Combined average of Moat (AI Resilience), Growth, and Valuation scores.
Moat Score
Cognex is the largest Western pure-play in machine vision and image-based barcode reading, with roughly mid-teens share of a ~$7B served market: its readers and vision systems are written into sortation lines, electronics assembly and packaging inspection, but the lock lasts one line build, and Keyence's direct-sales model and Chinese vision vendors contest every new one.
Cognex's moat is a DataMan reader or In-Sight camera already qualified into a running line, backed by vendor-owned pattern-matching and decode algorithms, but it is a per-line lock in a fragmented market, not a franchise that holds share by itself:
- The Qualified Line: Once a vision job is tuned and validated on an automotive, packaging or electronics line, swapping the camera means re-teaching the inspection and re-validating yield, so Cognex keeps the installed line and its replacement and upgrade business. The weakness is that its biggest end markets re-tender often: consumer-electronics lines are rebuilt around each product cycle and e-commerce distribution centres are bid project by project, which is why Cognex's revenue is far lumpier than Keyence's.
- Vendor-Owned Vision and Decode IP: PatMax geometric pattern matching, the image-based decode engines in DataMan, and the deep-learning tools now folded into the OneVision cloud platform are Cognex's own software, funded by R&D of about 14% of revenue (~$139M in 2025). That is real IP, but deep-learning vision is also the layer where general-purpose models, cheap smart cameras and Chinese vendors such as Hikrobot are closing the gap fastest.
- Not Keyence's Margin, Not Its Channel: Cognex sells through a mix of direct key accounts, distributors and integrators to roughly 30,000 customers, and its 2024 China price cuts showed the share it holds is priced, not owned. The 2025–26 reset (one sales organisation, about 9,000 new accounts added in 2025, $35–40M of net cost savings) lifted adjusted EBITDA margin from ~21% to a 32% quarter, but Keyence still runs operating margins above 50% on a direct model Cognex has not copied.
Moat Verdict
Cognex is a net AI beneficiary on demand, because deep-learning inspection widens what vision can do on a line and OneVision is built for it, but AI is also what lowers the barrier for rivals. Its AI-resilient pillars, line embedding and process validation, are real but intact rather than category-defining, while its AI-vulnerable vision and decode IP is where general-purpose models and Chinese smart cameras press hardest. It is a durable, good franchise in a fragmented market, not a fortress.
65.0 resilient · 65.0 vulnerable · 80/20 = 65.0 · = 65
Open a moat to read its note.
Integrators and line engineers trained on In-Sight Explorer, VisionPro and the DataMan setup tool tend to specify Cognex again, but no-code deep-learning setup lowers the cost of learning a rival's tool.
PatMax pattern matching, image-based decode engines and the OneVision deep-learning stack are vendor-owned IP, but general-purpose vision models and Chinese smart cameras are closing the gap; intact, the same rating as Keyence.
Cognex sells vision hardware and software; it controls no public data source.
Machine-vision and imaging engineers are scarce, but Keyence, Zebra and Chinese vendors hire from the same pool (M8).
Readers, vision systems, lighting and the OneVision software are sold together, but no attach or renewal figure shows the bundle holds customers better than its parts (M9).
Decades of inspection and decode images train Cognex's deep-learning tools, but the line images belong to customers and the dataset is not a class no rival holds (M5).
Vision jobs validated into pharma, food and automotive PPAP processes require re-qualification to replace; Keyence, SICK and others are validated into the same processes, so it bars entrants, not peers (M4), the same rating as Keyence.
The integrator and distributor bench is channel density, not a network effect; one customer's line does not make Cognex more valuable to another.
Readers and vision systems are embedded in a line for its life, the same fact Keyence rates intact; Cognex's largest end markets (consumer-electronics assembly and e-commerce sortation) also re-tender at each product cycle or distribution-centre build, and its 2024 China price cuts show the installed base did not protect price, so it rates intact.
Vision job files and inspection results feed the customer's MES; no downstream system defers to Cognex as the record, the same call as Keyence.
With roughly mid-teens share of a fragmented market and Chinese rivals selling cheaper hardware, Cognex has no structural unit-cost lead to rate.
Buyers are enterprises choosing on switching cost, integration and performance, which the other pillars rate. The name carries reputation, not a price premium it could hold on brand alone.
Moat in 5–10 years
Cognex's moats hold at intact. Line embedding and process validation stay for each line's life, and the 2026 margin recovery is cost-out plus volume, not new pricing power. The pillar to watch is business logic: if Chinese vision vendors and general-purpose models win logistics or electronics lines on price in 2027, or Cognex has to cut China prices again as it did in 2024, that would be the signal it is narrowing; nothing on the slate has moved yet.
No moat trending · 65 holds
Combined average of Moat (AI Resilience), Growth, and Valuation scores.
Moat Score
Cognex is the largest Western pure-play in machine vision and image-based barcode reading, with roughly mid-teens share of a ~$7B served market: its readers and vision systems are written into sortation lines, electronics assembly and packaging inspection, but the lock lasts one line build, and Keyence's direct-sales model and Chinese vision vendors contest every new one.
