What are the top edge AI startups by revenue today?

In our edge AI market deck, you will find everything you need to understand the market
SUMMARY
Anduril is the biggest private edge AI startup by current disclosed revenue, at $2.2 billion in 2025, while Symbotic is slightly larger if we include public companies that grew out of the startup ecosystem.
The top of edge AI is already much bigger than the phrase “startup market” suggests. Symbotic, Anduril, Mobileye and Samsara each operate around or above the $2 billion level on revenue or ARR, but they are not all equally clean edge AI comparisons.
The metric problem becomes important almost immediately. Verkada has surpassed $1 billion in annualized bookings, Samsara reports $2.125 billion of ARR, and several robotics companies are known mainly through estimates, so a simple league table would overstate how comparable these numbers really are.
Warehouse automation, automotive AI and defense autonomy are the clearest large revenue pools. These are also categories where customers can tie local AI directly to throughput, vehicle functionality, safety, labor or mission outcomes.
The private-company curve is steep. After Anduril and Verkada, Shield AI is around $300 million, Saronic around $200 million on an estimate, Locus Robotics around $180 million of estimated ARR, and Skild AI above $100 million of ARR.
Skild AI is the unusual software-heavy outlier. It crossed $100 million of ARR within ten months of its first commercial deployment and says $50 million had already been recognized, while most pure edge AI software companies remain much smaller than hardware-led systems businesses.
Humanoid robotics still looks far richer in funding and order announcements than in recognized revenue. Agility Robotics reported only about $1.8 million of 2025 net sales despite an order book above $300 million, while Figure AI and Apptronik still do not disclose company-wide revenue.
Automotive has the deepest bench of measurable edge AI businesses. Mobileye, ECARX, Horizon Robotics, Black Sesame, Cerence AI, indie Semiconductor, WeRide and Pony.ai already provide a broad ladder from roughly $90 million to almost $2 billion of annual revenue.
The $10 million to $100 million range is crowded. BrightAI, Rekor, Blaize, Standard Bots, Dexterity, Quadric, Gather AI and DEEPX show that commercial adoption is no longer rare; the harder jump is building repeatable distribution into nine-figure scale.
The practical takeaway is that edge AI revenue is concentrated around complete operational products rather than generic intelligence alone. Robots, vehicles, cameras, industrial vision systems and defense platforms still capture most of the largest revenue pools, even as software-first companies such as Skild begin to challenge that pattern.

This market map, featured in our edge AI market deck, highlights top companies and startups in the edge AI market
The ranking of top startups in the edge AI market by revenue
Below is a table ranking all the companies in this market by their current revenue scale. You can find our methodology at the end of this page.
If you want a deeper understanding of the market and its current dynamics, get our report covering the Edge AI Market.
| Ranking | Company | Latest Metric | Metric Type | Freshness | Disclosed When | Source Quality | Confidence | Segment | Why This Ranking |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Symbotic | $2.25B | Fiscal-Year Revenue | Fresh · 12mo | Nov 24, 2025 | Filed / Audited | High | Industrial / Warehouse Robotics | Largest recent recognized revenue found among edge-AI-focused companies; narrowly exceeds Anduril with stronger audited evidence. |
| 2 | Anduril Industries | $2.2B | Annual Revenue | Fresh · 9mo | May 13, 2026 | Company Disclosed | Medium | Defense Edge AI & Autonomy | Similar scale to Symbotic but privately disclosed rather than audited; comfortably above Mobileye. |
| 3 | Mobileye | $1.89B | Fiscal-Year Revenue | Fresh · 9mo | Jan 22, 2026 | Filed / Audited | High | Automotive Edge AI / ADAS | Large audited revenue with a particularly clean onboard-inference perimeter; below Anduril on amount and above Samsara. |
| 4 | Samsara | $1.62B | Fiscal-Year Revenue | Fresh · 8mo | Mar 2026 | Filed / Audited | Medium | Connected Operations / Edge Video AI | Very large filed revenue, but total company revenue includes telematics and cloud workflows beyond edge AI, reducing comparability. |
| 5 | Cognex | $994M | Annual Revenue | Fresh · 9mo | Feb 11, 2026 | Filed / Audited | Medium | Industrial Machine Vision | Near-$1B recognized revenue; outranks Verkada because Verkada's >$1B figure is bookings rather than recognized revenue. |
| 6 | Verkada | >$1B | Bookings | Fresh · 9mo | Dec 3, 2025 | Company Disclosed | Medium | Edge Video Security | Scale signal is larger than companies below, but it receives a substantial penalty versus Cognex because bookings are not revenue. |
| 7 | ECARX | $848M | Annual Revenue | Fresh · 9mo | Apr 2, 2026 | Filed / Audited | High | Automotive Edge Compute | Strong recent audited revenue; below Verkada's much larger commercial signal despite metric advantage, above Horizon on amount. |
| 8 | Horizon Robotics | Rmb3.76B | Annual Revenue | Fresh · 9mo | Mar 2026 | Filed / Audited | High | Automotive Edge AI Chips & Software | Large audited revenue, predominantly automotive solutions; below ECARX and above Geek+ on current scale. |
