Stem, Inc.: A Negative-Earnings Software Pivot With a Fragile Balance Sheet

The Quarter in Context

For the three months ended June 30, 2026, Stem, Inc. generated $33.7 million of revenue against a $7.7 million operating loss and a $14.4 million net loss, equating to a negative 22.8% operating margin and a negative 42.7% net margin. The headline revenue line was actually softer than the prior-year second quarter at $38.4 million, a 12.3% year-over-year decline, reflecting the deliberate exit from battery resale activity that historically anchored the top line. Six-month operating cash use was $8.0 million, and although trailing-twelve-month operating cash flow turned positive at $11.6 million, that figure is heavily distorted by a $14.3 million working-capital release in a single quarter rather than by structural earnings power. The 2025 net income of $177.5 million (six-month basis) is non-comparable, as it was driven entirely by a one-time $220.0 million gain on debt extinguishment from the June 30, 2025 senior secured note exchange, not by operations. The reporting period leaves the reader looking at a software transition that is shrinking the top line, persisting in operating losses, and consuming cash.

The balance sheet at June 30, 2026 carries $38.4 million of cash against $184.0 million of 2030 Convertible Notes, $135.2 million of 2030 Senior Secured Notes, and $37.2 million of financing obligations, producing a stockholder's deficit of $270.0 million. The company executed a 1-for-20 reverse stock split in June 2025, the senior secured notes carry an 11% cash-pay or 12% paid-in-kind toggle, and the ATM program issued 595,303 shares in the first half for $5.6 million of net proceeds. Investors should treat this quarter as a transition checkpoint for a software pivot, not as evidence of operational earnings power.

Business Model and Profit Engine

Stem sells AI-enabled software, edge hardware, professional services, and managed services to owners, operators, and developers of distributed clean-energy assets. The 10-Q breaks revenue into two streams: services and other revenue ($18.6 million in Q2 2026) and hardware revenue ($15.0 million in Q2 2026). Both lines declined versus the prior-year quarter, with hardware revenue shrinking faster, consistent with management's deliberate exit from lower-margin battery resale. Services and other revenue is the segment management is leaning into under the post-2024 software pivot, and it now represents roughly 55% of the Q2 2026 mix versus 54% in the prior-year period, a small mix shift that the filings do not yet support as a recurring-revenue breakthrough. Gross profit of $13.9 million implies a 41.3% Q2 2026 gross margin versus 33.4% in the prior-year quarter, a year-over-year improvement of roughly 790 basis points, but the lift comes from shrinking hardware rather than from services-led pricing power.

Filing facts do not provide a clean unit-economics or segment-contribution breakdown, because management reports a single operating segment. The filings do not disclose installed-base pricing, recurring-revenue renewal rates, services-margin expansion, or hardware-margin contraction at the line level, which the reader should treat as a data gap rather than as an inference. Capital intensity is comparatively low: investing activities consumed only $2.6 million in the first half of 2026, and that cash outflow went entirely to internally developed software rather than to physical capacity. Working capital is not a source of structural cash; rather, deferred revenue of $126.0 million acts as a customer-funded float, but it is gradually unwinding as the company recognizes revenue. The filings identify Customer A as 14% of Q2 2026 revenue and 17% of accounts receivable, and Customer B as 10% of accounts receivable, so customer concentration is observable and meaningful. The filings do not separately disclose pricing power, services attach rate, or hardware services cross-subsidy, so the reader cannot quantify how much of the gross-margin expansion is structural versus mix-driven. Observable conditions that would weaken the profit engine include a renewed downturn in solar-plus-storage project activity, additional customer concentration loss, and any compression in services attach rates.

Earnings Quality and Cash Conversion

Where the issuer reports cash flow and capital expenditure cumulatively, the standalone-quarter cash-flow figures are derived by subtracting the preceding cumulative filing. They are calculations from SEC facts rather than separately reported quarterly line items. Six-month net cash used in operating activities was $8.0 million, and Q2 2026 operating cash flow of $0.3 million is a standalone-quarter value calculated by subtracting the Q1 2026 operating cash use of $8.3 million from the $8.0 million six-month operating cash outflow, then backing into the quarter using filed cumulative filings, which is an author calculation rather than a directly reported quarterly line item. This $0.3 million cash result is not a durable inflection. Depreciation and amortization of $21.9 million in the first six months is by far the largest non-cash add-back, reflecting $113.0 million of intangible assets and $34.0 million of energy-storage systems on the balance sheet. Stock-based compensation of $3.5 million in the first six months, plus interest expense of $14.9 million, dominate the reported loss.

