Platforms on the Frontline: The Rise of the Platform Model in Defense Tech

[Marijn Hoijtink is an Associate Professor of International Relations at the Department of Political Science at the University of Antwerp. She leads the PLATFORM WARS project, which focuses on military applications of AI, how these shape new military practices, and the broader political-economic frameworks in which these take place.

Dr Jasper van der Kist is a researcher at the University of Antwerp. He is currently involved in the FWO-funded research project ‘Platform Wars’, where he studies the implications of Defence Tech’s digital platforms for contemporary warfare.]

With the wars in Ukraine and Gaza raging on, the military aspirations of the global technology sector continue to make headlines. International media commonly describe both conflicts as an ‘AI war lab’ or testing ground for the private technology sector. Among them are some of the largest and most powerful tech companies, such as Microsoft, Google, and Amazon, as well as newer (start-up) companies that specialize in the defense sector, such as Anduril, Shield AI, and Palantir – the latter rising as a billion-dollar ‘unicorn’ company which recently entered the S&P500. Most of these companies are based in the United States, but European military startups like Helsing AI are also joining the ranks.

While relations between the military and commercial technology sector have a more extended history that should be traced back to US military spending in the technology sector during the Cold War period, today’s developments warrant our critical attention. 

The aggressive mobilization of the platform as a software architecture and business strategy for defense is particularly important. In analogy with the success of dominant platforms in other domains – think of Airbnb, Facebook, or Android – defense tech market leaders claim that this ‘software-first’ approach has the potential to disrupt the traditional model of defense contracting and replace it with a more ‘agile’ and competitive approach to technological innovation. However, as we argue here, adopting the platform model in defense will likely lead to new monopolies and dependencies, which will be difficult to revert once the software platforms running military hardware and operations are in the hands of only a handful of defense tech companies. 

The Booming Market for Military AI

As Elke Schwarz describes in a recent contribution for Opinio Juris, current investment in military technology – and military Artificial Intelligence (AI) in particular – is booming. Roberto Gónzalez estimates that the US military and intelligence agencies have awarded contracts worth at least $53 billion to major tech companies between 2019 and 2022. These numbers may be considered trivial compared to the traditional defense industry dominated by prime contractors such as Lockheed Martin and Raytheon, with spending estimated to have risen to a record $2443 billion globally and $916 billion in the US in 2023. However, even if most defense spending goes to the primes, technology corporations and military startups still exert considerable influence in different ways.  

One notable way is by shaping key narratives driving investment in military technology. As Schwarz, Gónzalez, and others have argued, the deepening relationships between the technology sector and the military are fundamental in driving a set of narratives and beliefs underpinning the military AI hype. Among these are dominant narratives about the need to dominate or ‘win’ the new tech-driven arms race with China and defend liberal democracy against its enemies, the predictive power of big data, AI, and surveillance tools, and the overriding importance of experimentation and war as a ‘testing ground’ for technology development.

In the US, these narratives are further supported by familiar capitalist claims, which suggest that the central planning model of defense, based on a few large, expensive, heavily staffed military platforms, does not suffice and must be disrupted by free market competition between contractors providing smaller, cheaper, autonomous systems. Following the re-election of Trump, who maintains close ties with many of the new tech vanguards pushing for the disruption of the defense market, these logics will likely increase in the coming years, serving the interests of military startups and larger tech corporations.

Data Analysis and the Myth of ‘Perfect’ Knowledge

However, these companies do not only influence military culture and practice through popular narratives and sales pitches. As we argue here, they also exert influence via their software.

As the military produces increasing amounts of data generated by sensors built-in military hardware, passing through networked devices, and stored on different databases, the defense tech sector finds its competitive edge by developing software platforms for processing, sorting, and analyzing all this data. For example, Anduril is actively marketing Lattice, a software platform ‘built with a modular, open architecture that integrates a variety of data sources and sensors’ to enable ‘faster and better decision-making.’ By providing compelling user interfaces that connect data points across previously disparate systems, it promises military users a ‘simple, scalable, [and] extensible’ platform that ‘ultimately leverages machine intelligence to accelerate the closing of complex kill chains.’ 

Similarly, Palantir is promoting its own platform ecosystems, including its AI Platform (AIP) for Defense, which allows the military to ‘wield the power of LLMs [large language models] and cutting-edge AI.’ In an 8-minute promotional video from April 2023, Palantir shows how militaries may use AIP to fight an active war. The demo presents a fictional scenario involving a military operator ‘responsible for monitoring activity in Eastern Europe’ using a ChatGPT-style chatbot to make military decisions and operations. While Palantir ensures that it deploys its platforms in an ethical way, the video does not address any of the known limitations of large language models, such as unpredictability, lack of transparency, and (automation) bias. Nor does it reflect on the ethical and legal implications of using such technology in a military context. 