| 9 | Geek+ | Rmb3.17B | Annual Revenue | Fresh · 9mo | Mar 30, 2026 | Filed / Audited | High | Warehouse Robotics | 99.9% of revenue comes from robotics solutions, providing an unusually clean edge/physical-AI perimeter. |
| 10 | Ambarella | $391M | Fiscal-Year Revenue | Fresh · 8mo | Feb 26, 2026 | Filed / Audited | High | Edge AI Semiconductors | Current, audited and highly relevant inference-chip revenue; below much larger robotics vendors and above Shield AI. |
| 11 | Shield AI | ≈$300M | Annual Revenue | Fresh · 9mo | Dec 21, 2025 | Company Disclosed | Medium | Defense Edge AI & Autonomy | Strong direct revenue disclosure, but private and unaudited; sits below Ambarella and above UBTECH. |
| 12 | UBTECH Robotics | Rmb2B | Annual Revenue | Fresh · 9mo | Mar 31, 2026 | Filed / Audited | Medium | Humanoid / Service Robotics | Large audited revenue, though only part comes from humanoids; perimeter penalty keeps it below Shield AI. |
| 13 | Cerence AI | $252M | Fiscal-Year Revenue | Fresh · 9mo | Nov 19, 2025 | Filed / Audited | Medium | Automotive On-device Voice AI | Strong recognized revenue and a close edge-inference fit; below UBTECH's larger total and above indie Semiconductor. |
| 14 | indie Semiconductor | $217M | Annual Revenue | Fresh · 9mo | Mar 2026 | Filed / Audited | Medium | Automotive Edge Semiconductors | Filed revenue beats Saronic's similarly sized but estimated figure on source quality. |
| 15 | Saronic | $200M | Annual Revenue Estimate | Fresh · 9mo | Mar 6, 2026 | Third-Party Estimate | Low | Autonomous Naval Systems | Substantial estimated revenue, but weaker than indie's filed figure; still ahead of Locus on estimated commercial scale. |
| 16 | Locus Robotics | $180M | ARR | Very Fresh · 3mo | Jul 6, 2026 | Third-Party Estimate | Low | Warehouse Robotics | Very current scale but ARR and third-party estimated; above Evolv because the gap remains meaningful. |
| 17 | Evolv Technology | $146M | Annual Revenue | Fresh · 9mo | Mar 10, 2026 | Filed / Audited | High | On-site AI Security Screening | Clean recognized revenue evidence; below Locus's materially larger but weaker ARR signal, above Lantronix. |
| 18 | Lantronix | $121M | Fiscal-Year Revenue | Very Fresh · 3mo | Aug 26, 2026 | Filed / Audited | Medium | Edge AI / IoT Infrastructure | Extremely fresh audited scale, although only part of the business is edge AI; edges Black Sesame on approximate revenue. |
| 19 | Black Sesame Technologies | Rmb822M | Annual Revenue | Fresh · 9mo | Mar 31, 2026 | Filed / Audited | High | Automotive / Embodied Edge AI Chips | Most revenue comes from driving-assistance products; slightly below Lantronix in dollar scale and above CEVA. |
| 20 | CEVA | $110M | Annual Revenue | Fresh · 9mo | Feb 2026 | Filed / Audited | Medium | Edge AI & DSP IP | 86% of revenue is tied to smart-edge markets; strong direct evidence, but below Black Sesame on size. |
| 21 | Skild AI | >$100M | ARR | Very Fresh · 0mo | Sep 10, 2026 | Company Disclosed | Medium | Robotics AI Software / Physical AI | Exceptionally fresh ARR plus meaningful recognized revenue; placed just below CEVA's full-year revenue and above Gorilla due freshness and cleaner AI perimeter. |
| 22 | Gorilla Technology | $101M | Annual Revenue | Fresh · 9mo | Mar 2, 2026 | Filed / Audited | Medium | Edge Vision / AI Infrastructure | Recognized revenue is strong, but the company perimeter is less purely edge AI than Skild; above WeRide on amount. |
| 23 | WeRide | $97.9M | Annual Revenue | Fresh · 9mo | Mar 23, 2026 | Filed / Audited | High | Autonomous Mobility | Current recognized autonomous-driving revenue; slightly below Gorilla and above Helsing's incomplete corporate perimeter. |
| 24 | Helsing | £70.3M | Annual Revenue / Turnover | Very Fresh · 1mo | Aug 12, 2026 | Filed / Audited | Low | Defense Edge AI | Strong filed turnover but an incomplete group perimeter; positioned between WeRide and Pony with explicit caution. |
| 25 | Pony.ai | $90M | Annual Revenue | Fresh · 9mo | Mar 26, 2026 | Company Disclosed | Medium | Autonomous Mobility | Recent recognized revenue with direct AV fit; below Helsing's larger filed UK turnover and above BrightAI's older figure. |
| 26 | BrightAI | $80M | Revenue — period not specified | Aging · 22mo | Nov 19, 2024 | Company Disclosed | Medium | Industrial Edge AI / Infrastructure | Large direct figure retained despite age; penalized below newer revenue but remains above materially smaller companies. |
| 27 | Rekor Systems | $48.5M | Annual Revenue | Fresh · 9mo | Mar 31, 2026 | Filed / Audited | High | AI Roadway Intelligence | Current filed revenue; below BrightAI's much larger older figure and above Blaize. |
| 28 | Blaize | $38.6M | Annual Revenue | Fresh · 9mo | Mar 24, 2026 | Filed / Audited | Medium | Edge Inference Compute | Strong filed number but not a completely pure edge perimeter; above Aeva on recognized revenue. |
| 29 | Aeva | $18.1M | Annual Revenue | Fresh · 9mo | Feb 26, 2026 | Filed / Audited | Medium | Sensor Perception / AV | Lower nominal revenue than Standard Bots' estimate, but higher evidence quality and recognized revenue justify the higher rank. |
| 30 | Standard Bots | $24M | Revenue Run-Rate | Fresh · 9mo | 2026 | Third-Party Estimate | Low | Industrial Robotics | Larger nominal scale than Aeva but estimated/annualized rather than filed; above Quadric on estimated size. |
| 31 | Quadric | $15M–$20M | Licensing Revenue | Fresh · 9mo | Jan 22, 2026 | Company Disclosed | Medium | Edge AI Processor IP | Directly disclosed, well-defined licensing revenue outranks Dexterity's higher but modeled estimate. |