The 2025 net income figure of $177.5 million (six-month basis) and the Q2 2025 net income of $202.5 million should be flagged as non-comparable, because both were lifted by the $220.0 million gain on extinguishment of debt. Net cash used in investing activities of $2.6 million consisted entirely of internally-developed software capitalization. With operating cash flow near zero but positive only because of working-capital release, the reader cannot conclude that the underlying business is self-funding; the absence of a same-quarter capital-expenditure disclosure means free cash flow cannot be calculated, and the reader should treat it as unavailable rather than as positive.

Business Quality, Moat, and Industry Position

On the available evidence, the company is best described as an ordinary participant, not a demonstrated core supply-chain bottleneck. This classification describes strategic position only; it is not independent proof of a moat or valuation upside. Stem positions itself as a global leader in AI-enabled software for solar and storage asset optimization, citing 36 gigawatts of solar and approximately 2 gigawatt-hours of storage under management as of the latest 10-K. The product suite, PowerTrack, is offered across solar monitoring, energy management, supervisory control, and asset optimization. The 10-K narrative frames scale and operating data as the moat, on the theory that more operating data drives better software. The filings do not provide win-rate data, customer-renewal evidence, switching-cost quantification, or third-party benchmarks that would let a reader verify any of those claims. The most recent quarter's services-revenue trajectory is down year-over-year, which is not consistent with a strengthening moat. Without published switching-cost evidence, multi-year contract renewal disclosures, or competitor benchmarking, moat strength should be treated as a hypothesis, not as a fact.

The company reports a single operating segment and does not separately break out services-attached-to-hardware economics, so the filings do not let the reader quantify cross-subsidy between business lines. Industry context is supportive in the long run as the energy transition expands, but the 10-K itself explicitly warns that the One Big Beautiful Bill Act and macroeconomic pressures may affect customers' project pipelines, and the reader should treat any tailwind narrative as management expectation rather than as filing fact.

Management and Capital Allocation

Management's capital-allocation history is the most observable piece of evidence. The June 2025 exchange converted roughly $350 million of legacy convertibles into $155.4 million of 2030 Senior Secured Notes plus warrants, booked a $220.0 million accounting gain, but loaded the balance sheet with notes that pay 11% cash or 12% paid-in-kind. That transaction materially reduced future dilution risk in the short run, but it did not eliminate the underlying cash-interest burden. The ATM program, opened in March 2026, has issued 595,303 shares for $5.6 million of net proceeds in the first half, equating to a roughly $9.46 average price, which the reader should note without using it as a valuation anchor.

In April 2025, the company announced an approximately 27% workforce reduction and recorded $6.1 million of restructuring costs in fiscal 2025. The Q2 2026 statement of operations does not report restructuring costs as a separate line, but the year-over-year reduction in R&D from $10.0 million to $6.5 million and in general and administrative expense from $8.9 million to $8.2 million suggests the cost base has been reduced. Capital expenditure of $2.6 million in the first half went to internally-developed software. There were no share repurchases and no dividends in the available filings. Stock-based compensation of $1.5 million in Q2 2026 and $3.5 million for the half is a continuing dilution mechanism that the reader should track.

Valuation and What the Market Already Expects

The dated market capitalization of $37.5 million at a $3.90 share price as of September 27, 2026 sits atop a $270.0 million stockholders' deficit and roughly $319.2 million of senior secured plus convertible notes plus financing obligations, so the dated enterprise value implied is on the order of $318 million when cash is netted. The reader can discuss this qualitative implication, but a precise enterprise value depends on same-date debt and cash figures that the available SEC data does not fully align with the market-snapshot date. The Q2 2026 diluted loss per share of $1.58 cannot be annualized into a trailing P/E without a true trailing-twelve-month denominator, so any P/E ratio is unavailable rather than negative.

