Signals Inbox·July 21, 2026·Defense Tech

Shield AI vs Anduril vs Skydio: who owns military autonomy?

Anduril owns the broad military-autonomy race today, Shield AI leads the portable AI-pilot layer, and Skydio has built the strongest position in small tactical drones.

Q1Why do people compare Shield AI, Anduril and Skydio?

Shield AI, Anduril and Skydio belong in the same debate because each wants to become a default supplier of autonomous military aircraft, even though each company is trying to control a different part of the system.

Anduril wants to sell the complete package. A military customer can buy the vehicle, sensors, autonomy software, command system and, increasingly, the manufacturing capacity from the same company.

Shield AI starts with the pilot. Hivemind is designed to help different aircraft perceive their surroundings, make decisions and work together when GPS or communications become unreliable. Shield AI also builds V-BAT and is developing X-BAT, but its largest long-term opportunity lies in putting Hivemind inside vehicles made by other companies.

Skydio starts with the individual user. X10D helps a soldier, bomb-disposal team or security unit fly a reconnaissance drone without becoming an expert pilot. The company is trying to make autonomous flight common across thousands of local missions.

The comparison has become much more direct lately. The U.S. Air Force now separates the Collaborative Combat Aircraft airframe from its mission-autonomy software. Anduril supplies one of the aircraft, while Anduril and Shield AI compete to provide the software inside it. Skydio, meanwhile, has received the largest disclosed single-vendor tactical-drone order in U.S. Army history.

Skydio therefore belongs in this comparison across a smaller surface area. It competes directly with Anduril in compact autonomous aircraft, but much less in autonomous fighters, multi-agent combat software and force-wide command systems.

Q2Why is it so hard to declare one military-autonomy winner?

Anduril, Shield AI and Skydio lead three different military-autonomy races, so using one scoreboard would produce the wrong winner.

Skydio makes individual drones easier to operate. Shield AI gives aircraft more freedom to complete missions. Anduril connects aircraft, sensors and weapons into a wider operational system.

Those jobs require different evidence. Skydio can show how many drones it delivers. Shield AI can show how many unrelated aircraft Hivemind can pilot. Anduril can show how many military programs and battlefield systems Lattice connects.

Revenue does not settle it either. Anduril earns money from missiles, surveillance towers, rocket motors and underwater vehicles alongside autonomy. Shield AI is more concentrated around autonomous aircraft and software. Skydio’s business includes public safety, utilities and industrial inspection.

We judge the companies on five questions: who controls the most important programs, whose software works across the most hardware, who has entered real production, who can manufacture at scale and who can keep funding the race.

Q3Which company is actually biggest today: Anduril, Shield AI or Skydio?

Anduril is clearly the biggest business today, with reported revenue several times higher than either Shield AI or Skydio.

Anduril reported $2.2 billion in 2025 revenue after approximately doubling in one year. That implies an increase of roughly $1.1 billion, probably more than the entire current revenue of either rival.

Shield AI does not disclose exact revenue. Investor materials reported by the Financial Times valued the company at around 21 times projected 2026 revenue. Applied to its $12.7 billion valuation, that suggests a forecast close to $600 million, although it remains a projection.

Skydio says it generates “hundreds of millions” in annual revenue. The company does not separate military sales from police, utility and industrial customers.

Anduril is probably three to seven times larger, depending on where Shield AI and Skydio fall within their disclosed ranges. Not all of that gap comes from autonomy, but the additional revenue gives Anduril far more money to develop it.

Latest useful revenue evidence for Anduril, Shield AI and Skydio

Q4Who is growing fastest right now: Anduril, Shield AI or Skydio?

Anduril is adding the most revenue, Shield AI may be growing faster in percentage terms, and Skydio has disclosed too little to win the comparison.

Anduril doubled revenue to $2.2 billion. The percentage is impressive, but the dollar increase is more revealing: the company appears to have added around $1.1 billion in one year.

