Cognex Corporation CGNX

Key data

The business

Cognex sells machine vision: industrial cameras plus the software that reads what they see. A camera over a bottling line reads the lot code, one above a distribution-center conveyor reads every shipping label at speed, one on an assembly line checks that a chip's solder is right. Hardware is most of the revenue; service is under 10%. Customers are manufacturers and logistics operators across autos, semiconductors, consumer electronics, food and warehouse automation. The moat is switching cost and a two-decade library of vision algorithms: once a production line is validated around Cognex's cameras, swapping them means requalifying the whole line, so nobody does it casually. The weakness is the flip side. Cognex sells capital equipment into cyclical factory budgets, so when customers stop expanding, orders fall fast and together.

The numbers

The story is a deep cyclical trough now recovering, with margins snapping back faster than revenue.

Revenue set a record in the June quarter, up 17% on the year, and profit nearly doubled. The driver is margin, not just volume: gross margin was 71% against 67% a year earlier on better mix, and operating margin hit 29% in the quarter, close to the 30% Cognex earned at its 2021 peak. The low Q3 2025 profit was a tax quarter, not an operating stumble.

Here is the whole case in one frame. Revenue peaked at $1.04B in 2021, fell 19% into the 2023 automation recession, and has climbed back to a record run rate. But profit is still far below the old high, about $175M over the trailing year against $280M in 2021, because operating margin fell from 30% to the low 20s as revenue dropped and fixed costs stayed. The June quarter's 29% margin says that gap is closing. In half of 2026 the company already earned $124M, more than it made in all of 2024 or all of 2025. The variant view: the market pays 60 times earnings that are still depressed, and if margins fully normalize on record revenue the figure the bears quote collapses. The single print that settles it is whether operating margin holds near 29% for two more quarters or slides back toward 22%.

Management

Insiders are sellers, not buyers: no purchases in twelve months against $5.4M sold across eleven Form 4s, a director taking $2.7M in May and an officer $0.5M in August, plan status not disclosed on any. That is ordinary trimming into strength, not a stampede, but there is no insider vote of confidence at this price either. Capital allocation is the sharper question. Cognex bought back $151M of stock in 2025 and roughly $99M in Q1 2026, and authorized a fresh program in February. Buying steadily at 60 times earnings, near a twelve-year high, is buying its own record at its most expensive; the balance sheet can carry it, with $303M cash, token debt and net cash near $230M, but it is not obviously the best use of the money. Share count is barely down because buybacks mostly offset stock pay.

How it fails or surprises you

The cyclical rollover. Cognex sells into factory capital budgets, and the last year's surge rides an automation-spending rebound. If manufacturing PMIs roll or a heavy end market like consumer electronics or autos pauses, orders drop together, as they did in 2022-2023 when revenue fell 19%. Sixty times earnings on a down-cycle re-rates hard. First tell: sequential revenue and any softening in book-to-bill.

Margins that don't hold (the load-bearing doubt). the case for owning it is that 29% operating margin is the new normal, yet the trailing-year figure is still only about 23%, and one strong quarter is not a trend. If mix reverts and margin settles in the low 20s, earnings never grow into the multiple and the stock is simply expensive. The print: operating margin over the next two quarters.

Structural re-rating (right tail). If margin recovery sticks near 30% on record and rising revenue, trailing profit can reach the old $280M peak on a bigger base, and today's 60x quietly becomes roughly 35x on forward earnings while revenue compounds. The market is pricing a cyclical bounce, not a structurally more profitable company. Tell: gross margin sustaining 71%-plus with operating expenses flat.

Closing thoughts

The recovery is largely in the price. Operating margin over the next two quarters decides whether this is sustainable earnings power or a cyclical fade: if margin holds near 29%, today's multiple compresses naturally on rising earnings; if it slides back toward 22%, the stock is expensive for what it earns. The left tail is the fatter one, because machine vision has rolled over before and 60 times earnings leaves no cushion if orders soften; what is at risk is a multiple that halves on falling earnings. The right tail is real but needs two things at once, sustained margin and sustained demand, and the market has seen enough false dawns here to wait for proof.

The bet is still that factories, warehouses and chip plants keep buying more of Cognex's cameras, and that the sharp pickup of the last year is a real recovery in factory spending, not restocking. What breaks it is operating margin sliding back toward 22% while revenue growth cools; the pair to watch is quarterly revenue and operating margin, together, and if margin gives way first, the recovery story is the one that was wrong.