The qualitative implication of the dated market price is that the equity is being valued as an option on the software pivot, not as a claim on current cash flow. With trailing-twelve-month operating cash flow distorted by a single-quarter working-capital release and with six-month operating cash still negative, the market is implicitly requiring a normalization to recurring-revenue-driven positive free cash flow, a margin trajectory that lifts operating cash conversion above the current 41% gross-margin baseline, and a debt-paydown path that does not rely on further equity issuance. None of those requirements is currently evidenced in the filings.

Ten-Year Cash Flow to Fair Market Value

Direct compounding from a negative free cash flow base is invalid, because the available SEC data does not support a positive run-rate free cash flow from which to discount. Operating cash flow for the first half of 2026 was negative $8.0 million, capital expenditure was $2.6 million, and the calculated package reports free cash flow as unavailable because of the absence of a same-unit same-period capital-expenditure disclosure. Any year-five or year-ten normalization must be labeled as an author assumption. As a bear-case assumption, if revenue compounds at a 0% growth path off a $148 million trailing-twelve-month base and free-cash-flow margin stays at zero, the implied year-ten free cash flow is $0. As a base-case assumption, if revenue compounds at a 5% growth path off the same $148 million base, yielding approximately $241 million of year-ten revenue, and if free-cash-flow margin normalizes to 5%, the implied year-ten free cash flow is approximately $12 million. As a bull-case assumption, if revenue compounds at a 10% growth path off the same $148 million base, yielding approximately $384 million of year-ten revenue, and if free-cash-flow margin normalizes to 10%, the implied year-ten free cash flow is approximately $38 million. A conventional 10% to 12% discount-rate environment would discount the bull-case $38 million back ten years to a present value on the order of $13 million to $14 million, which is below the dated market capitalization of $37.5 million and which therefore does not independently support the dated price. Every input here is an author assumption rather than an audited fair value.

The reverse-DCF implication is qualitative: the dated market capitalization of $37.5 million appears to require the company to convert the current software pivot into positive recurring free cash flow within roughly five years, with a margin path that materially exceeds current levels and with a debt-paydown trajectory that does not depend on additional shareholder dilution. No filing evidence currently supports that operating normalization.

Five Value-Investing Lenses

Benjamin Graham Lens

Graham's discipline starts with downside protection, and the available balance sheet fails that test. Stockholders' deficit of $270.0 million, current liabilities exceeding current assets by $18.2 million, and $319.2 million of senior secured plus convertible plus financing obligations against $38.4 million of cash imply that, in any liquidation scenario, common-equity holders would receive nothing before secured and convertible creditors. The current ratio is below 1.0. Trailing-twelve-month operating cash flow of $11.6 million is misleading because it is dominated by working-capital release, and the most recent six-month figure is negative. A Graham-style test requires normalized earnings power that exceeds the cost of debt by a clear margin and a price that offers a margin of safety against a worst-case liquidation; neither condition is met here.

Benjamin Franklin Lens

Franklin's modern lesson is thrift, discipline, compounding, and the recognition that capital spent on one pursuit is capital not available for another. Stem has spent heavily on the software pivot, including a 27% workforce reduction, an impairment of assets held for sale of $2.8 million in the first half of 2026, and a $2.6 million investment in internally-developed software. The opportunity cost is the roughly $33.3 million of net loss incurred in the first half, plus the dilution embedded in the 595,303 ATM shares already issued. A disciplined operator would ask whether the next dollar of operating spend is producing measurable recurring-revenue traction or simply deferring recognition of structural losses. The filings do not currently answer that question.

Warren Buffett Lens

The Buffett-style test requires owner-level economics, identifiable moat, honest management, and capital allocation that compounds per-share value. The available filings do not show positive owner-level cash generation: trailing-twelve-month operating cash flow is positive only because of working-capital release, free cash flow is unavailable, and the $220.0 million debt-extinguishment gain was a non-recurring accounting event. Capital allocation has consisted of cost reduction, software capitalization, the 2030 Senior Secured Notes exchange, and a small ATM program. None of these actions clearly compounds per-share value when the company is still consuming cash and diluting shareholders. Honest disclosure appears adequate, but the moat claim around scale and operating data has not yet been evidenced in services-revenue growth.