Shield AI also says its latest fiscal-year revenue and orders doubled. During the same period, Hivemind expanded to 30 vehicle classes, gained a production position in the Air Force’s Collaborative Combat Aircraft software competition and reached new aircraft manufacturers.

Shield AI could be matching Anduril’s percentage growth from a much smaller starting point. Without a reported revenue figure, we cannot prove it.

Skydio describes its business as being in hypergrowth. Recent Army and Air Force orders support the direction of that claim, but the company gives no annual growth percentage or military revenue figure.

Anduril takes this one because it combines rapid percentage growth with by far the largest absolute increase. Shield AI is the plausible challenger hiding beneath the undisclosed numbers.

Q5Which company has the strongest military demand today?

Anduril has the strongest future demand because it sits inside more top-priority military programs, although Skydio has the cleanest volume order and Shield AI has the broadest recent run of foreign wins.

Anduril now has an Air Force production contract for the FQ-44 Collaborative Combat Aircraft. It also holds an $86 million Special Operations Command contract to integrate mission autonomy and led the Army’s $99.6 million Next Generation Command and Control prototype.

The Army has since selected Anduril to lead the common data layer as NGC2 expands. That position could be more valuable than a single aircraft order because other military applications, sensors and vehicles may need to connect through Lattice.

Shield AI has a Coast Guard contract with a ceiling of approximately $198 million for V-BAT surveillance services and a production position in the Air Force’s CCA software competition. Poland, Greece, Japan and India have also selected V-BAT or entered related production and software agreements.

Skydio has the clearest unit count. The Army ordered more than 2,500 X10D systems for over $52 million, showing that the drone is moving beyond trials into wide deployment.

Anduril comes out ahead because its programs influence how the wider military will connect and operate autonomous systems.

Strongest current military-demand evidence

Q6Which company has moved furthest from military demos to real production?

Anduril is furthest past the demo stage at the high end, while Skydio is already producing the largest disclosed number of tactical drones.

The Air Force’s latest testing makes Anduril’s progress difficult to dismiss. An FQ-44 fired an AIM-120 missile at a digital target after earlier trials had validated weapon carriage and data-link integration. A human retained authority over the release, while the aircraft completed the employment sequence.

The Air Force had already concluded that the FQ-44 met its mission requirements and was ready for full-scale manufacturing. Anduril moved from acquiring Blue Force Technologies in 2023 to winning a prototype position, flying the aircraft and securing production in less than three years.

Skydio is further ahead in unit output. Its Army order covers thousands of drones that already belong to an established program of record. X10D is also reaching Air Force explosive-ordnance-disposal and tactical-control units.

Shield AI has a production aircraft in V-BAT, with users in the United States, Ukraine and several allied countries. X-BAT remains in development and has yet to match the FQ-44’s flight, weapons and production milestones.

Anduril leads the harder production transition. Skydio leads in immediate quantity.

Q7Does Anduril’s Lattice or Shield AI’s Hivemind control the software layer?

Lattice controls more of the battlefield today, while Hivemind is the stronger candidate to become the reusable AI pilot inside many different aircraft.

Lattice collects information from sensors, builds a shared operational picture, sends tasks to machines and helps coordinate actions across air, land and sea.

During a recent division-wide Army deployment, Anduril and its partners connected 40 applications, 36 data feeds and 37 sensors or effectors across nine hardware types. Lattice is already doing far more than a drone autopilot.

Hivemind concentrates on what happens inside the vehicle. It helps an aircraft understand its surroundings, select a route, react to threats and coordinate with other aircraft. Shield AI says it now works across 30 vehicle classes.

Skydio Autonomy solves a more local problem. It keeps X10D stable, avoids obstacles, maps its surroundings and reduces the skill required to fly through difficult terrain.

Lattice leads battlefield coordination. Hivemind leads reusable AI piloting. Skydio leads autonomous flight inside a compact tactical drone.

