognition has run ahead of billing.

Two things keep it in proportion. A rising contract-asset balance establishes a gap between recognition and billing; it does not establish that the billing or the collection will be difficult, and this quarter says nothing either way about that. Year-to-date free cash flow also remained positive: Edwards put it at “a use of $181 million in the quarter and positive $42 million year to date.”^[9] In the second quarter Bloom generated $226.4 million of operating cash on $198.9 million of net income, the reverse of this one.

Management describes the shortfall as timing, and described it the same way on the call. Chief financial officer Simon Edwards said net income was “more than offset by an increase in working capital,” that “billing follows contract milestones that can occur in a different quarter,” and that the difference sits in “contract assets and receivables until billed and collected.” He added that the quarter was “not the beginning of a trend,” with “turning both our receivables and contract assets” now the primary focus.^[9] The next report gives a better test of that explanation: whether billing begins to catch up with recognised revenue, and whether operating cash flow improves alongside earnings. For now the contract-asset balance is something to watch rather than a sign that customers will struggle to pay.

2. What the backlog numbers actually tell us

Our September concern was the distance between management's roughly $20 billion commercial-backlog figure and the much smaller balance of remaining performance obligations disclosed in the filings — $494.1 million at the end of June.^[3] The two are defined differently and measure different things. Remaining performance obligations are contracted revenue that has not yet been recognised, which the company expects to recognise over the next one to two years. Commercial backlog is a wider measure the company defines itself and is not a disclosure under the revenue standard. We have not reconciled the gap between them against a company explanation and will not assert what causes it. For this update the question is narrower: how fast is the disclosed balance growing?

Three further measures sit between that backlog and cash, and it is worth being precise about what they are not. They do not form a funnel, their balances are not meant to reconcile, and a dollar does not pass tidily from one to the next — each is defined differently and recognised on its own timing. They are related indicators of the same build-out, read together:

Before the report we identified $750 million of remaining performance obligations as the level that would challenge our view of backlog conversion, and registered a call of below at a 70% probability.^[8] Bloom disclosed $717.4 million.

Bloom gave no fiscal 2027 outlook, which leaves the growth question open; an analyst on the call prefaced a question with “I obviously know you're not guiding to 2027 yet,” and Edwards answered in general terms about “the long game.”^[9] Chief executive KR Sridhar said “our backlog is continuing to expand faster than our revenue” but declined to quantify it.^[9] Remaining performance obligations are therefore the one disclosed measure of future contracted revenue we can work from.

The number came in below the threshold. It also rose 45.2% in three months and 71.1% since January, and the second fact matters more to the investment case than the first. We had questioned how quickly the commercial backlog would convert into disclosed obligations. This quarter is evidence that it is converting, and a narrow miss against a line we chose ourselves does not offset that.

3. A lower price does not settle the valuation

Bloom is cheaper than it was in September for two reasons: the price has fallen, and expected earnings have risen. Separating those matters more than the headline de-rating.

One convention needs stating, because getting it wrong counts the same claim twice. Essentially all of Bloom’s recourse debt is convertible — $2,450.9 million of zero-coupon notes due 2030 and $22.6 million of 3% Green Notes, against $3.0 million of non-recourse term debt — and the diluted share count already assumes it converts, under the if-converted method with a $3.5 million interest add-back to the numerator.^[5] Adding that debt to enterprise value while also using diluted shares counts it on both sides.

Our primary basis is therefore shares outstanding with all debt included: 295.5 million shares at $217.77 gives a $64.4 billion equity value, and after $2.5 billion of debt and $2.5 billion of cash an enterprise value of about $64.3 billion, or 14.3× the guided revenue midpoint. On the if-converted alternative — 324.3 million shares with the convertible debt removed, because the shares stand in its place — enterprise value is $68.1 billion and 15.1×. Both are here because the conclusion holds either way. Sixty-seven times a guided earnings number remains a demanding valuation for a business that consumed cash this quarter. But a demanding valuation is not by itself a reason to keep an old price target.

Why the target is under review

Our $235 figure was a probability-weighted output, not a single-point valuation. Our base case used 50–55 times a fiscal 2027 consensus of about $4.93, which produced a $245–270 range; the bear leg was anchored on a discounted cash flow that returns roughly $64 a share on the assumptions we set in September; the weights were 25% bull, 45% base, 30% bear.^[4] Those two anchors sit far apart on purpose — the DCF prices the business on its own cash generation, the base leg prices it on what the market will pay for near-term earnings, and the weighting between them carries the judgement.

Both anchors need updating. The fiscal 2027 consensus predates these results and will be revised within days. The discounted cash flow predates two of this year's three guidance increases.

The work is to decide how much of this quarter's revenue and margin improvement carries forward, and how much investment the next stage of growth requires. The cash flow statement makes that second assumption the important one: higher reported earnings do not convert dollar for dollar into value if delivering them ties up materially more working capital. We will publish the revised figure within two weeks with the scenario assumptions and weights shown. Until then $235 stands as published; we have not re-earned it.

4. What the share-price decline does and does not tell us

Bloom closed at $295.78 on October 6 and at $217.77 on October 22, down 26.4%, with the S&P 500 down 1.9% over the same window. Thursday alone took 8.7% off the shares before the results.

