Stem's Shift to Software Turns Profit Margins Around
Stem Inc. (NYSE: STEM) came to its Q4 and full-year 2025 earnings call with a narrative that the market has heard too many times in climate technology: moving away from being "heavy hardware" to becoming a software company. This time, however, the numbers back up the shift.
In 2025, Stem reported $156.3 million in revenue, a +8% year-over-year increase, and—more importantly—a deliberate reconfiguration of its revenue mix: software, services, and edge hardware grew 25% to $141.4 million, while battery resale decreased “as planned” to exit a low-margin, high-operational-risk business. This change translated into a profitability leap: the GAAP gross margin rose to 38% (vs. -8% in 2024) and the non-GAAP gross margin reached 46%. The year concluded with positive adjusted EBITDA of $6.7 million and operating cash flow of $6.9 million. Furthermore, the company guided for 2026 with adjusted EBITDA of $10-15 million and ARR of $65-70 million.
The headline of “first-time profitability” might distract some. A more insightful takeaway for a CEO or CFO is that Stem demonstrated that a transformation toward software can only be sustained when the company manages to elevate the willingness to pay of customers with an offering that ensures certainty, reduces friction, and transforms complex projects into recurring revenue. This is not corporate poetry; it’s product economics.
Stopping Battery Resale is Not a Strategy, It's Financial Hygiene
The clean energy market often punishes a common misconception: the belief that selling more “stuff” equates to building value. Reselling hardware—especially batteries and storage projects—tends to generate large revenue with compressed margins, lengthy collection cycles, exposure to project delays, and an accountability chain that typically ends in scope disputes.
Stem's results illustrate a decision that sounds obvious but that few companies execute in a timely manner: pushing the business toward lines where value is captured by performance and continuity. In Q4 2025, total revenues dropped 15% to $47.2 million, but software, services, and edge hardware increased 62% to $46.5 million. At the same time, the GAAP gross margin for the quarter reached 49% and the non-GAAP gross margin was 45%.
This arithmetic reveals the real mechanics of the shift: Stem accepted the need to forgo “easily inflated” revenues (hardware) to focus on income with the capacity to expand margins. It’s the type of decision that improves a profit and loss statement, but above all, reduces fragility. For a company operating energy projects, volatility does not kill it due to lack of demand; it kills it due to friction: change orders, delivery times, penalties, and service costs that were poorly budgeted.
The call also highlighted numbers pertaining to discipline: in Q4, Stem reported a 50% drop in cash operating expenses down to $18 million. For the year, the briefing mentioned cash operating expenses of $79.7 million, a reduction of 42%. In real transformation, cuts do not represent “austerity”; they signal a redesign of the system so that future growth does not require duplicating structures.
The Key Asset is ARR, but What’s Sold is Operational Certainty
Stem ended 2025 with ARR of $61.1 million (+16% year-over-year) and CARR of $67.2 million. In software, ARR is not a vanity metric; it serves as a proxy for two things: (1) the level of pain that the customer is willing to pay to alleviate each year and (2) the customer’s confidence that the provider will continue delivering.
In energy, “delivering” means sustaining performance in a challenging environment: climatic variability, interconnection, curtailment, maintenance, dispatch rules, and a stack of suppliers that never lines up perfectly. There, the promise of an EMS or monitoring system is not purchased for functionalities; it’s purchased for risk reduction. In the call, the company positioned its software as a control pillar for clean energy projects and reported growth in its PowerTrack suite: PowerTrack ARR of $41 million and 6 GW of added solar assets to reach 36 GW under management.
That detail—managing 36 GW—matters less as a marketing milestone and more as a commercial argument. The broader the operational history, the higher the perceived certainty and the lower the cost to convince the next customer. It's the type of advantage that allows maintaining high prices without resorting to permanent discounts.
The launch of PowerTrack Sage, a beta AI assistant with 80+ customers expected to have general availability by late March 2026, serves as a lever for efficiency and monetization. Stem mentions a “light” version for everyone and premium tiers for upselling. If designed and executed rigorously, it does something that many AI products struggle to achieve: making AI a component that reduces operational friction while also enabling a pricing tier based on captured value. Without measurable results, a “co-pilot” is mere decoration. With repeatable outcomes, it becomes margin.
The Transformation Test is Not EBITDA, It’s the Change in Sales Friction
The year 2025 marked Stem’s first year with positive adjusted EBITDA, and in the quarter, the adjusted EBITDA rose to $5.5 million. This is good, but it's not decisive. The tough test for a software model in energy is how much of the sale can be standardized without breaking the promise.
The company reported a 10% sequential increase in utility-scale bookings in Q4, largely driven by international solar. It also mentioned an engagement of 100 MWh of its PowerTrack Energy Management System (EMS) with Everyray (a developer and EPC in Germany), with commercial operations expected by summer 2027 according to the transcript. Additionally, it highlighted a “brownfield” agreement to operate the A4Site storage portfolio for a utility in Southern California.
These two examples illustrate the same thesis: profitable growth in this category cannot be won by “more features,” but by reducing the customer’s effort to achieve an outcome: (a) integrate, (b) operate, (c) demonstrate performance, and (d) maintain compliance. In utility-scale and brownfield projects, typical friction arises from migrations, compatibility with existing assets, and operational responsibilities. If Stem can package its EMS and managed services as a system with clear operational guarantees, pricing will cease to be a problem and will become a consequence.
The other friction indicator is cash. Stem closed the quarter with $48.9 million in cash and reported positive operating cash flow in both Q4 ($8.2 million) and for the year ($6.9 million). In a business that historically could appear “project-dependent,” this figure reduces perceived risk for both large customers and partners. The transformation towards software becomes credible when the provider stops behaving like a cash-strapped contractor and begins acting like a stable platform.
The 2026 Risk: Monetizing AI Without Falling into the Promise Bingo
Stem guided for 2026 with adjusted EBITDA of $10-15 million, operating cash flow of $0-10 million, and ARR of $65-70 million. The cautious cash range reveals that there remains sensitivity to billing timings, implementation, and commercial spending. Such caution is healthy: the quickest way to destroy a transformation is to accelerate sales with discounts or broad promises that later require overservice.
There’s also an undeniable market fact: after the earnings announcement, shares rose 5.79% to $10.05, but the briefing indicates a market capitalization of $85.75 million and a 33% drop for the year. This combination suggests a company that, despite operational improvements, remains under pressure regarding stock market credibility. And, when credibility tightens, many companies make the mistake of reverting to easy volume: pushing hardware revenue or unprofitable customizations to "meet the quarter."
The antidote is to maintain a business architecture that prioritizes certainty and repeatability. PowerTrack Sage can be a relevant piece if used to: (1) reduce internal delivery and support costs, (2) shorten diagnosis and resolution cycles, and (3) justify premium tiers with verifiable outcomes. If AI becomes a generic claim, the market will penalize it. If it becomes evidence and automation that reduces friction, the customer will pay more without dispute.
The earnings call makes it clear that Stem understands the movement: less resale, more recurrence; less fixed spending, more leverage; more assets under management, more trust; more margin, more oxygen. That is the difference between “transforming” and merely changing the slide deck.
Stem did not win in 2025 by bettering its software narrative. It won because it redesigned its offering so that the customer purchases operational certainty with less friction and more predictability, which elevates the willingness to pay. In any industry, commercial success is determined by this equation: reduce friction, maximize outcome certainty, and capture prices aligned with real value until the proposition becomes impossible to ignore.