Growth Score
FY2025 revenue was $994M, up 9% (7% excluding a one-time commercial-partnership benefit), with adjusted EBITDA margin of 21.5% and free cash flow of $237M. 2026 is a recovery year: Q1 revenue rose 24% to ~$268M and Q2 a record $291.3M (+17%), with adjusted EBITDA margin up to 32.2% and adjusted EPS $0.45 (vs $0.25). Logistics has posted nine straight quarters of double-digit growth, semiconductor grew over 20% in Q1, consumer electronics is growing and automotive has bottomed. Full-year guidance is $1.13–1.15B of revenue (+15%, +16% ex the one-time benefit), 29–31% adjusted EBITDA margin and $1.64–1.68 adjusted EPS; Q3 is guided to $300–320M, reported November 4. The 3–5 year rate has to be measured off the cycle, not 2026: revenue in 2025 was still below the 2021 peak.
Valuation Score
At the $53.99 close on October 16, 2026, CGNX sits about 61% of the way from the bear case ($40) to base ($63), a 75 on the ladder, after falling from a $72.88 52-week high. The base is about 30× a ~$2.10 FY2027 adjusted EPS, close to Keyence's ~28–32× despite a weaker moat and far lower margin, because Cognex's earnings are still mid-recovery (2026 adjusted EPS guided up about 60%) and Cognex has historically traded above Keyence on forward earnings; the base sits about 17% below the Street mean, which prices a full margin recovery that has not yet been through a downturn.
Designed Into the Line, Re-Won at the Next One
Cognex's moat is a DataMan reader or In-Sight camera already qualified into a running line, backed by vendor-owned pattern-matching and decode algorithms, but it is a per-line lock in a fragmented market, not a franchise that holds share by itself:
- The Qualified Line: Once a vision job is tuned and validated on an automotive, packaging or electronics line, swapping the camera means re-teaching the inspection and re-validating yield, so Cognex keeps the installed line and its replacement and upgrade business. The weakness is that its biggest end markets re-tender often: consumer-electronics lines are rebuilt around each product cycle and e-commerce distribution centres are bid project by project, which is why Cognex's revenue is far lumpier than Keyence's.
- Vendor-Owned Vision and Decode IP: PatMax geometric pattern matching, the image-based decode engines in DataMan, and the deep-learning tools now folded into the OneVision cloud platform are Cognex's own software, funded by R&D of about 14% of revenue (~$139M in 2025). That is real IP, but deep-learning vision is also the layer where general-purpose models, cheap smart cameras and Chinese vendors such as Hikrobot are closing the gap fastest.
- Not Keyence's Margin, Not Its Channel: Cognex sells through a mix of direct key accounts, distributors and integrators to roughly 30,000 customers, and its 2024 China price cuts showed the share it holds is priced, not owned. The 2025–26 reset (one sales organisation, about 9,000 new accounts added in 2025, $35–40M of net cost savings) lifted adjusted EBITDA margin from ~21% to a 32% quarter, but Keyence still runs operating margins above 50% on a direct model Cognex has not copied.
Moat Verdict
Cognex is a net AI beneficiary on demand, because deep-learning inspection widens what vision can do on a line and OneVision is built for it, but AI is also what lowers the barrier for rivals. Its AI-resilient pillars, line embedding and process validation, are real but intact rather than category-defining, while its AI-vulnerable vision and decode IP is where general-purpose models and Chinese smart cameras press hardest. It is a durable, good franchise in a fragmented market, not a fortress.
65.0 resilient · 65.0 vulnerable · 80/20 = 65.0 · = 65
Open a moat to read its note.
Integrators and line engineers trained on In-Sight Explorer, VisionPro and the DataMan setup tool tend to specify Cognex again, but no-code deep-learning setup lowers the cost of learning a rival's tool.
PatMax pattern matching, image-based decode engines and the OneVision deep-learning stack are vendor-owned IP, but general-purpose vision models and Chinese smart cameras are closing the gap; intact, the same rating as Keyence.
Cognex sells vision hardware and software; it controls no public data source.
Machine-vision and imaging engineers are scarce, but Keyence, Zebra and Chinese vendors hire from the same pool (M8).
Readers, vision systems, lighting and the OneVision software are sold together, but no attach or renewal figure shows the bundle holds customers better than its parts (M9).
Decades of inspection and decode images train Cognex's deep-learning tools, but the line images belong to customers and the dataset is not a class no rival holds (M5).
Vision jobs validated into pharma, food and automotive PPAP processes require re-qualification to replace; Keyence, SICK and others are validated into the same processes, so it bars entrants, not peers (M4), the same rating as Keyence.
The integrator and distributor bench is channel density, not a network effect; one customer's line does not make Cognex more valuable to another.
