Open in interactive viewer → charts, metric popovers & call review

Tecogen Stops Selling the Chiller and Starts Selling the Box

A modular retrofit package is a genuine capacity fix for a company that could only build 100 chillers a year — but the stock has already halved waiting for one purchase order
TGEN · Earnings Call · 2026-11-12

Tecogen Stops Selling the Chiller and Starts Selling the Box

Tecogen reported Q3 2026 on November 12, and the headline number was fine: revenue up 9.6% to $7.9 million, gross margin up seven points to 37.4%. But the real news was buried in the middle of CEO Abinand Rangesh's prepared remarks, and it is a quiet admission that the company's data center strategy was structurally blocked — plus a new idea for how to unblock it.

The problem was never the demo. It was the floor space.

For six quarters Tecogen has been converting hyperscalers with dual power source chiller demonstrations, then losing them on the follow-up question: can you deliver at volume? Turns out the constraint wasn't demand, it was the machine itself. In March, Rangesh told analysts the company could build “up to 100 chillers a year” — Abinand Rangesh, Chief Executive Officer · 2026-05-13 — a ceiling that caps the data-center business at roughly three to four times the company's best ever product year, or $30–40 million, per his own framing on that call. A hyperscaler planning gigawatt phases does not buy from a supplier whose entire output is 100 units.

So the company shrank the product. The new offering is not a chiller at all — it is a box containing the engine, the inverter, the emissions catalyst, the microgrid controls and the heat rejection, which bolts onto a chiller or dry cooler the customer has already specified. As Rangesh put it:

We have designed a modular dual-power-source package that can be retrofitted to third party chillers dry coolers, and any other data center load. It has all the advantages of the dual power source chiller, and is protected by the same patents, but is much easier to scale up production.

Abinand Rangesh, Chief Executive Officer · 2026-11-12

Three consequences follow, and each is a genuine change in the business model rather than a marketing reframe. First, capacity: the same factory floor, minus the refrigeration assembly, yields “more than 2x of what we could do in terms of the chiller” — Abinand Rangesh, Chief Executive Officer · 2026-11-12, and additional capacity is rentable space rather than tooling. Second, the supply chain — as Rangesh argued, “it uses a portion of the supply chain that is not constrained today” — Abinand Rangesh, Chief Executive Officer · 2026-11-12, because the engines Tecogen already buys in volume for its service fleet are not the scarce single large reciprocating engines the industry is scrambling for. Third, working capital: a smaller bill of materials against customer deposits of 25–40% (CFO Roger Deschenes laid that deposit structure out back in March) means deposits fund more of the material cost, which is exactly the cash problem that has dogged this company for a decade.

The vocabulary confirms the pivot is real, not rhetorical. This quarter's keyword set is dominated by power module, building block — Rangesh describes “stacking the building blocks together so that whatever the customer is using on-site, the building blocks just get bigger” — Abinand Rangesh, Chief Executive Officer · 2026-11-12 — and utility power, the scarce resource the box exists to conserve. Meanwhile the language of previous quarters recedes: big name and potential customer both fell away sharply in this period's keyword momentum, as did dual power source chiller, the flagship framing of 2025. The company has stopped hunting a marquee first customer and started designing itself into other people's drawings.

The base business is doing the heavy lifting, and the P&L shows it

Strip out the data center story and Q3 was the best operating quarter Tecogen has posted in years. Services revenue rose 13% to $4.5 million and services gross margin went from 25.3% to 42.6% — the payoff from the painful engine replacements and headcount cut that management flagged as a drag in prior calls. Product gross margin fell from 36.8% to 27.1% on hybrid drive air-cooler material costs, so all of the consolidated margin expansion came from service and energy production. That matters: service is annuity revenue, and it has now printed above $4.3 million for three straight quarters.

The long arc is less flattering. Revenue is still recovering from a decade of decline, and operating margins remain deeply negative — which is why a $1.4 million quarterly loss against $7.9 million of revenue is an improvement story only in the sense that gross profit is finally covering more of the overhead. SG&A still runs about 2x gross profit, so the breakeven math depends almost entirely on scaling revenue rather than trimming cost.

Two small credulity flags. Rangesh said the net loss "decreased... to $1.4 million compared to $1.3 million" — it widened. And he described the cash position as "higher than it was at the end of the third quarter" while reporting the third quarter. These are sloppy, not sinister, but on a call where the entire equity story rests on execution credibility, sloppiness reads louder than it should.

Three quarters of "stay tuned"

The most conspicuous absence is Vertiv. On the May call it was the approved-vendor route into 25–50 megawatts of specified projects; in August, when Eric Stine asked directly for an update on the master partnership agreement, Rangesh answered: “Yes. So I think things are in a very good place overall. With them. I just cannot comment at this point” — Abinand Rangesh, Unknown · 2026-08-13. This quarter, two analysts asked. Both got the same refusal: “All I am going to say on this 1 for the moment is stay tuned.” — Abinand Rangesh, Chief Executive Officer · 2026-11-12 A third of the year has now passed with the company's single most concrete commercial relationship described only as confidential. The new modular product may or may not sit inside that relationship — management won't say — and that ambiguity is itself the disclosure.

The other dropped thread is the legacy business framing. Prior calls treated the base business as the cash engine funding a data-center bet; this call folds both into one product platform, arguing the same box solves the same power-constraint problem for hospitals and industrial sites. That is an honest rationalization, but it also quietly concedes the data center alone was not going to carry the company.

The tape has already delivered its verdict

Tecogen's stock peaked at $4.19 on August 13 — the day of the last earnings call — and has since fallen 52.7% over ninety trading days, a drawdown of 53.5%. That is not idiosyncratic. The global tape shows AI data centers down roughly 16% over 30 days with 13 advancers against 57 decliners, and data center capacity down 24% over the same window with no advancers at all. Even Applied Materials, reporting this week, flagged a slowdown in data center buildout. Tecogen is pitching a data-center product into a market where the suppliers of the buildout are being de-rated.

Valuation is the uncomfortable corollary. At 6.4x revenue — after the drawdown — Tecogen is still priced as though the modular module works, ships, and converts. The Q3 gross margin improvement and the deposit-funded working capital profile are the first tangible evidence in years that the option has a path to being exercised. The 90-day collapse says the market wants proof of a purchase order, not another well-argued diagram of a building block.

What changed at this company is real: it found a way to sell its differentiated technology without having to manufacture the commodity around it. What did not change is the thing that has held the stock down for a decade — no order yet, and no willingness to say what is happening with the one partner everyone is asking about.