Who leads each part of the military-autonomy software layer?

Q8Which autonomy software works best on other companies’ aircraft?

Shield AI is clearly ahead when autonomy software has to fly somebody else’s aircraft.

The strongest example came from Anduril’s own YFQ-44A. During one flight, the aircraft ran both Anduril’s Lattice software and Shield AI’s Hivemind. The Air Force’s government-owned architecture allowed the same aircraft to switch between competing autonomy providers.

That flight belongs to a broader pattern. Mitsubishi Heavy Industries integrated Hivemind into its ARMD test aircraft and reached autonomous flight in under 60 days. Northrop Grumman flew Hivemind on its Talon IQ testbed. L3Harris used it on Green Wolf during an autonomous electronic-warfare exercise.

Four manufacturers have now flown or tested Hivemind on four distinct aircraft families. That is much stronger portability evidence than demonstrating autonomy only on your own hardware.

Anduril can integrate outside systems through Lattice, published APIs and its SOCOM role. Yet its clearest autonomous-aircraft results still tend to involve Anduril vehicles.

Skydio supports government and third-party applications on X10D, but Skydio Autonomy has not become the pilot inside several unrelated aircraft.

Hivemind already behaves like an independent product. Shield AI has a wide lead here.

Q9Who is winning the autonomous-fighter race: Anduril or Shield AI?

Anduril is winning the autonomous-fighter race now because it owns the aircraft moving into production, while Shield AI owns a valuable software seat inside that program.

Anduril captures more of the FQ-44’s economics. It designs and manufactures the airframe, integrates the mission systems and can compete to supply Lattice as the aircraft’s AI software.

Shield AI can earn licensing and integration revenue when the Air Force selects Hivemind. The company has already secured a production option alongside Anduril and RTX Collins Aerospace.

The Air Force continues to keep the software contest open. Six companies remain in the approved mission-autonomy pool, allowing the service to replace or combine suppliers rather than giving one company permanent control.

Shield AI’s larger counterattack is X-BAT. Its vertical-takeoff design could reduce dependence on long and vulnerable runways, especially across the Pacific.

X-BAT still needs to fly, complete military testing and secure a production customer. The FQ-44 has already crossed those stages.

Shield AI can win the pilot competition without winning the fighter competition. Today, the larger victory belongs to Anduril.

Q10Who is winning tactical military drones: Anduril, Shield AI or Skydio?

Skydio is the clear leader in platoon-level autonomous reconnaissance, where disclosed unit volume already reaches the thousands.

The Army’s purchase of more than 2,500 X10Ds gives Skydio a scale advantage that is easy to understand. These drones are intended for soldiers who need local reconnaissance without requesting a larger aviation asset.

Skydio has been chosen across both major tranches of the Army’s Short Range Reconnaissance program. X10D is currently the only system delivered under the second tranche, and the Air Force has expanded its use among explosive-ordnance-disposal teams.

Anduril’s Ghost family can carry broader payloads and connect directly into Lattice. Ghost may be more useful for electronic warfare, targeting or wider command networks, but Anduril has disclosed no comparable recent unit order.

Shield AI’s V-BAT flies for more than 12 hours and can launch vertically from ships or austere positions. It belongs to a heavier aircraft class with a different mission and price.

Skydio wins because X10D is becoming a standard tool for small units. Its next test is whether those thousands of drones remain useful under intense electronic warfare.

Q11Which company works best when GPS and communications disappear?

Shield AI has the best public evidence that its autonomy can keep working when GPS, communications and constant human control become unreliable.

Hivemind processes information onboard, allowing an aircraft to react locally instead of waiting for every decision to arrive from a remote operator.

Shield AI’s electronic-warfare flight with L3Harris moved that claim beyond a navigation test. Green Wolf aircraft detected electromagnetic threats, shared the information and used Hivemind to reroute follow-on aircraft through a safer area without human input.