| 32 | Dexterity | $21.2M | Annual Revenue Estimate | Fresh · 9mo | Nov 25, 2025 | Third-Party Estimate | Low | Warehouse Robotics | Nominally above Quadric but ranked below it because the figure is explicitly estimated; above Gather on estimated scale. |
| 33 | Gather AI | $15M | Annual Revenue | Very Fresh · 3mo | Jun 17, 2026 | Company Disclosed | Medium | Warehouse Drone AI | Direct recent disclosure outranks MemryX's larger third-party estimate. |
| 34 | MemryX | $17.5M | Annual Revenue Estimate | Fresh · 9mo | Sep 2026 | Third-Party Estimate | Low | Edge AI Semiconductors | Larger nominal figure than Gather, but substantially weaker sourcing; above DEEPX's bookings measure. |
| 35 | DEEPX | >$13M | Bookings | Very Fresh · 2mo | 2026 | Credible Reported | Medium | Edge AI Semiconductors | Very fresh commercial evidence but purchase orders are weaker than revenue; above Nota on disclosed scale. |
| 36 | Nota AI | >$7M | Annual Revenue | Fresh · 9mo | Dec 31, 2025 | Company Disclosed | Medium | Edge MLOps / On-device Optimization | Direct annual revenue outranks partial-period revenue below despite smaller companies showing rapid recent growth. |
| 37 | Serve Robotics | $6.22M | Six-Month Revenue | Very Fresh · 3mo | Aug 6, 2026 | Filed / Audited | Medium | Autonomous Delivery Robotics | Very fresh six-month recognized revenue; above Palladyne because it covers a longer period with a cleaner robotics perimeter. |
| 38 | Palladyne AI | $5.8M | Quarterly Revenue | Very Fresh · 3mo | Aug 6, 2026 | Filed / Audited | Medium | Defense / Industrial Physical AI | One-quarter figure signals strong current scale, but it is partial-period and includes activities beyond AI software; above Kodiak on latest-quarter intensity without annualizing. |
| 39 | Kodiak AI | $5.33M | Six-Month Revenue | Very Fresh · 3mo | Aug 7, 2026 | Filed / Audited | Medium | Autonomous Trucking | Strong current six-month evidence; below Palladyne's latest-quarter number and above Richtech's older annual figure. |
| 40 | Richtech Robotics | $5.04M | Fiscal-Year Revenue | Fresh · 9mo | Jan 20, 2026 | Filed / Audited | High | Commercial Service Robotics | Filed annual revenue beats the less reliable estimates and filing-derived private-company figures immediately below. |
| 41 | Aurora Innovation | $3M | Annual Revenue | Fresh · 9mo | Feb 11, 2026 | Filed / Audited | High | Autonomous Trucking | Lower than several estimates below in nominal terms, but recognized SEC-filed revenue receives substantially more weight. |
| 42 | Robovision | €3.43M | Annual Revenue / Turnover | Fresh · 9mo | Jul 28, 2026 | Credible Reported | Medium | Industrial Vision AI | Filing-derived turnover is useful but one step removed from the original NBB filing; sits below Aurora's direct public filing. |
| 43 | Lumana | $6.1M | Annual Revenue Estimate | Fresh · 9mo | 2026 | Third-Party Estimate | Low | Edge Video Security | Nominally larger than Aurora/Robovision but ranked lower because the evidence is a modeled third-party estimate. |
| 44 | Edgify | $4.1M | Annual Revenue Estimate | Fresh · 9mo | 2026 | Third-Party Estimate | Low | Retail On-device AI | Estimated scale below Lumana and above Nexa AI; no direct revenue disclosure found. |
| 45 | Nexa AI | $3.4M | Annual Revenue Estimate | Fresh · 9mo | 2026 | Third-Party Estimate | Low | On-device AI Software | Third-party estimate only; ranked between Edgify and OctaiPipe on the same weak evidence class. |
| 46 | OctaiPipe | $2.6M | Annual Revenue Estimate | Very Fresh · 0mo | 2026 | Third-Party Estimate | Low | Industrial Edge AI / Edge MLOps | Two estimate databases converge around $2.6M, but no company-reported revenue was found. |
| 47 | BrainChip | $1.89M | Annual Revenue | Fresh · 9mo | Feb 26, 2026 | Filed / Audited | High | Neuromorphic Edge AI | Small but clean audited revenue; slightly exceeds Agility's audited 2025 sales. |
| 48 | Agility Robotics | $1.78M | Annual Net Sales | Fresh · 9mo | Sep 4, 2026 | Filed / Audited | High | Humanoid Robotics | Audited revenue is slightly below BrainChip. More than $300M of multi-year Digit orders is deliberately not treated as current revenue. |
| NR | Axelera AI | Revenue Not Disclosed; 600+ Customers | Customers | Very Fresh · 0mo | Sep 15, 2026 | Company Disclosed | Low | Edge AI Semiconductors | NR — extensive commercial adoption, but customer count and reported deal pipeline cannot be converted into revenue. |
| NR | Waymo | Revenue Not Disclosed; 500,000 Paid Rides/Week | Transactions | Very Fresh · 6mo | Mar 27, 2026 | Company Disclosed | Medium | Autonomous Mobility | NR — unusually strong paid-usage evidence, but no actual company revenue disclosure supports a precise position. |
| NR | Brain Corp | Revenue Not Disclosed; 50,000+ Robots Deployed | Units Deployed | Very Fresh · 3mo | Aug 6, 2026 | Company Disclosed | Low | Robotics AI Software | NR — large installed base confirms scale, but deployed robots cannot defensibly be translated into annual company revenue. |
| NR | Spot AI | Revenue Not Disclosed; 1,000+ Customers | Customers | Very Fresh · 0mo | Not Stated | Company Disclosed | Low | Edge Video Security | NR — substantial customer base but no meaningful financial disclosure found. |
| NR | AiFi | Revenue Not Disclosed; 370+ Deployments / 40 Clients | Deployments / Customers | Very Fresh · 5mo | Apr 2026 | Credible Reported | Low | Autonomous Checkout / Retail Edge Vision | NR — deployment count is useful evidence of commercialization but not sufficiently comparable with revenue. |
