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

Xometry Clears $1 Billion and Finally Prints a Profit — Now It's Selling Its Brain, Not Just Its Marketplace

A 40% quarter and a first GAAP profit hide the real move: a quiet rotation from 'AI model' bragging to an embedded, agentic e-commerce playbook.
XMTR · Earnings Call · 2026-11-03

The milestone that finally showed up

For three years Xometry has been narrating a reacceleration that perpetually sat one quarter in the future. In Q3 2026 the arithmetic caught up with the story. Revenue rose 40% year-over-year to a record $252 million, the second straight quarter above 40%, pushing the company over a $1 billion run-rate. Marketplace gross margin touched a record 36.0%, and — the number that has eluded this company since it went public — GAAP net income came in at $1.9 million, versus a $11.6 million loss a year earlier. Adjusted EBITDA was $21.4 million, an 8.5% margin against 3.4% a year ago, and management lifted full-year revenue growth to roughly 37% from 33%–34%.

Note the timing wrinkle: the most recent filed fundamentals still describe Q2, when net income was -$5 million. The profitability turn lives in the earnings call, not yet in the 10-Q. Investors are being asked to trust a milestone that the audited tape hasn't caught up to — and the tape has rewarded them, with the stock up 18% over the last ninety trading days in a clean, unbroken uptrend.

What actually changed: the AI story lost its own headline

The most telling signal isn't in the numbers. It's in how the language moved. Two quarters ago the company's dominant extracted theme was AI model — rank one, by a wide margin, as management marketed an "interconnected" set of models spanning costing, sourcing and manufacturability. This quarter that phrase has dropped out of the top tier entirely. In its place: compounding growth, e-commerce platform, and the idea of an e-commerce playbook in custom manufacturing.

That is not AI disappearing — it is AI being absorbed into commerce language, which is precisely what a maturing infrastructure pitch looks like. The AI is no longer the product; the checkout is. Sanjeev Singh Sahni's preamble is a conscious reframing: “The three consecutive quarters of 40%+ growth in our marketplace are direct evidence that the product-led strategy put in place last year is working. We are defining the e-commerce playbook in custom manufacturing” — Sanjeev Singh Sahni, Chief Executive Officer · 2026-11-03. Read against the August call, where the same CEO sold the models themselves — “the AI models are not just independent models now. They are an ecosystem that's powering the growth” — Sanjeev Sahni, Chief Executive Officer · 2026-08-04 — the shift is unmistakable. Management has stopped selling the engine and started selling the storefront.

Siemens is no longer a promise; it is a distribution channel with a date

The strategic headline of the quarter is the conversion of the Siemens partnership from press-release excitement into a concrete product roadmap. In May, then-CEO Randy Altschuler framed it in distribution terms: “Siemens is the leading industrial software company globally. It has millions of users... we are embedding directly into their PLM and CAD software” — Randolph Altschuler, Chief Executive Officer · 2026-05-08. This quarter it has sprouted a schedule: a Designcenter beta in Q1 2027, a gradual customer ramp through the year, and — per the CFO — flow-through economics above the company's standard incremental margin because no sales and marketing is required to acquire those buyers.

In fact, I'd take this opportunity to also talk a little bit about agentic commerce, which is a true differentiation and strategic advantage for us... agents today can already interact with our site, learning about selection, categories, and capabilities, and taking it back to the foundation models.

Sanjeev Singh Sahni, Chief Executive Officer · 2026-11-03

This is where Xometry brushes against a genuinely global wave. Across this week's reporters, consumer and enterprise agents are everywhere — DoorDash, Twilio and Shopify all leaned on agent narratives, and the market's theme tape is littered with agentic infrastructure. Xometry's agentic purchasing and generative CAD language is company-specific in application but rides a broader tide. The difference is that unlike a consumer chatbot, Xometry is arguing its API-first rebuild makes it the rails those agents transact on. Whether that is vision or opportunism is the single most important question the stock now trades on.

What management stopped talking about

Here is the quieter, more important tell. The themes that powered Xometry's earlier narrative have faded from the discourse. The supplier network — a rank-eight corporate staple as recently as late 2024 — has slid into decline across recent quarters. large accounts as a discrete talking point weakened too, and even "manufacturability," the technical heart of the original pitch, lost momentum this quarter. The moat claim still rests on 5,000-odd active suppliers, but management's energy has migrated to the enterprise funnel, PunchOut integrations and the e-commerce buyer experience.

That may simply be prudent — enterprise is where the wallet share is. But it matters that the two-sided-network rhetoric is thinning just as the company reframes itself as digital infrastructure. A marketplace that increasingly looks like a software layer selling into large ERP systems is a different business model than a liquidity pool of small machine shops, and it deserves a different multiple. Meanwhile, the shopworn refrains persist: the opening innings framing — less than 1% of a $275 billion TAM — has been said on essentially every call for a year, which makes it durable conviction or durable boilerplate depending on your temperament.

The pattern analysts keep circling

Buried inside an obviously good quarter is a familiar shape. Q3 grew 40%; Q4 guidance implies total growth of just 32%–33% and marketplace growth of 35%–36%. That is a deceleration, and it echoes a ritual: in February and again in May, analysts pressed management on guidance that implied a step down. As Greg Palm of Craig-Hallum put it in May, “the full year guide... implies not just a pretty big deceleration in Marketplace growth in the second half” — Greg Palm, Analyst · 2026-05-08. The answer then, as now, was a blend of "we're being smart" and "the trends remain strong." The raise is real — roughly $7 million more EBITDA — but the beat-raise-decelerate cadence is now a pattern, not an accident.

The bar has moved, and so has the price

None of this is free. Price-to-revenue has roughly doubled to 6.8x over the past year — still well below the 2021 peak of 11.9x, but a valuation that now assumes the compounding story works. The balance sheet supports the ambition: liabilities-to-assets fell to 43.6% from a 61% peak, cash sits at $505 million, and free cash flow less stock comp is trending toward breakeven. But the "at least 20% incremental adjusted EBITDA" pledge is now the floor everyone already pays for, and the Siemens upside lands in 2027, not this year.

So the story is neither the $1 billion nor the first profit. It is that Xometry spent this quarter quietly changing what it sells — from an AI-native marketplace to an embedded intelligence layer distributed through someone else's software — while softening the two-sided-network language that once justified the business. If the agentic bet and the Siemens channel are real, this is the quarter the company grew up. If they are packaging, the multiple already reflects the wrapping.