V-BAT has also operated in Ukraine, where electronic warfare regularly disrupts navigation and communications. Shield AI has maintained a local team working with Ukrainian operators and used those missions to improve the aircraft.

Anduril is tackling the same problem through Lattice Mission Autonomy, Ghost and its Replicator work. Army exercises have tested Lattice under degraded network conditions, although public evidence from contested flights is less detailed.

Skydio says X10D can navigate without GPS through onboard cameras and terrain mapping. Independent evidence from severe electronic-warfare conditions remains thinner.

Shield AI leads on the evidence available today. Anduril is close enough that the result is not settled.

Q12What have Shield AI, Anduril and Skydio actually proved in Ukraine?

Shield AI has the strongest positive Ukraine record of the three, but none has matched the scale or speed of Ukraine’s cheaper local drone industry.

Shield AI disclosed more than 130 V-BAT sorties by April 2025 and has kept personnel in Ukraine to support operations and training. V-BAT has mainly provided surveillance, target detection and intelligence rather than autonomous strikes.

The number is modest beside the scale of the war, but repeated sorties in a heavily jammed environment provide better evidence than a controlled test range. Poland and Greece have since referred to V-BAT’s Ukrainian experience when expanding their own drone capabilities.

Anduril’s Ghost-X has reportedly operated in Ukraine since 2022. Reliable figures on sorties, availability and mission success remain scarce, so its record cannot be compared directly with V-BAT.

Skydio has the clearest setback. According to reporting by The Wall Street Journal, hundreds of its earlier drones struggled with Russian electronic warfare, went off course and proved difficult to repair. X10D was developed with those lessons in mind, but still needs new battlefield evidence.

Shield AI takes the comparison. The wider battlefield remains dominated by a fragmented local ecosystem that modifies and replaces drones much faster than traditional American procurement normally allows.

Q13Who can manufacture military-autonomy systems at serious scale?

Anduril has the strongest manufacturing position for the long term, while Skydio has better proof of high-volume output today.

Anduril is building Arsenal-1 in Ohio, a five-million-square-foot site ultimately intended to produce tens of thousands of autonomous systems each year. The FQ-44 is expected to become one of its first major aircraft programs.

The strategy is broader than one factory. Anduril wants aircraft, weapons, sensors and autonomous vehicles to share production equipment, components and suppliers. That could shorten development cycles and lower costs across several product families.

Skydio’s operation is smaller, but its present output is easier to verify. X10D is already built in California, and the Army has ordered more than 2,500 systems.

Shield AI manufactures V-BAT in the United States and is creating an additional production base in India through a partner investing $90 million.

Anduril has the largest credible path to future capacity. Skydio has the strongest proof that promised capacity is already turning into unit deliveries.

Q14Which company has the best military customers and distribution?

Anduril has the most influential U.S. military positions, Shield AI has the strongest allied-country network, and Skydio has the easiest product to buy repeatedly.

Anduril sells into programs that shape the wider system. The Air Force has selected its combat aircraft, the Army uses Lattice in its emerging NGC2 architecture, and Special Operations Command chose Anduril for mission-autonomy integration.

Shield AI is expanding fastest across allied governments. V-BAT has gained customers or production relationships in Japan, India, Greece, Poland and the United States. Several agreements include local manufacturing or sovereign software work, making Shield AI more attractive to governments seeking domestic control.

Skydio says every U.S. military branch and 29 allied countries use its systems. X10D’s approved procurement status and compact size allow units to buy additional batches without creating a major aviation program.

Anduril ranks first because its U.S. positions influence decisions above the individual-product level. Shield AI has built the strongest alternative route through allied governments.

Q15Which company has enough money to keep winning?

Anduril can afford a longer and more expensive fight than Shield AI or Skydio, and the financial gap is enormous.

Anduril raised $5 billion at a $61 billion valuation. That capital can fund factories, aircraft development and early production before government orders fully arrive.