| NR | Hayden AI | Revenue Not Disclosed; 2,000+ System Installations | Units Deployed | Fresh · 12mo | Sep 16, 2025 | Company Disclosed | Low | Mobile Edge Vision / Transit Enforcement | NR — large deployed footprint and clear edge inference, but installation count does not establish company revenue. |
| NR | Hailo | Revenue Not Disclosed; 300+ Customers | Customers | Aging · 29mo | Apr 2, 2024 | Company Disclosed | Low | Edge AI Semiconductors | NR — major market company, but the strongest public commercial count is aging and cannot support a revenue rank. |
| NR | Edge Impulse | Revenue Not Disclosed; 250,000+ Developers | Users / Developers | Fresh · 9mo | Dec 2025 | Company Disclosed | Low | Edge AI Development Platform | NR — large ecosystem but developer count is substantially weaker than paying-customer or revenue evidence. |
| NR | Bluewhite | Revenue Not Disclosed; 75,000+ Autonomous Operating Hours | Other Scale Signal | Fresh · 18mo | Mar 18, 2025 | Company Disclosed | Low | Agricultural Autonomy | NR — real commercial field usage is clear, but operating hours are not a revenue measure. |
| NR | Figure AI | Revenue Not Disclosed; 30,000+ Bmw Vehicles Supported By 2025 Deployment | Other Scale Signal | Fresh · 9mo | Nov 19, 2025 | Company Disclosed | Low | Humanoid Robotics | NR — Figure is revenue-generating, but no amount is public; production output at one customer cannot be converted into company sales. |
| NR | Jaipur Robotics | Revenue Not Disclosed; >5M Tonnes Of Waste Analyzed Annually | Other Scale Signal | Very Fresh · 0mo | Sep 7, 2026 | Company Disclosed | Low | Industrial Edge Vision | NR — impressive live industrial throughput, but tonnes analyzed are not financially comparable. |
| NR | Luxonis | Revenue Not Disclosed; 25,000+ Customers | Customers | Aging · 21mo | 2024 | Company Disclosed | Low | Edge AI Cameras / Spatial Vision | NR — meaningful customer scale, but no revenue number and the disclosure is now old. |
| NR | Hellbender | Revenue Not Disclosed; ≈$3M Customer Contract | Bookings / Customer Contract Value | Very Fresh · 0mo | Sep 2026 | Filed / Audited | Medium | Edge AI Cameras / Physical AI Hardware | NR — a $3M contract is concrete commercial evidence, but one customer's contract value cannot represent total annual revenue. |
| NR | Rapta | Revenue Not Disclosed; 50+ Shipping Defects Prevented At Named Customer | Other Scale Signal | Very Fresh · 0mo | 2026 | Company Disclosed | Low | Industrial Edge Vision | NR — clear production deployment but no business-wide customers, deployments or revenue figure. |
| NR | Innatera | Revenue Not Disclosed; First Named Odm/Customer Deployments | Other Scale Signal | Fresh · 9mo | Dec 10, 2025 | Company Disclosed | Low | Neuromorphic Edge AI | NR — commercial products now exist, but disclosed customer/deployment evidence is too limited for a revenue position. |
| NR | RealSense | Revenue Not Disclosed; 3,000+ Customers | Customers | Fresh · 14mo | Jul 11, 2025 | Company Disclosed | Low | Edge AI Cameras / 3D Vision | NR — one of the larger commercial footprints in edge vision, but no standalone revenue has been disclosed since the spinout. |
| NR | Apptronik | Revenue Not Disclosed; 6 Verified Commercial Deployments | Deployments | Very Fresh · 0mo | 2026 | Credible Reported | Low | Humanoid Robotics | NR — recent $935M financing establishes investor scale, not revenue; verified customer deployments are the stronger commercial evidence. |

As this chart shows, and as featured in our edge AI market deck, search interest in edge AI has increased sharply
Which edge AI startups make the most revenue today?
The biggest edge AI companies now generate billions of dollars a year, with Symbotic and Anduril sitting at the top of the market on the strongest available revenue figures.
Symbotic reported $2.25 billion of revenue in FY2025. Anduril disclosed $2.2 billion for 2025. Mobileye generated $1.89 billion, while Samsara has since crossed $2.1 billion of ARR as its connected-operations platform keeps growing.
The next group is already much smaller. Cognex generated about $994 million in 2025. Verkada has surpassed $1 billion in annualized bookings, which shows huge commercial scale even though bookings cannot be treated as recognized revenue. ECARX generated $848 million. Horizon Robotics reported RMB3.76 billion, and Geek+ generated RMB3.17 billion.
From there, the curve drops quickly. Ambarella generated $391 million in its latest completed fiscal year. Shield AI disclosed roughly $300 million for 2025. Several other companies cluster around $100 million to $250 million.
That steep drop is hard to miss. Edge AI already has several very large businesses, while most of the market is still far smaller.
| Company | Best recent commercial figure | Metric | Main edge AI market |
|---|---|---|---|
| Symbotic | $2.25B | FY revenue | Warehouse robotics |
| Anduril | $2.2B | Annual revenue | Defense autonomy |
| Mobileye | $1.89B | FY revenue | Automotive AI |
| Samsara | $2.125B | ARR | Connected operations |
| Cognex | $994M | Annual revenue | Machine vision |
| Verkada | >$1B | Annualized bookings | AI physical security |
| ECARX | $848M | Annual revenue | Automotive compute |
| Horizon Robotics | RMB3.76B | Annual revenue | Automotive AI chips |
| Geek+ | RMB3.17B | Annual revenue | Warehouse robotics |
| Ambarella | $391M | FY revenue | Edge AI semiconductors |
What is the biggest private edge AI startup by revenue?
