Garrett Motion: A $300 Million Promise Inside a Shrinking Car Business

Industrial gensets keep beating guidance while the stock hands back a third of its re-rating — and management still won't size the cooling prize

GTX · Earnings Call · 2026-10-29

The ladder nobody expected to climb this fast

In October 2025, Garrett Motion told investors it was already doing $100 million a year in turbochargers for stationary power generation — big gensets, most of them backing up data centers. Nine months later, on the Q2 2026 call, Olivier Rabiller lifted the full-year figure to about $200 million and conceded it was “a little bit ahead of what we even shared with you 2 months ago.” — Olivier Rabiller, CEO · 2026-07-29 Yesterday he did it a third time. industrial sales should now reach $230 million this year and roughly $300 million in 2027 — 30% organic growth, and on CFO Sean Deason's framing, overwhelmingly genset. Three raises in four quarters, on the smallest line in the P&L. It is the only part of the Garrett story where the company has consistently beaten its own numbers, and it is margin-friendly for a boring reason: the capacity to build big industrial turbos is already installed. As Deason put it, “that capacity is in place and, for us, we know how to do it.” — Sean Deason, Chief Financial Officer (CFO) · 2026-10-29 Then the deflation: $230 million is about 6% of this year's $3.8 billion revenue base. Real, compounding, and still not the thing that moves the multiple.

A second customer, and still no number

Buried in the same release is the more consequential shift. Through the winter and spring, industrial cooling was a one-partner story. Analysts chased Trane exclusivity on both the February and April calls, and Rabiller kept demurring: “I've told you that we are having discussions... it goes beyond what we've announced with Trane.” — Olivier Rabiller, Chief Executive Officer · 2026-04-30 Yesterday he produced proof — a new E-Cooling Compressor award with Multistack, a chiller customer Garrett had been cultivating "in the background for quite some time." The keyword record captures the pivot neatly. Trane was among the biggest gainers in Garrett's own vocabulary in Q1 2026 and among the sharpest decliners by Q3. The partner name that defined the narrative has been replaced by a vertical. But watch what happens when four analysts in a row try to size it. Deutsche Bank floats $500 million. Stifel asks for magnitude and timing. BNP asks what sales level triggers material margin uplift. JPMorgan asks about 2027. The answers are a masterclass in enthusiastic deflection:

I would say it is a little bit premature to give you exactly the size of this. One more time, for the same reason I answer before, it is a very fluid marketplace. We are getting into a new technology. That new technology will have an adoption curve.
Olivier Rabiller, Chief Executive Officer (CEO) · 2026-10-29

That adoption curve is doing a lot of work. "Fluid marketplace" and "in flux" recur so often they became their own keyword this quarter. The hedge is defensible — Trane field-testing only just confirmed the performance case, and customers haven't yet committed to applications or programs. But it means the $300 million industrial number is the only hard figure investors own, and it rests on gensets, not on the cooling technology that is supposed to re-rate the company. In February, management put industrial cooling at "in excess of 5% of revenue by the end of the decade" and called it accretive from day one. Nine months on, the shape of that ramp is still a promise.

The core is doing the opposite of the story

Set the industrial ramp against everything else. Net sales of $930 million grew 3% organically — but only after roughly a 3-point drag from tariff pass-through, which lowers the top line without touching profit. Light vehicle volumes are, in Deason's words, “off slightly.” — Sean Deason, Chief Financial Officer (CFO) · 2026-10-29 And the gasoline softness Hamed Khorsand probed is essentially a China problem: “it's much more movements that we've seen in Asia and especially in China... it has been very much down on the domestic front.” — Olivier Rabiller, Chief Executive Officer (CEO) · 2026-10-29 Rabiller would not forecast 2027 for autos at all — “I'm not sure I'm having a crystal ball that is more accurate than anyone else” — Olivier Rabiller, Chief Executive Officer (CEO) · 2026-10-29 — and pointedly declined to handicap mooted European quotas on Chinese vehicle imports, beyond noting Garrett sells to the exporters too, making China exports the one genuinely two-way piece of the portfolio. So the shape is a mid-single-digit organic grower whose growth is increasingly concentrated in industrial and commercial vehicle, bolted onto a light-vehicle base that is flat-to-down and geographically lurching. Full-year guidance was reaffirmed at the midpoint — $580 million adjusted EBIT, a 15.3% margin — only because strong operating performance is expected to offset FX headwinds concentrated in Q4. That is a margin step-down hidden inside a reaffirmation, and it follows a third quarter that printed 16.3%. Operating margin of 12.9% in the June quarter was up 1.7 points year over year, extending a run of five straight quarters above 10% after the 2022–23 trough. Revenue, though, has been boxed into a $900 million–$1 billion band since 2023. The margin is being earned, not rented with volume.

Why a better company is a worse stock

This is the tension that makes the report worth reading. Garrett spent six quarters re-rating: price-to-revenue went from under 0.6x to 1.8x, and price-to-earnings from about 7x to 19x, as the market discovered the data-center adjacency. Then it broke. The shares are down 32.6% in ninety days and 38% from the June 30 peak of $36.23 — a drawdown that began essentially at the Q2 print and continued straight through a Q3 that beat on margin and cash conversion. The tape explains a good deal of it. The AI data center complex — the theme Garrett's gensets and chillers lever to — is the worst-performing cluster in the global keyword tape over the last ninety days, roughly seven advancers against forty-nine decliners. The same window rewarded oil and gas producers instead. Garrett is, functionally, being traded as a data-center derivative at precisely the moment that trade unwound. Which makes the capital return the tell, not the footnote. The company repurchased $57 million of stock in the quarter, $172 million year-to-date, repaid $50 million of term loan, and authorized a fresh $250 million program for 2027 — against net debt of roughly $1.2 billion that has been flat since 2024. Management is not pre-funding a cooling capacity build; it is buying its own equity at a 38% discount to the summer. Debt is over 95% fixed and swapped into euros at an all-in cash rate near 5.5%, with no near-term maturities, so the balance sheet can afford the bet either way. Whether it should be spending it on the ramp is the question the buyback implicitly answers.

The one thing to watch

Garrett no longer has an industrial credibility problem — three consecutive raises have settled that. It has an arithmetic problem. Revenue of $976 million in the June quarter was up 7%, but the three-year trend is a plateau. Lifting industrial from $230 million to $300 million is easy to say. Getting the consolidated top line to compound while light vehicle and China domestic shrink is harder, and a Q4 FX headwind suggests the translation math is already working against them. The next real datapoint is not the Q4 print. It is whether Multistack and the Trane field tests convert into a second and third named OEM — with a dollar figure attached. Until then the market owns a $300 million promise, a shrinking car business, and a stock that has already handed back a third of its re-rating.