I’m a propagandist, I’ll twist the truth, I’ll put forward only my version of it if I think that’s going to propagandise people to believe what I need them to believe.
This is not a soundbite from a particularly ebullient moment in the hit television show Mad Men. These words were uttered by Palmer Luckey, the CEO of Silicon Valley’s hottest military technology startup. Luckey’s company, Anduril Industries, specialises in artificial intelligence-enabled systems, including autonomous weapon systems. With a valuation of US$14 billion, Anduril is one of the darlings of the defence startup scene and its newly emerging venture capital (VC) ecosystem where big promises, big bets and a tendency toward propaganda are a staple necessary for success.
The integration of artificial intelligence (AI) into defence programmes, let alone weapon systems, remains controversial. The UK Artificial Intelligence in Weapon Systems Committee has urged caution over procurement processes for AI-enabled weapons, yet – as is so often the case when it comes to Silicon Valley products – the development, procurement and roll out of AI defence programmes has sharply accelerated in recent years.
Founded only in 2017, Anduril has already been awarded multiple multi-million dollar contracts by the US Department of Defense (DoD), as well as the UK Ministry of Defence (MoD). Against the background of the ongoing Russia-Ukraine war, the war in Gaza and rising global tension, this may not seem a surprising development.
In my latest research on military AI, I identified that one of the key drivers of the accelerated procurement of military startup products, such as autonomous drones and other AI-enabled systems, is the influx of enormous sums of venture capital money and influence. These venture capital companies need defence organisations to adopt the technology industry’s ethos of speed and scale and the venture capital world’s appetite for risk and revolution. This makes these firms not only financial players but also political ones.
My research, published in Finance and Society, suggests that this trend toward shaping defence in the image of Silicon Valley, motivated by venture capital interests, is likely to become more pronounced and widespread. With this in mind, it’s worth looking more closely at the dynamics in play when venture capital sets its eyes on matters of life and death.
The emerging military financialisaton
The military AI industry and global defence spending are both booming. At current estimates, the global military AI market was worth US$ 13.3 billion in 2024, with a projected growth to US$35 billion in the next seven years. These numbers vary, depending on the market data services consulted, but they have been revised upward on a regular basis in the last 12 months. Global defence budgets have also ballooned against the backdrop of ongoing conflicts and a general shift toward militarisation in the last 24 months.
Global defence spending reached a record level of just over US$ 2 trillion in 2023. With US$ 877 billion, the US accounted for nearly 40% of global defence spending in 2023. The NATO alliance will be spending US$ 1.47 trillion in 2024. These are big, attractive numbers for big tech and finance companies with intentions to gain a foothold in the defence market.
The Insights section is committed to high-quality longform journalism. Our editors work with academics from many different backgrounds who are tackling a wide range of societal and scientific challenges.
Meanwhile, defence organisations are starting to spend more money on cutting edge technologies, including, inevitably, AI. A 2024 Brookings Institute Report found that defence contracts for AI-related technologies increased in value by nearly 1200% in the 12 months from August 2022 to August 2023.
For most new AI products, civilian or otherwise, some form of venture capital funding is often involved, especially if the AI venture in question might prove to be too risky to be funded through bank loans or other financial instruments. Venture capital is willing to take bets on innovation that other funders would be unwilling, or unable to take.
In the past two decades, this type of funding has primarily focused on Silicon Valley products for the civilian market, where the dynamics have allowed for extraordinary gains to be made for investors. But as the defence market is growing, and the opportunities for extraordinary venture capital returns in the commercial spheres wane, those with large amounts of capital to invest see a new opportunity for huge gains in defence within their grasp.
It is unsurprising, then, that in the past five years, venture capital investment in defence technologies has surged. From 2019 to 2022, US venture capital funding for military technology startups has doubled, and since 2021, the defence technology sector has seen an injection of US$130 billion in VC money.
Venture capital spending is also at an all-time high for the European defence sector; private VC investments are projected to reach a record US$ 1 billion, driven mostly by US venture firms. There is a palpable buzz in the air about the possibilities for VC backed endeavours and the possibility to reshape the defence landscape.
The venture capital-military-Silicon Valley nexus
Venture capital has always been connected to the military sector in some way. In fact, the current boom in venture capital defence investing could be seen as a return to its early days. The origins of venture capital are typically traced back to the American Research and Development Corporations (ARDC) founded in 1946, just after the second world war in which the US was buoyed by a victory achieved, at least in part, by cutting edge technologies.
