Military AI revolution heightens competition for defence tech contracts: Peter Apps

LONDON, Sept 5 (Reuters) - In London’s Shoreditch district, long home to UK tech start-ups, a French artificial intelligence firm is crunching data that would have allowed Ukrainian commanders to more quickly counter Russia’s 2022 full-scale invasion.

Initially fed a fictitious set of orders for a Ukrainian brigade tasked to stop the Kremlin’s initial armoured column thrust south from Belarus towards the capital Kyiv, the AI program swiftly works its way through the available information including detailed mapping.

Sign up here.

Within the space of several minutes it has synthesised that information to include likely enemy advance routes, potential courses of action and other considerations for Ukrainian defenders.

Based in Paris, Comand AI was founded 20 months ago and recently received 12 million euros of venture capital funding.

That makes it a relatively new arrival on an ever-growing roster of “defence technology” or “miltech” firms.

But it has ambitious plans for growth, hoping to sell to the French, German and British militaries with a view to becoming a key European player.

Media coverage has often focused on autonomous “killer robot” drones making their own targeting decisions. But those watching defence technology most closely say the real revolution revolves around crunching data for much broader decision-making – something that modern militaries hope will give them the edge when it comes to hitting many targets quickly.

It is an arena that is growing fast.

In his new book "AI, Automation and War”, published this month, Exeter University Professor Antony King described the Gaza and Ukraine conflicts as potentially "a turning point in the history of war", suggesting "they may mark the moment when AI first began to be indispensable to military operations".

He said those wars showed no sign of AI taking over control of systems entirely – but AI systems were now becoming “more potent every week” and ever more central to military planning.

That process has already significantly revamped how military headquarters operate – including the integration of technology firms in a way never seen before.

That is bringing with it increasingly savage competition between firms with sometimes overlapping products, all in a race for military talent as well as programmers to deliver legitimacy and access.

As Comand AI’s UK Vice President Paul Billings puts it, the purpose of its programs is not to replace human decision-making and military planning. Instead, it is to make “a smarter decision faster”– or more accurately, multiple decisions much faster and better than the enemy.

“You have a lot of information coming in – the number of sensors on the battlefield is increasing very fast – and so your problem becomes cognitive overload,” said Billings, a former parachute officer. “Our aim is to process that and deliver something usable.”

For now, U.S. firms continue to dominate the market and debate, demonstrating a level of ambition few others in the West have yet come close to matching. The largest of those is Palantir, established in 2003 and the largest of several defence tech firms backed by PayPal billionaire Peter Thiel.

UKRAINIAN, ISRAELI LESSONS

Senior leaders at Palantir have talked publicly about being invited into Ukraine in the first days of the war to build the kind of systems for the government in Kyiv they had previously built for the U.S., Israel and other Western allies.

Both Ukraine and Israel have also built up their own expertise and companies.

In the case of Israel, such massive data-crunching enabled the Israeli armed forces to identify hundreds of targets they believed included certain Hamas militants in Gaza, Hezbollah leaders in Lebanon and much of the top Iranian military command targeted and killed during Israeli airstrikes this June.

Events in the Middle East remain controversial enough that some firms are holding back their involvement – or at least keeping quiet about it. But when it comes to stepping up engagement with the Pentagon and U.S. allies in Europe and the Pacific, some of the world’s largest tech firms including Microsoft, Google and Amazon are now also growing players.

Some rivals have already fallen by the wayside.

In 2024, Rebellion Defence, which held lavish parties for British generals and journalists at venues like the Tower of London, shuttered its UK operations, saying the UK market was too small, although it continues to negotiate new contracts with the Pentagon.

Even industry leader Palantir has in some ways been a victim of its own success.

As arguably the highest-profile listed company tying together the two growth areas of AI and defence, Palantir has seen its stock price rocket to roughly five times its value over a year ago – and up an astonishing 1,700 percent since it first listed in September 2020. That surge has made it one of the 10 most valuable tech firms in the U.S.