Anduril is the biggest private edge AI startup we can identify from credible current revenue disclosures, with $2.2 billion of revenue in 2025.
The company said revenue had more than doubled over the previous year. That growth is unusual at this scale: going from roughly $1 billion to $2.2 billion in a year puts Anduril well beyond the normal commercial range of venture-backed edge AI companies.
Verkada is the closest private company on a headline commercial metric after surpassing $1 billion in annualized bookings. The metric difference is important because signed business can be recognized as revenue over a later period.
Shield AI is further behind at roughly $300 million of 2025 revenue. Locus Robotics is estimated at around $180 million of ARR, and Skild AI has now crossed $100 million of ARR.
So Anduril's lead is substantial. Its reported annual revenue is more than seven times Shield AI's and roughly twenty times Skild AI's current ARR.

This chart, featured in our edge AI market deck, illustrates yearly venture capital investment in edge AI startups
Is Symbotic actually bigger than Anduril?
Symbotic is still slightly ahead of Anduril on their latest completed annual revenue figures, although the gap is small enough to put both companies in the same commercial tier.
Symbotic generated $2.247 billion in FY2025. Anduril reported $2.2 billion during calendar 2025. The difference is only about $47 million, or roughly 2%.
Symbotic has also kept growing since that annual figure. The company recently reported $721 million of quarterly revenue, up 22% from the same quarter a year earlier. That is already an annualized pace approaching $2.9 billion if someone simply multiplied the quarter by four, although we do not use that calculation as reported annual revenue.
The important distinction is company status. Symbotic is publicly listed, whereas Anduril remains private.
For readers using "startup" in the strict private-company sense, Anduril is the answer. If we include companies that grew out of the startup ecosystem and later listed, Symbotic currently has the larger completed-year revenue figure.
How big is Mobileye's edge AI business now?
Mobileye remains one of the largest edge AI companies in the world, with $1.89 billion of 2025 revenue and current guidance putting annual revenue close to $2 billion again.
The company recently reported $508 million of quarterly revenue. It now expects roughly $1.97 billion to $2.02 billion for the full year.
Mobileye is especially useful in this ranking because the connection with edge AI is unusually direct. Its EyeQ processors and driving software run inside vehicles and handle perception and driving functions where local processing is essential.
That makes Mobileye cleaner to compare with edge inference companies than a large diversified technology business that happens to sell some edge products.
The growth rate is less explosive than Anduril's or some younger robotics companies. Commercial scale is a different question, though, and Mobileye already operates at a level that most private edge AI startups are nowhere near.

This chart, featured in our edge AI market deck, shows how Hailo is winning in edge AI
How big is Samsara's edge AI business?
Samsara has become a huge physical-operations AI company, reaching $2.125 billion of ARR after growing that figure 30% year over year.
The company also recently reported $508.4 million of quarterly revenue, up 30%. More than $500 million of ARR now comes from customers spending over $1 million a year.
We should still be careful with the edge AI label here. Samsara sells a broad connected-operations platform covering cameras, vehicle telematics, equipment monitoring, safety, workflows and other operational software. AI is becoming increasingly important across that stack, especially in video and physical operations, but the entire $2.125 billion cannot reasonably be described as pure edge AI revenue.
Samsara therefore shows how large an edge-AI-enabled platform can become. It is less useful as a clean measure of the standalone edge AI market.
Once company revenue reaches this scale, a loose market definition can add billions of dollars to a ranking very quickly.
How much money are warehouse robotics companies making?
Warehouse robotics has already become one of the biggest commercial edge AI markets, led by Symbotic, Geek+ and Locus Robotics.
Symbotic generated $2.25 billion in its last completed fiscal year and has since posted $721 million in quarterly revenue. Geek+ generated RMB3.17 billion in 2025, up 31.6% year over year. Almost all of Geek+'s revenue came from robotics solutions.
Locus Robotics sits lower but still represents substantial scale. Sacra estimates the company reached roughly $180 million of ARR as its Robots-as-a-Service model expanded across fulfillment customers.
Smaller businesses fill out the market. Dexterity is estimated above $20 million, Gather AI has disclosed roughly $15 million, and Standard Bots is estimated around a $24 million annualized revenue pace.
The pattern is pretty clear: warehouse AI is producing both billion-dollar systems companies and recurring-revenue robotics businesses. Customers can tie the investment directly to picking speed, labor requirements and throughput, which makes the buying case easier than in more experimental edge AI categories.
| Company | Latest useful figure | Metric |
|---|---|---|
| Symbotic | $2.25B | FY revenue |
| Geek+ | RMB3.17B | Annual revenue |
| Locus Robotics | ~$180M | Estimated ARR |
| Standard Bots | ~$24M | Estimated run-rate |
| Dexterity | ~$21M | Revenue estimate |
| Gather AI | ~$15M | Revenue |
| Agility Robotics | ~$1.8M | Annual net sales |

This chart, featured in our edge AI market deck, illustrates yearly funding for edge AI startups
Are automotive edge AI companies already billion-dollar businesses?
Automotive edge AI is already a multi-billion-dollar commercial market, with Mobileye alone operating close to $2 billion in annual revenue.
ECARX generated $847.9 million in 2025. Horizon Robotics reported RMB3.76 billion, up 57.7%, with automotive solutions accounting for the overwhelming majority of its business. Cerence AI generated around $252 million in its latest completed fiscal year, while indie Semiconductor generated roughly $217 million.
Black Sesame Technologies added another RMB822 million of annual revenue. Autonomous-driving companies are smaller but growing quickly: WeRide reached $97.9 million in 2025 after revenue jumped almost 90%, while Pony.ai generated $90 million.
These figures show how far automotive edge AI has moved beyond R&D. ADAS processors, cockpit systems, voice interfaces, autonomous-driving software and vehicle compute platforms are already generating hundreds of millions of dollars for individual vendors.
