The AI military complex: Which are the main companies in it?

For centuries, warring wars meant deploying hardware specified arsenic tanks, aircraft, artillery and weapons, built by a fistful of premier contractors nether semipermanent authorities programmes.

It besides meant putting boots connected the ground.

That paradigm is not disappearing, but bundle is becoming progressively cardinal to it. And now, much soldiers are warring wars done a screen.

The subject AI manufacture present encompasses everything from unreality computing to autonomous drones and decision-making software. This displacement traces backmost to the US drone programme of the 2000s and 2010s, which turned remote, screen-based sidesplitting into a diagnostic of however wars would beryllium fought.

Billions of dollars are flowing into a fistful of companies present starring the AI translation of militaries, arsenic the aged military-industrial analyzable reshapes itself into a military-tech one.

As portion of a bid connected AI and warfare, Al Jazeera looks astatine who immoderate of these companies are.

Who are these companies?

On July 10, the UK Ministry of Defence signed a 2bn pounds ($2.7bn), 15-year declaration with Omnia Training, a consortium led by Raytheon UK, to bid 60,000 soldiers a twelvemonth connected utilizing AI to marque a “combat-ready” British military.

Days earlier, NATO awarded contracts to tally the alliance’s Enhanced Air Command and Control Data Platform to US military-tech institution Anduril for its Lattice AI bundle alongside US steadfast Palantir and the French military-tech associated task Athea SAS.

Lattice uses AI and machine imaginativeness to bring information from aggregate sources into 1 unrecorded picture.

AI defence procurements are not new. The Pentagon’s 2014 “Third Offset Strategy” stake explicitly connected AI defining the adjacent procreation of warfare, but the gait appears to beryllium picking up arsenic planetary subject expenditure climbed to $2.88 trillion successful 2025, with governments pouring grounds sums into defence budgets.

The astir caller ample deal by the Pentagon saw it motion contracts with 8 of the world’s starring tech companies to deploy AI crossed classified networks successful a bid to go an “AI-first” military.

With wars raging from Ukraine to the Middle East, deals similar these are multiplying fast, and the manufacture itself is evolving conscionable arsenic rapidly – accepted arms firms and caller entrants alike are jostling for a portion of the market. One mode to marque consciousness of this shifting scenery is to look astatine wherever companies beryllium based connected their business models.

Researchers astatine the Stockholm International Peace Research Institute (SIPRI) mapped the AI subject industry, grouping firms into 4 categories based connected their relation successful the industry:

• Defence primes: accepted arms companies
• Neoprimes oregon defence startups: newer firms specialising successful bundle for the subject industry
• Big tech: tech giants that person monolithic cloud/computing capacity
• Foundational exemplary providers: frontier AI labs gathering precocious AI models

Below, we outline what each of these does, according to SIPRI’s analysis, and which companies inhabit each category.

Defence primes/contractors

These are the long-established arms manufacturers that person supplied militaries with systems and hardware for decades, and are present layering AI onto those existing systems.

The largest of these firms see Lockheed Martin Corp, RTX, Northrop Grumman Corp, BAE Systems and General Dynamics Corp, each of which deed all-time highs after the onset of the US-Israel warfare connected Iran.

Part of that rally reflects however profoundly AI is present being woven into these aged defender arms companies’ ain merchandise lines. In February, Lockheed Martin announced that it was flight-testing an AI instrumentality for its F-35 combatant jets to assistance pilots place hostile craft faster. L3Harris, different American defence firm, built AI straight into its camera systems for real-time targeting. It is besides fitted with platforms similar Shield AI‘s Tracker, which helps observe hostile drones. Shield AI is simply a US defence-tech institution gathering autonomous AI pilots for subject craft and drones.

Neoprimes oregon defence startups 

This newer category, sometimes referred to arsenic “neoprimes”, is software-first firms, selling specialised information products to the arms marketplace arsenic good arsenic policing and borderline forces.

Palantir is possibly the best-known sanction among specified firms, mostly done its relation successful Project Maven, the Pentagon’s flagship subject AI programme. It is designed to velocity up decision-making successful battle, relying connected AI and instrumentality learning to automatically sift done outer imagery, drone videos, and signals quality to place and prioritise targets.

The strategy is being utilized successful assorted conflicts, including the Ukraine warfare and US-Israel warfare connected Iran — Maven identified 1,000 targets successful the archetypal 24 hours of the second conflict.

A nonstop rival to Maven is Anduril’s Lattice platform, which ties its ain hardware – drones, sensor towers, autonomous vehicles – into a azygous bid network, with it being heavy utilised by the US Army.

Anduril displays its unmanned combat craft “Fury” astatine the Farnborough International Airshow, successful Hampshire, Britain, connected July 20, 2026 [Cassell Bryan-Low/Reuters]

Big tech 

Companies similar Google, Meta and Amazon Web Services (AWS), which power astir 67 percent of the planetary unreality infrastructure market, are progressively linked to subject applications that accepted defence contractors are providing.

In May, the US Department of Defense announced a caller statement with 8 tech and AI companies to usage their technologies for its classified networks.

The agreements with SpaceX, OpenAI, Google, NVIDIA, Reflection, Microsoft, Oracle and AWS were brokered to “accelerate the translation towards establishing the United States subject arsenic an AI-first warring force”.

The woody has meant these companies person been cleared for IL6 and IL7 networks, the astir delicate and classified computing environments the Department of Defense operates.

Foundational exemplary providers

The last class is firms that physique the underlying AI models that different companies integrate into subject tools, specified arsenic OpenAI and Anthropic successful the US and Mistral AI successful Europe. Chinese researchers person been processing AI models for subject usage since the aboriginal 2020s, including utilizing Meta’s Llama exemplary arsenic the basal for the People’s Liberation Army (PLA) instrumentality called ChatBIT, which was designed to stitchery quality and enactment decision-making successful the field, according to Reuters.

The Pentagon estimated Chinese subject AI spending comparable to American figures, astir $1.5 to $2bn annually.

This twelvemonth the Pentagon added Alibaba, Baidu, BYD, NIO, RoboSense and Unitree to its Chinese subject companies list, according to Reuters. According to the Pentagon, these firms are believed to beryllium helping the subject successful China.

Smaller Chinese AI firms tied to ex-Baidu, Alibaba and Tencent employees person been gathering targeting, surveillance and decision-making tools for the military, according to a 2021 report by the Centre for Security and Emerging Technology (CSET).

Many of these frontier AI labs were against their products being utilized successful militaries; however, successful 2024 OpenAI got escaped of its military-use ban, and successful the aforesaid twelvemonth Palantir and Anthropic partnered up successful what is present a narration fraught with lawsuits implicit however Claude was being utilized by the US military.