Palantir earnings have also grown spectacularly largely on the back of mainly U.S. government contracts exceeding $1 billion for the first time in the most recent quarter.

But that remains only a fraction of the earnings of similarly valued companies.

If Palantir and other firms cannot keep up a spectacular level of growth, they risk ending up grotesquely overvalued – something that has already seen Palantir’s stock price drop markedly since the start of August.

The largest U.S. government contracts – either with the Pentagon or other departments such as Homeland Security – can still be enormous and are growing fast.

Last month Palantir signed a multi-year contract with the U.S. Army to pull together existing AI and analysis work and build out new projects that could be worth some $10 billion.

Unapologetically American and with its leaders close to the Trump administration, Palantir has also won contracts with multiple U.S. allies wanting interoperability with the U.S.

The same is true of Anduril, founded in 2017, which unlike Palantir but like German rival Helsing also produces autonomous weapons systems as well as command, control and intelligence systems.

Increasingly, however, those same allied nations also want access to their own “sovereign” non-U.S. systems, demand for which appears to have increased markedly since the temporary shutdown of U.S. intelligence and some other support to Ukraine earlier this year.

Comand AI - which hopes to win contracts with the French, British and German armies as well as smaller European counterparts - and many of the other firms have made a point of hiring recent military talent alongside AI programmers.

Comand AI’s “lessons learned” tool can go through data and documents and draw conclusions in the style used by the French Armed Forces, while UK defence AI firm Adarga touts its alignment to the British military.

AI 'HALLUCINATIONS'?

Those contracts, however, are often a dramatic magnitude smaller than their U.S. equivalents.

Adarga, founded in 2016, recently announced that its latest contract with Britain's Ministry of Defence was worth “up to 12 million pounds” ($16 million), with the firm still described officially as a “start-up”.

As efforts to bring the Ukraine war to an end appear to falter, the relationship between these firms – both U.S. giants and smaller U.S. and European firms – and their Ukrainian counterparts is becoming ever deeper, the war increasingly seen as a “testbed” more than a profit opportunity.

Comand AI, for example, says it is working with Ukraine's Griselda, which describes itself as a team of "experienced programmers, military veterans and volunteers", to develop "an automated system of input, processing and transmitting information using artificial intelligence".

Ukrainian firms tend to operate on a very different financial scale to their Western counterparts.

In March, Griselda announced it had secured $600,000 in funding, a different order of magnitude to the multimillion or higher injects of the venture capital and other money of larger Western firms. But Ukrainian firms can often have a key advantage: access to large volumes of real-time combat data.

Increasingly, Western defence technology insiders say they are finding Ukrainian counterparts and the Ukrainian government itself are withholding access to such information, suddenly aware that it represents potentially lucrative intellectual property coveted by foreign firms and even governments.

Then there is Ukraine’s experience of how AI can shift the military mindset when it comes to decision-making in general and targeting in particular.

In both Ukraine and Israel, commanders are anecdotally seen more likely to accept a recommendation to take action if they believe it has come from computer-driven analysis, perhaps a sign that they feel less responsible.

That could prove a problem.

Speaking last year, U.S. 18th Airborne Corps deputy commander Brigadier General John Cogbill noted that AI “hallucinations” – incidents in which the AI effectively “makes up” information including details – continued to pose a periodic problem when using that technology.

Whether that remains the case going forward will depend heavily on how good AI becomes at catching its own errors, how much effort tech firms put into ensuring that it does – and to what extent commanders and defence ministries decide they really care. AI is already changing warfare fast – and the legacies of decisions made now may last for years to come.

By Peter Apps; editing by Mark Heinrich

Our Standards: The Thomson Reuters Trust Principles., opens new tab

Opinions expressed are those of the author. They do not reflect the views of Reuters News, which, under the Trust Principles, is committed to integrity, independence, and freedom from bias.