Readers and vision systems are embedded in a line for its life, the same fact Keyence rates intact; Cognex's largest end markets (consumer-electronics assembly and e-commerce sortation) also re-tender at each product cycle or distribution-centre build, and its 2024 China price cuts show the installed base did not protect price, so it rates intact.
Vision job files and inspection results feed the customer's MES; no downstream system defers to Cognex as the record, the same call as Keyence.
With roughly mid-teens share of a fragmented market and Chinese rivals selling cheaper hardware, Cognex has no structural unit-cost lead to rate.
Buyers are enterprises choosing on switching cost, integration and performance, which the other pillars rate. The name carries reputation, not a price premium it could hold on brand alone.
Moat in 5–10 years
Cognex's moats hold at intact. Line embedding and process validation stay for each line's life, and the 2026 margin recovery is cost-out plus volume, not new pricing power. The pillar to watch is business logic: if Chinese vision vendors and general-purpose models win logistics or electronics lines on price in 2027, or Cognex has to cut China prices again as it did in 2024, that would be the signal it is narrowing; nothing on the slate has moved yet.
No moat trending · 65 holds
Growth Analysis
Growth Drivers
Key Risk
Logistics and consumer-electronics capex digests in 2027 after the 2026 rebound while Chinese vision vendors keep pressing on price, so organic revenue growth falls back below 5% and adjusted EBITDA margin slips under the 25% exit-rate floor management set for 2026.
Score Derivation
71.4 base + 1.3 trajectory + 4 margin − 5 risk = 72
Base 71.4 (9% midpoint of 7–11%) + 1.3 trajectory (semiconductor accelerating; logistics held stable because management guided it to moderate from double digits; consumer electronics stable) + 4 expanding margin (adj. EBITDA 21.5% in 2025 to 29–31% guided for 2026) − 5 moderate risk (factory-automation and consumer-electronics capex cycle plus China pricing, the same moderate step as Keyence) = 72
Price Scenarios (12–24 Months)
Price vs Targets
Dashed lines mark the 12–24 month bear, base and bull targets. Daily closes.
Valuation Multiples
| Trailing P/E (GAAP) | ~52× |
| Forward P/E (FY26) | ~33× |
| Forward P/E (NTM) | ~27× |
| PEG Ratio | ~1.8× |
| Price / Sales (FY26) | ~7.9× |
| Price / FCF | ~38× |
On NTM earnings Cognex is at a premium to the S&P 500 but at the low end of its own history and slightly below Keyence's ~28–32×, and the drop from 52× trailing to 27× NTM is the margin recovery showing through rather than a cheap multiple. A PEG near 1.8× says the price already credits mid-teens EPS growth, so the margin of safety rests on the 30%+ EBITDA quarter holding through the next automation downturn.
Approximate figures as of October 18, 2026.
Where We Are vs Targets
Loading live price…
The 2026 rebound proves to be a capex spike: logistics and consumer-electronics customers digest, China pricing pressure returns, and FY2027 adjusted EPS stalls near $1.75 on a ~23× multiple.
- Q3 or Q4 2026 revenue lands at or below the low end of guidance ($300M for Q3) as e-commerce distribution-centre projects slip
- Adjusted EBITDA margin falls back below the 25% exit-rate floor in 2027 as volume deleverage outruns the $35–40M cost savings
- Hikrobot and other Chinese vision vendors force another round of price cuts in Greater China, roughly 16% of revenue
Revenue grows high single to low double digits off the 2026 recovery, adjusted EBITDA margin holds near 30%, and FY2027 adjusted EPS of ~$2.10 earns about 30×.
- FY2026 revenue lands inside the $1.13–1.15B guide and adjusted EPS inside $1.64–1.68
- Adjusted EBITDA margin stays at or above 28% through 2027 as opex grows slower than revenue
- Buybacks under the $500M authorisation keep the share count flat to down despite stock compensation
Semiconductor and logistics strength compounds, OneVision and the new AI products widen the customer base toward management's doubling goal, and FY2027 adjusted EPS of ~$2.40 earns about 36×, back toward the 52-week high and Street-high targets.
- Revenue growth holds above 12% into 2027 as semiconductor and packaging join logistics in double digits
- Adjusted EBITDA margin moves into the upper half of the raised 25–31% through-cycle band on a sustained basis
- Customer count rises well past the ~30,000 Investor Day baseline as the new-account rate of 2025 (~9,000) repeats
CGNX: common questions
Is Cognex stock a buy?
We rate Cognex Hold, with a composite score of 71/100: moat 65, growth 72 and valuation 75. The valuation score is struck against our bear, base and bull targets and moves with the price.
What is Cognex's moat?
Cognex scores 65/100 on moat. Cognex is the largest Western pure-play in machine vision and image-based barcode reading, with roughly mid-teens share of a ~$7B served market: its readers and vision systems are written into sortation lines, electronics assembly and packaging inspection, but the lock lasts…
How fast is Cognex expected to grow?
We estimate 7-11% a year over the next three to five years, for a growth score of 72/100.
What is Cognex's price target?
Our base case is $63, with $40 in the bear case and $86 in the bull case.