Mayville's Data Center Dream Meets Its Execution Tax

Revenue up 18% and critical-power demand booming, yet the stock is 57% off its June peak — the market stopped paying for the promise and started pricing the ramp

MEC · Earnings Call · 2026-11-04

Mayville's Data Center Dream Meets Its Execution Tax

On November 4, Mayville Engineering reported a third quarter that was, on its face, a growth story: revenue up roughly 18% year-over-year to $170.4 million, and its data center and critical power (DCP) end market up about 65%, now 22% of sales. And yet the stock sits 57% below where it traded four months ago. The gap between the headline and the tape is the entire story here — this is not a demand problem, it is a company paying an execution tax to build the capacity that demand requires, and a market that has abruptly stopped financing the wait.

The tape already voted — and not on Mayville alone

MEC's own price action tells the story before the transcript does: a 77% run over nine weeks, then a 63% collapse over fifteen, leaving the shares down 57.3% over the last 90 days from a June 30 peak near $37. The report lands into a name already deep in a drawdown, not a breakout. Crucially, this isn't idiosyncratic. The global AI-adjacent data center tape has rolled over broadly — the AI data center cluster ranks among the 30-day decliners with 56 negative tickers against only 11 positive, and names tied to capacity buildout (data center capacity, critical IT load, high-bandwidth memory) are all fading across windows. Mayville is a small-cap metal fabricator riding a wave the whole market is now re-underwriting. Its fate is levered to whether the wave is durable, and right now the tape is voting no.

What actually changed: $20 million moved, and the margin base reset

Buried in the Q&A is the real number. Asked about the Q4 margin step-down, CEO Jagadeesh Reddy was blunt: roughly “about $20 million approximately of DCP revenues pushed out of second half of 26 into 2027” — Jagadeesh A. Reddy, President and CEO · 2026-11-04, spread across more than one customer still wrestling with their own supply chains. CFO Rachele Lehr did the arithmetic out loud — at a 20% adjusted EBITDA margin, that shift “brings our midpoint down from that 56 to that 50” — Rachele Lehr, Chief Financial Officer · 2026-11-04. Guidance moved; Q4 is now framed as "a transitional quarter." The mechanism is the same one that has dogged this name all year: ramp related costs, premium labor, and outsourcing costs — about $3 million in the quarter. The labor detail is vivid: welders at certain plants paid "$90 an hour plus per diem" against a sub-$30 starting wage, while the company outsources laser and paint work because its own machines are still being installed. Management insists these are in line with expectations and temporary — “These costs were in line with our expectations though... we continue to view them as temporary.” — Jagadeesh A. Reddy, President and CEO · 2026-11-04

The promise keeps sliding right

The tension is historical. On the March call, the framing was that launch costs would bite in the first half and largely be done by the second; Lehr then said “we do anticipate a little bit trailing into the second half of the year” — Rachele Lehr, Chief Financial Officer · 2026-03-04. By May it was firmer: “we really expect to be at full run rate for the second half of the year” — Rachele Marie Lehr, Chief Financial Officer · 2026-05-06. August flagged timing uncertainty. Now November says ramp costs moderate in early 2027, margins improve sequentially, and the benefit is "most visible as we exit the year and move into 2028." Three calls, one word — temporary — and a runway that keeps extending into the out-year. What genuinely is new is candor about the learning curve. The company has been in industrial markets 81 years, but in DCP roughly one.

We have been in business for 81 years in industrial markets. We have been in DCP for a year. And it is a highly, you know, fast moving industry. So there were definitely some learnings in 2026.
Jagadeesh A. Reddy, President and CEO · 2026-11-04

That admission pairs with a shift on capacity: in August management was only exploring reservations and admitted it “have not signed any particular customer to a contract like that” — Jagadeesh Reddy, President and CEO · 2026-08-05; now it says it will go back to customers in Q4 and start "charging them for delays, charging them for incremental headcount." Moving from absorbing the variance to billing for it is a real, if defensive, maturation.

Why it matters: a real cycle, a priced-in promise

The bull case is not fake. The qualified DCP pipeline jumped to over $200 million from $125 million, penetration among top OEMs is under 5%, and products sit late in a build that broke ground years ago — the data center build is a genuine multiyear runway. The Georgia facility acquired October 1 is a ~$25 million investment expected to add ~$50 million of annual revenue capacity by 2028. But the financials show how much is being spent ahead of the payoff. Latest-quarter revenue of $163 million carried a negative operating income of about $3 million, a gross margin of just 10.9%, and a net cash position of roughly -$124 million. Free cash flow was a use of $3.3 million in Q3, and with capex guided to $25-35 million, the free cash flow line stays negative. Leverage is 2.9x against a 2.5x target. The irony is in the multiple. Price-to-revenue of 1.6x, up 164% year-over-year, is exactly what you'd expect from a stock that was re-rated on a secular narrative and is now de-rating as the narrative meets a lumpy P&L. So which is it? Mayville offers two stories that collide. The demand side is real and, by management's telling, accelerating — the CEO closed with "I could not be more bullish about 2027." The margin side keeps slipping into the out-year, propped up by an equity raise and a new plant, while the market's appetite for AI-adjacent buildout cools. The bull case now rests entirely on whether the "temporary" costs actually leave by 2027-28 against the promised ~15% EBITDA margin. For a $400 million-cap fabricator, that is a high-wire act — and the tape, unlike the transcript, is not being patient.