Among all the edge AI verticals we reviewed, automotive has one of the deepest groups of companies with measurable revenue.
Which edge AI chip companies make the most money?
Horizon Robotics and Ambarella currently sit near the top of the edge AI semiconductor market on reported commercial scale, followed by Black Sesame Technologies and CEVA.
Horizon generated roughly RMB3.76 billion in 2025. Its business includes both hardware and software for automotive intelligence, so it has a broader revenue perimeter than a pure chip supplier.
Ambarella provides a cleaner semiconductor datapoint. It generated $390.7 million in its latest completed fiscal year, up 37.2%. The company specifically attributed much of that growth to higher shipments and a richer mix of AI inference processors.
Black Sesame generated RMB822 million, mainly from automotive driving-assistance products. CEVA generated roughly $110 million across edge-oriented semiconductor IP and related markets.
After those companies, revenue falls sharply. Blaize generated $38.6 million. Quadric disclosed $15 million to $20 million of licensing revenue. DEEPX has reported more than $13 million of purchase orders, while BrainChip remained below $2 million in annual revenue.
| Company | Latest useful figure | Metric |
|---|---|---|
| Horizon Robotics | RMB3.76B | Annual revenue |
| Ambarella | $390.7M | FY revenue |
| Black Sesame Technologies | RMB822M | Annual revenue |
| CEVA | ~$110M | Annual revenue |
| Blaize | $38.6M | Annual revenue |
| Quadric | $15M–$20M | Licensing revenue |
| DEEPX | >$13M | Purchase orders |
| BrainChip | $1.89M | Annual revenue |

This chart, featured in our edge AI market deck, compares the main business model options for edge AI accelerator companies
Is Verkada really a billion-dollar edge AI startup?
Verkada has crossed the billion-dollar commercial threshold, but its disclosed figure is annualized bookings rather than revenue.
The company said annualized bookings had surpassed $1 billion when it announced its latest valuation. It also reported more than two million devices deployed across 171 countries and more than 30,000 customer organizations.
Those operating numbers make the scale credible. Verkada is selling cameras, access control, sensors, alarms and other physical-security products around a software platform that increasingly uses AI for monitoring and investigations.
Still, $1 billion of bookings cannot be dropped into a revenue table as though the accounting treatment were identical to Cognex's $994 million of annual revenue.
We keep Verkada near the top because its commercial footprint is enormous. The metric simply carries a little less weight than a comparable amount of recognized annual revenue.
How much revenue does Shield AI make compared with Anduril?
Shield AI generated roughly $300 million in 2025, leaving Anduril more than seven times larger by annual revenue.
That difference is striking because the two companies are often grouped together as leading private defense-AI businesses.
Shield AI is already substantial. A $300 million autonomous-systems company is bigger than most private edge AI startups we found. Its Hivemind autonomy software and aircraft programs also put AI at the core of the product rather than treating it as a feature added to a broader software platform.
Anduril has simply reached a different level. Its $2.2 billion of 2025 revenue reflects a much broader product range, much larger defense programs and unusually rapid scaling.
Defense edge AI is concentrated. Several startups are winning serious government business, but one company has pulled far ahead on revenue.

This chart, featured in our edge AI market deck, shows revenue distribution by customer segment in the edge AI market
How fast is Skild AI growing?
Skild AI has become one of the fastest-growing commercial companies in physical AI, crossing $100 million of ARR only ten months after its first commercial deployment.
The company says it now has more than 60 paying customers across warehouses, factories, delivery, security, food preparation, construction and data centers.
The strongest detail is the recognized revenue behind the ARR. Skild said $50 million had already been recognized since commercial deployments began. That gives us more confidence than an ARR figure supported mostly by newly signed contracts or pilots.
Most of Skild's revenue also comes from manipulation tasks rather than mobility, according to the company. That suggests demand is spreading into actual industrial work rather than being concentrated in robot navigation.
Skild is still far smaller than Anduril, Symbotic or Mobileye. The speed of the ramp is what stands out. Going from first deployments to nine-figure ARR in less than a year is rare even in the current AI market.
Is Locus Robotics bigger than the new humanoid robot startups?
Locus Robotics currently has a much larger proven commercial business than most private humanoid robot companies.
Sacra estimates Locus at roughly $180 million of ARR. The company has had years to build recurring deployments in warehouses through its Robots-as-a-Service model, so it benefits from an installed fleet and customers expanding robot usage over time.
The humanoid category looks very different. Agility Robotics reported only about $1.8 million of 2025 net sales in recently filed financial information, despite having more than $300 million of multi-year orders.
Figure AI and Apptronik have announced major customers and factory deployments, but neither has disclosed a company-wide revenue figure strong enough for us to place confidently in the ranking.
This is a useful reminder when looking at robotics headlines. Humanoid startups may carry much larger valuations and fundraising totals, while an older warehouse robotics company can still have far more current revenue.

This chart, featured in our edge AI market deck, shows how on-device AI assistant technology has evolved over time
Are humanoid robot startups actually making much revenue yet?
Humanoid robotics is still much larger in funding, valuations and promised orders than it is in recognized revenue.
Agility Robotics gives us one of the rare audited datapoints. The company generated roughly $1.8 million in net sales during 2025, even though its multi-year order book has moved above $300 million.
That gap makes sense at this stage of the market. Large orders can cover future robot deliveries, software and services over several years. Revenue only appears as the company actually delivers what the customer bought.
Figure AI has shown production deployments with BMW. Apptronik has announced commercial programs with major companies. Those deployments matter because they move humanoids beyond demonstrations, but there is still no solid public company-wide revenue figure that lets us compare either business with Locus, Symbotic or Geek+.
For now, humanoid robotics remains an early commercial market. Skild AI is scaling much faster financially, although its robot intelligence is designed to work across multiple hardware types rather than one humanoid platform.
Which smaller edge AI startups have already crossed $10 million in revenue?
