Cognex Corporation will pay approximately $500 million in cash to acquire RealSense, the depth-sensing camera specialist spun out of Intel in 2025, in a deal that folds 3D robotic perception into Cognex's industrial machine vision portfolio. The transaction, expected to close in the fourth quarter of 2026 subject to customary conditions, also commits Cognex to a three-year $56.5 million cash retention program for RealSense staff and restricted stock units valued at roughly $50 million under Cognex's 2023 Stock Option and Incentive Plan. Evercore advised Cognex on the deal.

What does RealSense bring to a 2D/3D measurement shop?

RealSense was founded by Intel in 2014 and operates a 3D depth-sensing platform that the company positions as the "Visual Cortex of Physical AI." Its cameras serve fixed-arm perception-guided robotics, autonomous mobile robots, quadrupeds, and humanoid platforms — applications where 2D machine vision alone cannot resolve the geometry a manipulator or mobile base needs to grasp, dock, or step.

"RealSense was built on a simple idea: robots and autonomous systems can't act intelligently in the physical world if they can't first perceive it accurately," said Nadav Orbach, Chief Executive Officer of RealSense.

Cognex has historically anchored its revenue in industrial identification and 2D/3D machine vision measurement for factory inspection. Adding stereoscopic and structured-light depth perception extends that measurement stack from "does the part look right on the line?" to "can a robot safely enter this cell?"

How large is the addressable market?

Cognex estimates the robotic perception segment at roughly $600 million today, growing more than 25% annually to approximately $1.6 billion by 2030. RealSense is projected to generate $80 million to $90 million in 2026 revenue — a 50% jump over the prior year.

That revenue mix matters for instrument engineers. A 25%-plus CAGR in perception is well above the single-digit growth typical of mature machine vision inspection, and it puts Cognex on a steeper adoption curve than its legacy bar-code-reading base.

What does the integration look like financially?

The full $500 million purchase price is funded from Cognex's existing cash and short-term investments. The $56.5 million retention pool carries performance modifiers, and the $50 million RSU grant depends on the grant-date share price. Cognex has not disclosed a target closing-quarter revenue or margin for the combined unit, leaving the integration model — and the bridge from RealSense's growth rate to Cognex's through-cycle margin framework — as a forward question for analysts.

Which product line is leaving, and why?

Before closing, RealSense will spin its Facial Authentication product line into an independent company. Cognex framed the move as a way to keep that biometrics business on its existing growth track, but it also removes a regulated-product exposure from the combined entity — relevant for buyers in healthcare, financial services, and government procurement channels that must weigh biometric-capture vendors against their own data-handling obligations.

What does the deal change for integrators?

Matt Moschner, President and Chief Executive Officer of Cognex, tied the strategic case directly to the integrator's bill of materials: "Combined with Cognex's machine vision expertise and global reach, RealSense positions us to offer a full-stack visual intelligence platform, spanning industrial ID, 2D and 3D machine vision measurement to 3D depth perception and robotic navigation, built to serve the full spectrum of Physical AI."

The practical question for systems integrators is convergence at the API and SDK layer: whether depth-camera frames will feed Cognex's existing tooling, and whether 2D inspection libraries can consume depth maps without a separate runtime. Cognex has not yet published a roadmap. Until that integration plan is concrete, the $500 million bet is a market-expansion claim, not a tested platform — and integrators evaluating Physical AI cells will want to see latency, synchronization, and calibration behavior on a single bench before signing off on a multi-camera bill of materials.