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

908 Devices: A Record Quarter Into a 22% Drawdown

Record device shipments and NIRLab's first full quarter arrived into a market that had already sold the stock down 22%; the quiet story is AVCAD turning from subcontract to COTS vendor
MASS · Earnings Call · 2026-11-10

The print versus the tape

908 Devices reported the best operating quarter in its history on November 10 and got no credit for it in advance. Revenue of $19 million, up 35%, more than 300 devices shipped — a company record — and an installed base of roughly 4,400 units. “We delivered revenue of $19 million, up 35% year-over-year, driven by continued strength in our core business” — Kevin Knopp, Chief Executive Officer and Co-Founder · 2026-11-10, Kevin Knopp told the call. Guidance was raised at the low end to $68.5–70 million, and the adjusted EBITDA loss is tracking to the mid-single-digit millions, roughly a third of 2025's level. And yet the tape tells the opposite story. The shares are up 24% over ninety trading days, but they peaked at $12.87 on October 28 and were down 22.4% from that high by the report date. A $413 million market cap that had re-rated hard through the autumn — trailing price-to-revenue sat at 5.4x at the Q2 mark, up about 149% year over year — de-rated into a beat-and-raise. That is usually a positioning story rather than a fundamental one, and it sets up the question the call actually had to answer: is this a hardware quarter or a platform quarter? It was mostly hardware. FTIR was 55% of Q3 revenue, handheld product and service revenue rose 35% to $17.8 million, and roughly $4 million of the quarter came from a single Asia Pacific ProtectIR award. “We did mention the Asia Pacific, about $4 million came from that ProtectIR opportunity. So that was a high concentration” — Joseph Griffith, Chief Financial Officer · 2026-11-10, Joe Griffith conceded under a question about 41% of revenue landing with two customers. He was emphatic that there were no order shifts or pull-forwards from Q4 — but a quarter that good on two customers is a quarter that has to be re-earned.

AVCAD: a slip that may be an upgrade

The most consequential thing on this call was buried in Q&A. AVCAD — the Army's next-generation aerosol and vapor detection program, and one of three named growth catalysts for years — has moved from subcontractor to candidate commercial vendor. Management delivered MX908 units to the office running analytical testing and evaluation, and framed it as a signal that the program is seriously considering a commercial-off-the-shelf path, with potential orders in 2027. The history matters. In March, the award was imminent:

I remain very encouraged and positive that we will see an award this spring, and hence we are factoring it into some of the guidance discussion today.

Kevin J. Knopp, Chief Executive Officer and Co-Founder · 2026-03-03
By August it had slipped: “On AVCAD, it's taking a little bit longer than we and the market expected, but our confidence in winning hasn't moved” — Kevin Knopp, Chief Executive Officer and Co-Founder · 2026-08-11, with the $10 million-plus opportunity pushed to the future and stripped from guidance. Today it carries zero revenue in the 2026 guide. What is new is the economics of the path now being contemplated:

You'd expect the commercial margins to be more in that mid-50s, mid- to high 50s that we see with our other products under COTS versus more OEM level margins that might have been in the 30% to 40% range.

Joseph Griffith, Chief Financial Officer · 2026-11-10
That is a genuine trade: later revenue, but roughly 1,500–2,500 basis points more gross margin on the same program, because 908 would sell the finished COTS solution rather than subsystems. It also ties the company to a market-level theme rather than a company-specific hope — the Department of War procurement reform push toward speed and commercial products is showing up across defense reporters this week, from 3DA.AX to ELMT. AVCAD is no longer a single-program bet; it is a bet on a procurement doctrine.

The recurring-revenue paradox

The company's top-ranked theme this quarter is connected services, and Knopp spent more airtime on the connected vision — expert Reachback, Team Leader fleet management, cloud subscriptions — than on any product. The mechanics are attractive when described: “You sell a device and then it's about -- after that, it's about 50%, then again in a subscription fee each year thereafter” — Kevin Knopp, Chief Executive Officer and Co-Founder · 2026-11-10. But the numbers went the other way. Recurring revenue was $5.1 million, up only $0.2 million year over year, and fell to 27% of the total because device placements were so strong. The subscription model the company bought with NIRLab is real, but it is currently being outgrown by its own hardware. The same aspiration has now been uttered on two consecutive calls: “starting to creep up into the 30s, kind of approaching that 40% is what we aspire to” — Joseph Griffith, Chief Financial Officer · 2026-08-11. Watch whether the mix moves once the lapsing mass-spec federal service contract — the December 2025 funding loss — rolls off in Q4, which management says removes the drag.

NIRLab, next-gen MX, and the 2027 setup

NIRLab is the cleanest new fact. Roughly 100 devices sold in its first full quarter of ownership versus 35 in the first two months, more than half to customers new to NIRLab, with roughly $1.2 million of revenue and a path to $5 million-plus in 2027: “we see an opportunity as a key growth driver to get to $5 million plus in 2027” — Joseph Griffith, Chief Financial Officer · 2026-11-10. The usage stat matters more than the unit count — more than 1 million cumulative scans, roughly one per minute, up 74% year over year. That is the metric that makes a subscription durable, and it is being reinforced by a genuine regulatory tailwind: tightening controls on kratom compounds and the drift away from colorimetric kits. The second 2027 driver is the next-generation MX908, with pilot manufacturing running in Danbury and early-access shipments targeted for year-end, ahead of an upgrade cycle across more than 3,000 installed units. This has been promised since 2025, so the milestone rather than the promise is the news. The third — aerosol detection for A-series nerve agents and pharmaceutical-based agents, where a handful of MX908 units went to a U.S. military customer to close an immediate capability gap — is the clearest example of platform leadership translating into orders rather than slides. Financially, the company can afford patience: $99.1 million of cash, no debt, $2.4 million consumed in the quarter, and free cash flow turned marginally positive in Q2 after four years of burn. Gross margin, at 51.9% GAAP in Q2 and 53% in Q3 with 58% adjusted, is a mix story now — lower-margin international ProtectIR volume against high-margin subscription revenue — and management still guides the mid-to-high 50s. The bear case writes itself: a $413 million company whose biggest quarter leaned on two customers, whose flagship defense program slipped a year, and whose marquee strategic focus is shrinking as a percentage of revenue. The bull case is narrower but sharper — NIRLab compounding at a rate nobody modelled, a genuine product cycle in MX908, and a possible conversion from 30–40% OEM margins to mid-50s commercial margins on a program the Department of War wants to move faster. The stock's 22% retreat into a good quarter says the market is pricing the first case; the second is what the next two prints have to prove.