A surprisingly broad group of edge AI companies has moved beyond the $10 million commercial threshold, although several figures remain estimates or alternative metrics.
BrightAI disclosed $80 million of revenue when it emerged from stealth. Rekor Systems generated $48.5 million in 2025. Blaize generated $38.6 million. Standard Bots is estimated around a $24 million annualized pace, and Dexterity around $21 million.
Quadric disclosed $15 million to $20 million of licensing revenue. Gather AI has reached about $15 million according to its CEO. DEEPX has reported more than $13 million of purchase orders since moving into mass production.
The interesting break comes above $100 million. Plenty of edge AI businesses can now reach eight figures, but relatively few progress into nine-figure revenue or ARR.
There is no shortage of early commercial adoption. Building a large repeatable distribution engine is where the list gets much shorter.
| Company | Latest useful figure | Evidence |
|---|---|---|
| BrightAI | $80M | CEO-disclosed revenue |
| Rekor Systems | $48.5M | Filed annual revenue |
| Blaize | $38.6M | Filed annual revenue |
| Standard Bots | ~$24M | Estimated run-rate |
| Dexterity | ~$21M | Revenue estimate |
| Quadric | $15M–$20M | CEO-disclosed licensing revenue |
| Gather AI | ~$15M | CEO-disclosed revenue |
| DEEPX | >$13M | Purchase orders |

In our edge AI market deck, we identify pain points entrepreneurs should prioritize
Are edge AI software startups smaller than hardware companies?
Pure edge AI software businesses are generally smaller than the companies selling complete robots, vehicles, cameras or industrial systems.
Skild AI is the strongest recent exception. Its software has crossed $100 million of ARR while working across different robot platforms.
Most other software-focused edge AI companies remain much smaller. Nota AI disclosed annual revenue above KRW10 billion. Nexa AI and Edgify appear to remain in the low-single-digit millions based on third-party estimates. Edge Impulse built a very large developer ecosystem before Qualcomm acquired it, yet no clean revenue figure was publicly disclosed.
Compare that with Symbotic at $2.25 billion, Mobileye at $1.89 billion, ECARX at $848 million, Geek+ at RMB3.17 billion and Ambarella at $391 million.
Customers seem willing to spend far more when AI arrives inside a finished operational product. Software can eventually capture a large independent layer, and Skild may be starting to prove that. So far, the biggest pools of revenue still sit closer to the physical system.
Which edge AI sectors are making the most money right now?
Warehouse automation, automotive AI and defense autonomy currently stand out as the strongest revenue pools in edge AI.
Warehouse robotics has Symbotic above $2 billion, Geek+ in the hundreds of millions of dollars and Locus at an estimated $180 million of ARR.
Automotive is even broader. Mobileye operates close to $2 billion, ECARX approaches $850 million, Horizon Robotics has moved well beyond RMB3 billion, and a long tail of chip, software and autonomous-driving companies already generates meaningful revenue.
Defense has fewer large private companies, but Anduril alone generated $2.2 billion. Shield AI adds another roughly $300 million.
Physical security is also developing into a substantial category through Verkada, Cognex-adjacent machine vision use cases, Evolv and other vision platforms.
The biggest sectors share a practical advantage: customers can usually tie the technology to a costly real-world process. Warehouses can measure throughput. Automakers can spread a platform across hundreds of thousands of vehicles. Defense customers can fund large autonomous-system programs.

This chart, featured in our edge AI market deck, shows revenue breakdown by region across Europe, Asia, North America, Africa, and South America in the edge AI market
How concentrated is edge AI revenue?
Edge AI revenue is heavily concentrated in a small group of companies, even though the market contains dozens of funded startups.
At the top, Symbotic and Anduril each generate more than $2 billion. Mobileye is close behind. Samsara also operates above $2 billion of ARR, although its business perimeter is broader.
The next cluster is already around the $500 million to $1 billion range. Once we move below Ambarella and Shield AI, the list quickly enters the low hundreds of millions.
Then the company count explodes.
We found many businesses between $10 million and $100 million, plus another group with single-digit-million revenue or only deployment and customer metrics.
That shape tells us more than a market-size estimate would. Edge AI has clearly produced large companies, but commercial success is far from evenly distributed. A relatively small number of businesses have found a product and distribution model capable of reaching serious scale.
Which edge AI companies are growing fastest right now?
Anduril, Skild AI, Horizon Robotics, Geek+ and several large public edge AI companies are currently posting some of the strongest growth rates we found.
Anduril more than doubled revenue to $2.2 billion in 2025. Doubling at that size is much harder than doubling from a small base.
Horizon Robotics increased revenue 57.7% to RMB3.76 billion. Geek+ grew 31.6% to RMB3.17 billion. Ambarella's latest completed fiscal year grew 37.2% to $390.7 million.
Samsara has recently maintained 30% growth in ARR, reaching $2.125 billion. Symbotic's latest quarter grew 22% to $721 million.
Skild AI is harder to compare because it started from such a small commercial base, but its move beyond $100 million of ARR within ten months of first deployment makes it one of the fastest ramps in the entire group.
The strongest growth is happening at both ends of the market: some billion-dollar companies are still compounding quickly, while a handful of young physical AI startups are jumping into the nine-figure range.
| Company | Recent growth marker | Current scale |
|---|---|---|
| Anduril | Revenue more than doubled | $2.2B revenue |
| Skild AI | >$100M ARR within 10 months | >$100M ARR |
| Horizon Robotics | +57.7% annual revenue | RMB3.76B |
| Ambarella | +37.2% annual revenue | $390.7M |
| Geek+ | +31.6% annual revenue | RMB3.17B |
| Samsara | +30% ARR | $2.125B ARR |
| Symbotic | +22% latest-quarter revenue | $721M quarterly revenue |
| WeRide | +89.6% annual revenue | $97.9M |

This chart, featured in our edge AI market deck, illustrates yearly venture capital investment in edge AI startups
Which famous edge AI startups still do not disclose revenue?