We reviewed Bloom's SEC filings from September 20 through the report and found nothing that explains the decline: three Form 4s and a Form 144, the most recent on October 2, and no 8-K, offering, guidance change or customer announcement.^[2] The one open-market sale — 8,437 shares by the chief commercial officer at an average of $277.17 — was executed under a Rule 10b5-1 plan adopted in advance, and the filing says so. That is as far as the filings go; news outside them could still explain it.

In September we attributed the fall to the discount rate, having described Bloom as behaving “less like a mature industrial company and more like a long-duration asset.”^[4] Higher yields were consistent with the initial decline: the ten-year reached a 5.31% cycle high on October 5, the day before Bloom's peak.

Yields have since returned to roughly their starting point and Bloom has not. Rates may explain part of the decline but not the whole move. Given how broadly the group fell, a shared driver looks likely without being proven. For this update the more useful question is what the lower price buys: a stronger earnings base than we expected, at a still-demanding valuation, with cash conversion yet to be shown.

5. What would change the rating

Our hold band is a twelve-to-eighteen-month probability-weighted target within roughly 10% of the market price either way. At Thursday's close of $217.77 the $235 target sits 7.9% above, inside the band; at the after-hours level of $204.50 it sits 14.9% above, outside it. We are not changing a rating on an after-hours price, and the more useful statement is what evidence we need rather than where the number currently falls.

The next operating test is the fourth-quarter report, and our revised target inside two weeks. Management did not give a date on the call beyond noting that the fourth quarter ends with the calendar year; Bloom has reported the December quarter in February in each of the last two years, and we will take the date from the company rather than estimate it, having been wrong about a reporting date on this name once already.

6. An accountability note

The September initiation published two trim levels, $248–254 and $277–280. Between September 8 and October 22 the shares closed above $277 on ten sessions and above $248 on twenty-six; a resting sell order at either level would have been marketable on those days. None was ever placed.

The failure is earlier than it first looks. We drafted both orders on October 20, and at that point our policy permitted the trade, because no publication was yet scheduled. Deciding to publish this note is what pulled October 20 inside the restricted window, and entering an order days before a note goes out is the thing the rule exists to prevent. The orders drafted on the 20th were therefore never the fix. The levels should have been resting from the day we published them in September, when twenty-six qualifying sessions and nothing in the policy stood in the way.

Twenty-five shares worth $7,394.50 at the October 6 peak closed Thursday at $5,444.25, $72.50 above a $214.87 cost. That is an execution failure, and it is a failure of sequencing rather than of judgement: we identified the levels and left the orders unplaced until our own rules closed around them. Readers should judge the published framework and the account’s realised results separately. We will address the process changes elsewhere; the holding and order status at publication appear in the disclosure below.

Housekeeping

Scope. This note reflects the results release and the 10-Q, both filed October 22, and the conference call held that afternoon. Company financials are taken from the filings; call quotations are from a published transcript rather than our own recording, and are labelled as such.^[9]

Two corrections to the coverage. Early reports described management as pointing to capacity able to support “four times 2025 revenue.” The transcript shows Edwards saying capital spending was “four times what we spent in all of 2025” — investment, not revenue — alongside year-to-date revenue of $3.05 billion, “already 50% above what we generated in all of 2025.”^[9] Separately, one aggregator reported the shares up as much as 20% after hours; the tape shows them in a $201.50–$213.40 range, below the close throughout.^[6]

Trading window. Our policy restricts trading in a covered name from ten days before a publication to five days after it. Because the restriction runs backwards from the publication date as well as forwards, a note published October 23 carries a window of October 13 through October 28 — it covers the fortnight already elapsed, including the report itself. No order was resting when the window opened, so no resting-order exception applies, and the orders described in section 6 cannot be entered until October 29. Protective stops and kills at previously approved levels remain available at any time and are logged when used; none was triggered.

Sources

1. Bloom Energy Form 8-K of October 22, 2026, and Exhibit 99.1 — third-quarter results, revised full-year guidance, condensed balance sheets and statements of cash flows. SEC EDGAR
2. SEC EDGAR submissions index, CIK 1664703, reviewed October 22. Form 4 of October 2 carrying the Rule 10b5-1 affirmation; Form 144 of October 1. Filing index
3. Bloom Energy Form 10-Q for the quarter ended June 30, 2026 — remaining performance obligations of $442.4M product and $51.7M service. SEC EDGAR
4. Our initiation: “The Index Doesn’t Read the 10-Q”, September 9, 2026 — sensitivity table, scenario weights and trim levels.
5. Bloom Energy Form 10-Q for the quarter ended September 30, 2026, filed October 22 — unsatisfied performance obligations of $661.2M and $56.2M. SEC EDGAR
6. Yahoo Finance daily closes for BE, the cohort and the ten-year yield; after-hours prints from the same source, labelled as such. Price history
7. Consensus estimates from StockAnalysis, a secondary source dated October 16 and pre-print. Forecast page
8. Our pre-registration for this report, frozen October 20 at 15:31 Eastern, before the results. This is an internal record and has no public URL; the threshold, the direction and the 70% probability are quoted in section 2 exactly as they were written.
9. Bloom Energy third-quarter 2026 earnings call, October 22, 2026, 2:00pm Pacific. Quotations are taken from the published transcript at Investing.com, a secondary source, and are attributed to the named speaker. Transcript