Charlie Munger Lens

Munger would invert the thesis and ask how this investment could permanently lose money. The most plausible path is straightforward: software-and-services revenue continues to decline because of the deliberate exit from battery resale, the company fails to convert the pivot into positive free cash flow, debt-service obligations of roughly $18.4 million of cash interest over six months overwhelm the $38.4 million cash balance, and management is forced to issue additional equity at unfavorable prices or to negotiate a further debt restructuring that dilutes existing holders. A second inversion asks whether the market is paying a premium for optionality that the balance sheet cannot afford to fund. The current price appears to assume a successful pivot that the filings do not yet evidence. Munger would also flag that the qualitative thesis, "energy transition plus AI plus scale," is exactly the kind of narrative that his lollapalooza warnings target.

Li Lu Lens

Li Lu would test the circle-of-competence boundary and the durability of advantage. Stem sits at the intersection of energy software, asset optimization, and AI-enabled services, which is a crowded space with established competitors and where switching costs are not yet documented in the filings. The 10-K narrative is heavy on vision and light on durable, externally verifiable advantages. The 36 GW of solar and 2 GWh of storage under management is a meaningful operational footprint, but the reader cannot tell from the filings how much of that footprint is recurring-revenue-producing versus one-time project revenue. Li Lu would also emphasize that culture is hard to read from filings, but observable evidence of executive turnover, including separation agreements with multiple senior officers, suggests organizational strain that the reader should note rather than dismiss. The dated market capitalization of $37.5 million looks more like a venture-stage option than a high-quality compounder, and that is not Li Lu's circle.

Risks, Disconfirming Evidence, and What Would Prove This Thesis Wrong

Prominent growth narratives can amplify attention bias and the price investors are willing to pay before cash flows arrive. That does not negate the operating evidence, but it raises the burden of proof for valuation and makes execution shortfalls more costly. The most plausible thesis breaker is renewed deterioration in services revenue, because the company's pivot assumes that software and services can replace the hardware revenue being deliberately exited. If the Q3 2026 services-revenue line prints flat or down year-over-year, the pivot narrative collapses. A second thesis breaker is any covenant trip or amendment to the 2030 Senior Secured Notes, because the notes carry an 11% cash-pay / 12% paid-in-kind toggle, are secured, and would impair equity value if restructured. A third thesis breaker is a large equity raise at a steep discount to the dated price, because the ATM program is open and additional dilution would meaningfully reduce per-share value. A fourth thesis breaker is a goodwill or intangible-asset impairment, given the $113.0 million of intangibles on the balance sheet.

Disconfirming evidence in either direction would include disclosure of multi-year recurring-revenue contract backlog, services-renewal-rate evidence above 90%, services-attached-to-hardware attach rate above a defined threshold, and operating cash flow that turns positive on a trailing-twelve-month basis without working-capital support. The filings do not provide any of those metrics, so the reader should treat the absence as a gap rather than as a positive. Management's stated expectation of profitable operations is forward-looking and is not evidence of current performance. The 10-K flags risks including execution of the software-and-services strategy, supplier delivery, macroeconomic pressures, and the impact of the One Big Beautiful Bill Act; the reader should weigh each of those risks rather than dismiss them.

Investment Conclusion

The current dated market capitalization of $37.5 million sits above any bear-case and base-case ten-year free-cash-flow normalization built from the available SEC data, while the bull-case path requires the dated market to underwrite an operating margin and free-cash-flow conversion that the filings do not currently evidence. The balance sheet, with $270.0 million of stockholders' deficit, $38.4 million of cash, and $319.2 million of debt plus financing obligations, does not provide the margin-of-safety cushion that a Graham-style or Li Lu-style investor would demand. Stock-based compensation, ATM dilution, and a senior secured note coupon of 11% to 12% are all continuing drains on per-share value. The dated price reflects optionality on a software pivot, not a margin of safety on current cash generation. The investment conclusion is "Avoid pending better evidence" because the dated market capitalization prices in a successful software pivot and operating normalization that the filings do not yet support.

This article is for educational purposes only and is not investment advice.