Summary

  • Cognex’s pending US$500 million cash acquisition of RealSense is paired with an issuer forecast of US$80–90 million in 2026 revenue. The forecast’s treatment of a product line scheduled to spin out before closing is not disclosed.
  • The resulting 5.6–6.25-times purchase-price-to-sales screen is conditional, not a valuation. Cognex’s cash balance shows it can finance the announced price; it does not establish what the acquired business will earn or convert to cash.

The close date changes; the perimeter changes first

A transaction can be pending and still have a defined price. Its operating perimeter can be less settled. On September 22, Cognex said it had signed a definitive agreement to buy RealSense for US$500 million in cash, subject to customary adjustments. The deal is expected to close in the fourth quarter of 2026, subject to customary conditions. In the same announcement, Cognex said RealSense’s Facial Authentication product line would be spun out into an independent company before closing.

That timing matters. The announcement also projected RealSense revenue of US$80–90 million for 2026, more than 50% above the prior year. It does not say whether the forecast includes Facial Authentication, whether that line’s revenue has already been removed, or how much revenue will transfer to the new company. There is no published post-spinout revenue baseline. The price is visible; the denominator attached to the business Cognex expects to own is not.

Dividing the announced price by the two ends of the forecast gives a rough range of 5.6 to 6.25 times revenue. That is useful as a screening calculation, provided it is described narrowly. It is not an enterprise-value-to-sales ratio: the filing gives no target debt or cash position from which to derive enterprise value. Nor is it a fair-value conclusion. Most importantly, the numerator and denominator may refer to different perimeters. If the sales range includes a line Cognex will not acquire, the calculation overstates revenue available to the buyer and understates the price relative to that revenue.

If the range already excludes it, the calculation means something different. The announcement does not settle which case applies.

The product is concrete; the commercial bridge is not

RealSense sells depth-sensing cameras and related vision technology used to help machines estimate distance and perceive three-dimensional surroundings. Its product pages describe camera families for robotics and industrial applications. The D455 combines stereo depth and RGB imaging, while the D457 adds GMSL/FAKRA connectivity aimed at robotic and warehouse platforms. RealSense says that the D457 requires a platform-specific deserializer and drivers. That is a useful reminder that a camera’s technical capability is only the first step toward a production sale: integration, validation, service and customer acceptance also matter.

Cognex says the acquisition will add a 3D robotic-perception platform to its existing machine-vision business and create a broader offer spanning industrial identification, measurement, depth perception and robotic navigation. That is a plausible product adjacency, not a demonstrated cross-sell. The announcement does not disclose overlapping customers, how many RealSense cameras are in production deployments, the share that could be sold through Cognex’s channels, or the cost of supporting multiple host platforms. It also does not give target gross margin, operating profit or cash flow.

Cognex’s own current margin cannot be used as a substitute: a buyer’s established product mix is not evidence of the target’s margin structure.

A market estimate is not the acquired company’s sales ledger

Cognex estimates the robotic-perception market at US$600 million today and approximately US$1.6 billion by 2030, with annual growth above 25%. Those figures explain the strategic case the buyer is making. They do not establish RealSense’s share, the revenue it can capture, or the amount customers will pay for cameras, software and integration over the life of a robot fleet.

The distinction is especially important when a market category is defined around a function. Depth perception can be one component in a robot’s bill of materials, alongside the robot, controller, safety system, navigation stack, software and systems integration. A company may address a large market without booking an equivalent portion of the finished system’s value. Conversely, a modest camera vendor can become strategically important if its hardware is embedded, supported and difficult to replace. The disclosures contain no data that quantify either possibility for RealSense.

There is a further measurement gap: the Facial Authentication spinout could change the mix of products and customers in the revenue projection before the transaction closes. Biometrics and industrial perception may share imaging expertise without sharing the same buyer, procurement cycle, margin, regulatory burden or support model. Without a separate revenue and cost bridge, the outside reader cannot tell whether the projected growth belongs to the business being purchased, the business leaving, or both.

Cognex can fund the price; funding is not a return

The buyer’s balance sheet provides context. Cognex reported US$755 million of cash and investments and no debt as of July 5, 2026. The US$500 million announced purchase price is about 66% of that reported balance. It is a scale comparison, not a closing-date cash reconciliation: the figure predates the September agreement and does not show subsequent cash movements. Cognex’s second-quarter revenue was US$291 million, up 17% year over year; the company also reported US$68 million of quarterly free cash flow. Its August full-year guidance called for US$1.13–1.15 billion of revenue, before any disclosed contribution from RealSense.

These figures make the transaction financeable on its announced terms. They do not say whether paying cash is the best use of that capital, how much integration expenditure will follow, or what incremental profit the acquisition can generate. The buyer’s ability to fund a transaction and the acquired asset’s ability to earn a return are separate propositions. Treating one as proof of the other would convert a liquidity fact into a performance claim.

People incentives are another ledger

The SEC filing describes a separate retention package for certain continuing RealSense employees: up to US$69 million of cash payments over three years and restricted stock units with an economic value of US$45–55 million, depending on Cognex’s share price on the grant date. The cash amount includes up to US$25 million subject to performance modifiers. The stock units are expected to vest 20%, 30% and 50% on the first, second and third anniversaries. Earning either form generally requires continued employment, with specified acceleration protections for certain key employees.

Cognex’s press release describes US$56.5 million of cash retention at target and approximately US$50 million of RSUs. Target and maximum are different measures, and neither should be quietly added to the US$500 million cash purchase price. The RSUs are not purchase cash; the payments are spread over time and depend on conditions. They are, however, economically relevant. In a technology acquisition, employee continuity can determine whether the buyer receives not just cameras and intellectual property, but the engineering knowledge and customer relationships needed to maintain them.

The filings do not state how many employees are covered or how the retention package maps to specific products.

The useful next disclosure is therefore not another estimate of the total addressable market. It is a bridge: the product line leaving; revenue and costs moving with it; the sales base Cognex acquires; and, after closing, the acquired business’s shipments, gross margin, operating costs and cash generation. Evidence of Cognex channel sales or RealSense products entering repeatable production programs would help test the strategic rationale. Without that bridge, a 2026 revenue forecast remains a forecast for a perimeter the buyer has not yet described precisely.

Evidence register

  • Cognex’s September 22 acquisition announcement supplies the US$500 million price, US$80–90 million revenue forecast, market estimates, pending close and planned Facial Authentication spinout.
  • The SEC Form 8-K states the cash funding source, conditions, and maximum cash and stock retention terms.
  • Cognex’s second-quarter 2026 results support the buyer-scale comparison: cash and investments, debt, quarterly revenue and cash flow as of July 5.
  • RealSense’s D455 materials describe stereo depth-camera functions; its D457 materials describe the GMSL/FAKRA interface and platform integration requirements. These product pages do not establish sales or margins.

The transaction remains pending at the research cutoff. The evidence supports a conditional sales multiple and a clear financing capacity; it does not yet support a matched-perimeter valuation or a conclusion about returns.