BorgWarner Beat, Raised — and Fell 31% From Its Peak. The Tape Is Repricing the Data Center Story, Not the Quarter

A record-ish margin quarter and a $300M turbine-generator launch-in-waiting meet a market that has stopped paying for 'early 2027'

BWA · Earnings Call · 2026-10-29

The quarter that should have been enough

BorgWarner reported on October 29 and, on the arithmetic, there was little to fault. Sales came in just under $3.5 billion. Adjusted operating margin was 10.9%, up 20 basis points year over year, while adjusted operating income held at $381 million on $112 million less organic sales. Adjusted EPS rose 12%. Free cash flow of $195 million went straight back out the door — $175 million of buybacks plus $35 million of dividends — and Craig Aaron used the call to nudge full-year EPS guidance to $5.20–$5.35 a share, “driven by our third quarter and currently planned fourth quarter share repurchases and a decline in our full-year effective tax rate” — Craig Aaron, Chief Financial Officer · 2026-10-29. And yet BWA sits 30.8% below its June 4 peak of $77.03, and is down 21.8% over the last 90 trading days alone. A beat-and-raise into a 22% quarterly drawdown. That gap — compounding earnings versus a collapsing multiple — is the real story of this report, and it has almost nothing to do with light vehicles.

From 'product readiness' to 'quote activity'

The cleanest way to see what changed inside this company is the language rotation. A quarter ago the top theme was product readiness — the language of engineering milestones. This quarter, the industrial program has moved into the commercial register: microgrid inverter, quote activity, silicon carbide. The specifics are real. BorgWarner has shipped microgrid inverter samples to seven customers, begun initial quoting with its first major customer, and expects UL certification in early 2027. The North Carolina final-assembly plant for the turbine generator is built, with production trials by year-end and $300 million of 2027 revenue still the plan. A new 800-volt power supply unit now reaches inside the data hall, not just the grid edge — a genuine extension of the addressable market, and one that puts BorgWarner in the same architectural conversation as the 800V DC rack-power crowd reporting this week. So why the multiple collapse? Because the second derivative got worse, not better. Asked whether permitting and political noise was affecting negotiations, Joe Fadool conceded something new:

We generally see some of the current political environment as possibly slowing the finalization of some of the program awards or program agreements. But overall, the activity is there... We see this as probably a very short-term issue. We are looking beyond the midterms.
Joe Fadool, President and CEO · 2026-10-29

That is the first time in this program's short history that management has put a timing caveat on the quoting funnel. For three quarters the message was capacity-constrained demand; now it is demand awaiting paperwork.

The missing capacity decision

The most conspicuous silence is the capacity call. In August, Colin Langan asked directly whether they would add turbine-generator capacity in the second half of 2026, and Fadool answered: “we do anticipate we're going to have to make a decision in the second half of this year on capacity” — Colin Langan, Analyst · 2026-08-05. Three months later, on the Q3 call, the second half is nearly over and no decision has been announced. Instead, Chris McNally's buy-side question about the shape of the ramp met: “We will not be giving you any kind of color until we get into early 2027” — Joe Fadool, President and CEO · 2026-10-29. Two gigawatts of installed capacity that management once described as a subset of demand, with no visible decision to add to it, is a data point the bulls have to explain. There is a second, quieter tension in the financials. The company is spending $10–15 million of incremental industrial R&D to buy its way into these markets — while total research and development fell 3% year over year to $176 million. That is roughly 4.8% of a $3.6 billion revenue base. The industrial push is being self-funded, not resourced like a startup.

Why the tape is voting no

The stock is not being judged as an auto parts company right now; it is being judged as a levered play on AI infrastructure, and that complex is bleeding. In the global theme tape, AI data center shows a 90-day return of roughly -5% with 49 decliners against 7 advancers — the widest negative breadth anywhere in the dataset — while data center capacity is down about 8% over the same window. BorgWarner is a small, newly minted member of that cohort, so it gets the beta without the multiple. The company's own keyword history tells the same story internally: 'data center market' was one of the sharpest momentum decliners this quarter, and 'Endeavour' — the go-to-market partner that carries the hyperscaler relationships — fell hard as a theme even before the call. The editorial read of the transcript is that the data center narrative got thinner, not thicker, in Q3. Meanwhile the boring 80% of the business is quietly fine. Light vehicle eProduct sales grew 33% to $630 million, mostly Europe, on a gross margin of 19.8%, up 2.2 points, and the light-vehicle production theme sits in the advancers column of the 360-day tape. The core is not the problem.

What 2027 has to prove

Management has been promising this outgrowth inflection since at least February, when Fadool said plainly, “I can also say I'm not satisfied with the outgrowth we've been seeing” — Joseph Fadool, President and CEO · 2026-02-11, and pointed to 2027 as the payoff year. That year is now twelve months away and the promise has narrowed to three components: modest growth over market in automotive, $300 million of turbine-generator revenue, and mid-teens conversion on the increment. Q4 itself is guided down to roughly $3.43 billion with a mid-teens decrement plus $10 million of industrial R&D. The balance sheet gives them room. Effective net cash position of -$1.6 billion, 30% better year over year funded an EPS-accretive tender, and $1 billion of 2026 free cash flow supports the stated 70% payout. But valuation is the pressure point: at 19x operating income, up 113% year over year, this is no longer a cheap cyclical being marked up on a hidden asset. It is a fairly-valued auto supplier carrying an option on data center power, and options get marked to market when their underlying complex wobbles. The fair read: nothing broke this quarter. What changed is that the data center story stopped being free. Until the UL certificate lands, the first inverter award is signed, and turbine generators actually ship, BorgWarner's improving eProduct sales and expanding margins will keep fighting a tape that has already decided how much it is willing to pay for 'early 2027.'