Several well-known edge AI companies still cannot be ranked responsibly by revenue because the strongest public numbers describe customers, deployments or usage instead.
Axelera AI says it has more than 600 customers. Hailo previously disclosed more than 300. RealSense said it served more than 3,000 customers when it became independent.
Waymo has disclosed hundreds of thousands of paid rides per week, which is powerful evidence of real commercial activity. Brain Corp has tens of thousands of robots deployed. Figure AI and Apptronik have announced high-profile industrial deployments.
These numbers tell us that the companies have moved beyond pure R&D. They still do not tell us annual company revenue.
Trying to fill the gap with assumptions would make the ranking look cleaner while making it less reliable. A customer count cannot tell us average contract value, a robot deployment cannot tell us pricing, and paid rides cannot tell us how much revenue remains after incentives or other operating arrangements.
| Company | Strongest public scale evidence | Why there is no revenue rank |
|---|---|---|
| Axelera AI | 600+ customers | Revenue undisclosed |
| Waymo | Hundreds of thousands of paid rides weekly | No clean standalone revenue figure |
| Brain Corp | 50,000+ robots deployed | Deployment count only |
| RealSense | 3,000+ customers | Standalone revenue undisclosed |
| Spot AI | 1,000+ customers | Customer count only |
| Hailo | 300+ customers | Revenue undisclosed |
| AiFi | 370+ deployments | Deployment count only |
| Figure AI | Major factory deployments | Company-wide revenue undisclosed |
| Apptronik | Multiple commercial deployments | Company-wide revenue undisclosed |
So what is the best current ranking of private edge AI startups by revenue?
Anduril is the clear private edge AI revenue leader, while Verkada, Shield AI, Saronic, Locus Robotics and Skild AI form the next commercial tier on a mix of bookings, revenue, estimates and ARR.
Anduril has the strongest combination of scale and a direct company disclosure at $2.2 billion of annual revenue.
Verkada's >$1 billion figure is annualized bookings, so we place it below Anduril despite its huge commercial footprint. Shield AI's roughly $300 million is a cleaner annual-revenue datapoint. Saronic is estimated around $200 million, but the source quality is weaker. Locus Robotics sits around $180 million of estimated ARR.
Skild AI has now crossed $100 million of ARR, with $50 million already recognized since deployments began. That gives the company one of the strongest fresh commercial stories among young physical AI businesses.
BrightAI follows at $80 million based on an older direct disclosure. Below that level, the ranking becomes much more sensitive to whether we trust recognized revenue, run-rate estimates or company-specific operating metrics.
| Rank | Private company | Best available commercial figure | What the figure actually is |
|---|---|---|---|
| 1 | Anduril | $2.2B | Annual revenue |
| 2 | Verkada | >$1B | Annualized bookings |
| 3 | Shield AI | ~$300M | Annual revenue |
| 4 | Saronic | ~$200M | Revenue estimate |
| 5 | Locus Robotics | ~$180M | Estimated ARR |
| 6 | Skild AI | >$100M | ARR |
| 7 | BrightAI | $80M | Revenue, older disclosure |
| 8 | Standard Bots | ~$24M | Estimated revenue run-rate |
| 9 | Dexterity | ~$21M | Revenue estimate |
| 10 | Quadric | $15M–$20M | Licensing revenue |
The wider market is much larger than this private-company table suggests. Symbotic, Mobileye, Samsara, Cognex, ECARX, Horizon Robotics, Geek+ and Ambarella show what happens once successful edge AI businesses reach public-market scale.
The main conclusion is straightforward. Edge AI already supports several billion-dollar businesses, but only a small group has made that jump. Most companies remain below $100 million, rely on weaker estimates, or disclose commercial deployments without telling us revenue.
The money is clustering around very practical applications. Warehouses, vehicles, defense systems, industrial vision and physical security account for most of the largest companies we found. These are environments where local AI can change an expensive real-world process immediately, and customers are already paying accordingly.

In our edge AI market deck, we like to quantify things to make things easier to understand
OUR METHODOLOGY
This analysis ranks edge AI companies by the strongest available evidence of current commercial scale. We use edge AI broadly enough to include robotics, automotive AI, defense autonomy, machine vision, physical security, connected operations and edge-oriented semiconductors, but we keep the underlying business model visible so a broad platform is not mistaken for a pure-play edge AI company.
We keep annual revenue, fiscal-year revenue, ARR, annualized bookings, run-rate figures, purchase orders, estimates and deployment metrics in their original categories. Direct company-wide revenue carries the most weight; ARR and other recurring-revenue measures are useful secondary indicators, while bookings, orders, customer counts and deployments are treated as evidence of commercial scale rather than synthetic revenue.
Company status matters as well. The strict private-startup ranking is led by Anduril, while companies such as Symbotic, Mobileye, Samsara, Cognex, ECARX, Horizon Robotics, Geek+ and Ambarella are used as broader market benchmarks because they have already reached public-market scale or no longer fit a narrow startup definition.
When several disclosures existed for one company, we favored the figure that best represented current company-wide commercial scale, then considered source quality and freshness. We did not manufacture revenue by multiplying prices, users, robots, customers, rides, contracts or other operating figures.
Key sources include Symbotic's FY2025 results, Symbotic's latest quarterly results, Reuters reporting on Anduril's 2025 revenue, Mobileye's FY2025 results, Samsara's latest results, Verkada's annualized bookings disclosure, ECARX's FY2025 results, Horizon Robotics investor materials, Geek+'s 2025 annual results, Ambarella's FY2026 results, Skild AI's ARR and recognized-revenue disclosure, SEC-filed Agility Robotics transaction materials, and TechCrunch's report on BrightAI's $80 million revenue disclosure.

In our edge AI market deck, we tell you what to focus on