ARDC was one of the first companies to systematically raise capital from institutional investors to finance start-up type companies with high potential but which were too risky for bank loans. With this approach ARDC was the first venture capital outfit to create investment portfolios which often relied on one or two extraordinary successes in order to offset the majority of companies which only made very modest returns or indeed losses. In this way, ARDC was the first so-called “unicorn” company.
Unicorns are young companies that receive a valuation of US$ 1 billion or more (up until recently an exceedingly rare occasion for a startup and something every investor covets in their portfolio). This is at the heart of venture capital investing: it is risk capital with potentially very high rewards.
In the early days especially, just after the second world war, many investments went toward supporting startups that would deal in military innovation and technologies. This brought about various analytical instruments, high-voltage generators, radiation detection technology, as well as early mini-computer companies, like the Digital Equipment Corporation.
The digital landscape, as we know it today, has its roots in the military. Innovations in communications theory were purposed for military missile technology in the 1950s, the grandfathers of AI almost all worked on mid-century military projects and even the internet itself emerged from a military project, then named Arpanet.
Read more:
How Britain got its first internet connection – by the late pioneer who made it happen
Many Silicon Valley firms remained entangled with the military sector over the decades and, as the anthropologist Roberto Gonzales has written, almost “all of today’s tech giants carry some DNA from the defence industry, and have a long history of cooperating with the Pentagon”. So, venture capital’s DNA is folded into this relationship.
But, it is worth stressing that traditionally it was the needs of the military organisations and the governments that largely dictated the pace, structure and process for technological innovations.
Now, the pace and focus for military technology and innovation is increasingly set by a progressively vocal and powerful technology startup industry and their funders who have unleashed a raft of “Patriotic capital” initiatives, such as American Dynamism; the Special Competitive Studies Project, Rebooting the Arsenal of Democracy and America’s Frontier Fund. These enterprises were conceived by a handful of prominent companies and individuals in the new defence tech domain to shape defence and military priorities and make good returns while doing so. Propped up by vast amounts of venture capital, unicorn companies are proliferating in the defence space, including new military technology unicorns like Anduril Industries, Shield AI, Skydio, Scale AI and Palantir (Palantir is technically no longer a startup since it went public in 2020, but it is still part of a cohort of new military technologies).
This is a recent development. In the two decades from the mid-90s to 2014 the venture capital sector focused its efforts on a thriving civilian technology landscape, where the sky was the proverbial limit for returns from technology startups like Google, Microsoft, Facebook and PayPal.
The defence market, in contrast, was considered mature and consolidated, with strict acquisitions rules and regulations, and too little opportunity for outsized returns on investments. For a government contract to come to fruition, it would often take many years. Defence was also dominated by a handful of key industry players – the so-called primes which include Lockheed Martin, RTX Corporation, Northrop Grumman, Boeing, General Dynamics and BAE Systems.
These primes divvied up the lion’s share in the defence market between themselves and there seemed to be little opportunity for technology startups to get a foot in the door without major efforts. For example, companies like Space X and Palantir sued the US Airforce and US Army in 2014, respectively, for the opportunity to bid for certain contracts. Using the law to break open defence for military startups has since become more widespread.
In addition to these structural hurdles for VC investment in the defence sector, there was a greater nominal moral cost associated with the idea of profiteering from war. Since venture capital investors are often endowments, foundations, insurance companies, universities and pension funds, there was an outward reluctance to be seen as investing in “a defence portfolio” – or in other words, in instruments of death. European venture capital investors were particularly cautious.
However, the speed with which these trepidation seems to have subsided in less than a decade, is remarkable, suggesting either that the investors propping up venture capital firms come from different backgrounds which might have less hesitation when it comes to benefiting from the business of war, or, that it was always primarily simply a matter of maths rather than morals.
Unicorns and hypergrowth
Today, everyone wants to invest in a unicorn because it has the potential to skyrocket in valuation.
But in order to get a foot in the door with an unproven product or concept, some startups can be motivated to make big, bold claims about the revolutionary, change-making nature of their products. And even once a company has secured funding, the ethos of overpromising often remains enshrined in order to sustain success toward hypergrowth.
In the worst case scenario, overpromising is done at such scale that it amounts to criminal fraud, as it was the case with the notorious blood testing startup Theranos, which went from being one of the most exciting healthcare startups, valued at US$ 10 billion at its peak in 2015, to a complete bust in four short years.