Meanwhile, the use of large language models for military purposes will only increase. In January last year, OpenAI deleted a prohibition in its usage policy on using its technology for ‘weapons development’ and ‘military and warfare.’ Its tools now support U.S. Africa Command, or AFRICOM, in its warfighting missions across the African continent. OpenAI has also signed a recent deal with Anduril to combine OpenAI’s models with Anduril’s hardware and software for detecting and shooting down drones. Meta, in collaboration with military startup Scale AI, is promoting its flagship large language model, Llama, as a convenient tool for military planning and decision-making. As reported by The Intercept, ‘Defense Llama’ has already raised concerns among experts about the ‘flawed,’ ‘worthless,’ and ‘irresponsible’ answers given in an online demo using the tool for airstrike planning. 

While we should be careful to distinguish between the promotional visions of these companies and the actual deployment of new military AI tools on the battlefield, the overall direction is clear: towards more datafication, automation, and the further systematization of killing in warfare. The software provided by companies such as Anduril, Palantir, and Scale AI – or developed in collaboration with them – is thereby crucial as they offer the broader infrastructure for collecting, fusing, and analyzing data and making it actionable. These tools, in turn, embed and fixate ideas of perfect knowledge, a fully transparent battlefield, and the desirability and ability of precision strikes. However, the realities of war on the ground, as we know, remain much different.

Platform Capitalism Meets Defense Tech

Defense tech is not just a provider of software solutions for data integration and analysis. In the words of Palantir CEO Alex Karp, the aim is to ‘platfomisize’ everything. This means offering flexible software tools that let third parties (the military, defense primes, or other start-ups) build their own applications on top of them. Anduril, for instance, has recently deepened its commitment to open-source, participative software with Lattice SDK, allowing ‘developers to build decentralized, edge-ready applications that integrate seamlessly into the Lattice ecosystem.’

Building these software platforms is a proven business strategy for Silicon Valley companies. The idea is to create network effects by getting more and more users on their platform so that the companies can scale. But, whereas many platform companies like Google typically try to attract as many users as possible to their platform through software services (whose data they can then monetize with targeted advertising), defense tech platforms stand out by their contracted software solutions for the defense sector. 

For example, Palantir has made a name for itself by ‘forward deployed software engineers’ who work with the US Department of Defense to train personnel and ensure its software platforms are tailored to the Department’s needs. Anduril, for their part, specializes in building military hardware or buying up start-up companies that already do, using their hardware as a ‘Trojan horse to get software in the [DoD’s] front door.’ In both cases, these business models are about getting as many defense agents as possible into their platform ecosystem.

By rapidly exploiting these network effects and scaling up this way, the most successful defense tech companies set out to disrupt the defense market in the years to come. Early movers such as Shield AI, Anduril, Rebellion Defense, and Palantir are rewarded by this model and raise the most funding from venture capital. They are now starting to compete with industrial defense companies, either directly by building the military hardware themselves or by partnering up and renting out software solutions for legacy hardware systems. Having access to large amounts of venture capital, these early movers can buy up and lock in competition. For example, in the past years Anduril has already made several acquisitions of defense tech start-ups with the intent to connect these technologies to its Lattice platform. 

Whether defense tech succeeds in breaking up the market dominance of the defense primes remains to be seen. However, if platforms in sectors such as social media (Facebook, X), the gig economy (Uber, Airbnb), and e-commerce (Amazon) offer any indication of where things might be going, it is that platform software architectures and business models have strong monopolistic and oligopolistic tendencies with clear implications for democracy and human rights law.

The implications of the ‘platform economy’ for the public sector are widely known and debated, but not so much yet for the defense sector.  For instance, lock-in effects make the military increasingly dependent on these platform companies, making it too costly to change software or migrate to another platform. Such dependencies raise questions about states’ ability to control these companies and align them with emerging governance frameworks such as the Responsible AI in the Military Domain (REAIM) initiative – if they are to survive the next Trump Administration at all. 

Indeed, the US military initially expressed reservations about these dependencies. However, the sales pitch of defending democracy efficiently and effectively with technology, and the responsibility entrepreneurs and investors have in its realization seems to resonate all too well at a time of geopolitical turbulence.

The ‘tragedy’ of the current geopolitical context offers military startups the perfect opportunity to promote their platforms. It also enables and encourages ‘the hubris it was intended to caution against.’ Earlier this year at the ‘AI Expo for National Competitiveness,’ sponsored by Palantir, Alex Karp insisted in a panel with Google CEO Eric Schmidt, CIA deputy director David Cohen, and former chairman of the joint chiefs of staff Mark Milley that ‘we are the peace activists.’ The drive towards platformization in the defense sector brings together two value narratives. The first is financial and pretty straightforward: war is incredibly profitable. On the other hand, the almost religious faith that peace – understood as the absence of war – can only be realized by readying oneself for the next virtuous war.

The authors receive funding within the framework of the Odysseus program from the Research Foundation – Flanders (FWO).