Shield AI raised $1.5 billion in equity and $500 million in preferred financing at a $12.7 billion valuation. Blackstone also committed a further delayed facility. Shield AI now has enough money to expand V-BAT, absorb Aechelon and continue X-BAT development.

Skydio raised $110 million at a $4.4 billion valuation. The company says its core business increasingly funds its own investment needs.

Anduril’s latest round was 2.5 times larger than Shield AI’s announced package and roughly 45 times larger than Skydio’s round.

That advantage carries unusual weight in defense, where companies often finance years of engineering, testing and manufacturing before production revenue appears. Anduril can survive more delays and pursue more programs simultaneously.

The risk is overextension. Aircraft, missiles, underwater vehicles, sensors, software and factories all compete for management attention, however much money the company raises.

Q16Which company has the strongest long-term moat?

Anduril has the strongest combined moat today, while Shield AI has the clearest chance to build a more focused software moat.

Anduril’s advantage grows as its products connect. A new aircraft gives Lattice another platform. A new command contract brings more sensors and applications. A larger factory can support several product families.

Military customers also face practical switching costs. Replacing a command network connected to applications, sensors, weapons and training procedures is much harder than changing one drone model.

Shield AI’s potential moat comes from reuse. Hivemind now works across 30 vehicle classes. Improvements developed for one aircraft may transfer to others, allowing Shield AI to spread its research cost across multiple manufacturers.

Aechelon could deepen that advantage by giving Hivemind access to high-fidelity simulated environments. The software can practise rare or dangerous missions before entering a real aircraft.

Open military architectures reduce lock-in for both companies. The Air Force can already run Lattice and Hivemind on the same aircraft, so neither supplier can rely on closed interfaces alone. Shield AI must keep winning on performance, while Anduril can still earn revenue from the aircraft, command system or factory when another company supplies the pilot.

Skydio has millions of customer flights, strong visual-navigation experience and a tightly integrated design. Those advantages improve X10D, but small military drones face constant price pressure and fast imitation.

Anduril’s moat remains stronger because it combines software, hardware, contracts and production. Shield AI could eventually build the better pure-software business if Hivemind becomes the preferred pilot across several major aircraft programs.

Q17Who owns military autonomy right now: Shield AI, Anduril or Skydio?

Anduril owns the broad military-autonomy race today by a clear margin, Shield AI leads the portable AI-pilot race, and Skydio leads the smaller tactical-drone race.

Three advantages decide the overall result.

First, Anduril controls more high-value programs. Its position covers autonomous combat aircraft, Army command software, special-operations autonomy, counter-drone systems and underwater vehicles. Neither rival influences as much of the military system.

Second, Anduril makes money across more of the chain. It can earn revenue from aircraft, sensors, software, integration, weapons and production. Shield AI captures more of the autonomy layer, while Skydio captures the tactical-drone layer.

Third, Anduril has far more capital and industrial capacity. Reported revenue reached $2.2 billion, the latest funding round brought in $5 billion, and Arsenal-1 is being built to support several product families.

Shield AI remains the most dangerous challenger. Hivemind has flown aircraft from several manufacturers, won an Air Force production position and completed an autonomous electronic-warfare mission on an L3Harris aircraft.

To overtake Anduril, Shield AI must turn those integrations into large recurring software revenue, reach major X-BAT flight milestones and win production work across several aircraft families.

Skydio must prove that X10D performs reliably under severe electronic warfare, convert its Army deployment into sustained repeat orders and move from autonomous reconnaissance toward coordinated multi-drone missions.

Anduril still has to deliver. FQ-44 delays, weak operational adoption of NGC2 or slow progress at Arsenal-1 would reduce the value of its current lead. Successful execution across all three would make the gap much harder to close.

The market will remain open enough that no company can lock every competitor out. For now, Anduril controls more of the valuable territory than anyone else.

Military-autonomy leaders by competitive criterion