In the Theranos case, the company’s charismatic founder had wildly overpromised the technology’s capabilities, claiming that it would enable a whole raft of tests which could be done from just one small drop of blood. This ground-breaking technology “could revolutionise medicine and save lives the world over”.
It was a future-oriented promise – the technology could not do what was promised yet – nonetheless the company claimed to already have a functioning testing device, which turned out to be a lie. Theranos folded in 2018 and the charismatic founder, Elizabeth Holmes, went to prison.
Selling a fantasy
There are many other, less dramatic stories that play out in a similar, although not fraudulent way: companies that promise to revolutionise the way we do mundane things with ground-breaking technology, which turn out to be unsustainable, unworkable, or simply fizzle out.
But the upshot is that investors lose money, and, more importantly, that people who have come to rely on the technology’s promise come to harm.
In the defence context, the promises of new military technology revolves around selling powerful deterrence, of protecting democracy, of being able to have comprehensive, accurate, real-time knowledge, of a fully transparent globe, and, first and foremost of a clean, swift and decisive victory with smooth and effortless connectivity.
This can foster, at worst, a fantasy of omniscience and omnipresence and, at best, it stokes a desire for an impossible revolution in warfare that is too attractive to resist and ultimately draws in an ever wider audience into its wake. These narratives are often underwritten by a general hype that a future with AI is inevitable. This makes for a powerful storyline which mythologises and valorises a technology that may never deliver what is promised. It is a potent mix that often resists more sober voices that urge caution.
The claims made by defence unicorns may often seem plausible, but they are usually unverifiable, because they address the future. And often that future reflects a vision shaped by fiction and science-fiction, which is always some degrees removed from the social and political challenges of reality.
This temptation of overpromising and the mythologising of possible technology is shaping programmes that work toward realising global transparency and global reach at speed. The Joint-All-Domain Command and Control (JADC2) programme is one such effort initiated by the Pentagon. It aims at connecting all domains – land, air, sea, space and cyber – into a single network for “predictive analysis” and “high-speed battle”.
To make the programme palatable to Congress, JADC2 is often likened to the ride-sharing platform Uber, promising seamless interaction between systems and platforms for speedy interventions. This refocuses attention on AI as an infrastructural necessity for all military assets and platforms. Without expanding military AI, this vision will be impossible. It is here where the opportunity for military startups resides.
Two prominent military tech companies are contractors for JADC2 – Anduril and Palantir. Both companies make little secret of their ambitions to disrupt the defence sector, unseat the current primes and carve out a monopoly slice of the market in order to secure increasing gains.
Palantir has set its eyes on “becoming the central operating system for all US defense programs”; Anduril has declared that it will be going “after everything that’s on the [Defense Department’s] list” in order to dominate in the sector. For both companies, this is the battle – the battle for growth.
As Anduril’s Luckey says: “you have to fight and win across multiple areas”. (He means that in terms of corporate strategy, not actual battlefields). Similarly, CEO and co-founder of Palantir, Alex Karp, acknowledged that, in order to break defence as a market wide open, he is proud to “have dragged and kicked and cajoled and humiliated” various lawmakers, policymakers and government to help further this goal. Move fast and break things.
Making a unicorn requires a concerted effort, and an aggressive posture, on the part of those who stand to gain the most, financially, in this domain. This is best done in alliance with like-minded others. In the current defence venture capital landscape there is a close entanglement of founders and funders.
Peter Thiel, for example, is co-founder of Palantir, he also runs the Founders Fund VC outfit which has investments in Space X, Anduril and Scale AI, among others. The VC company Andreessen Horowitz also funds SpaceX, Anduril, Shield AI and Skydio. The managers of these VC companies have long-standing ties with one another. Similarly, there is interlacing between companies. Anduril, for example, was started by former Palantir employees who took their experience from Palentir and applied it at Anduril. Palmer Luckey, formerly of Oculus Rift, was installed as its charismatic and outspoken CEO.
Peter Thiel and Eric Schmidt (formerly CEO of Google and Chair of the US National Security Commission on Artificial Intelligence) are investing in the America’s Frontier Fund, and so on. There is a tightly knit and very well connected network of financiers and startups that all work to double down on the key driving message: the defence sector is in need of disruption and we are the ones to shake things up.
At a recent panel giving evidence to the US Armed Services Committee were representatives of five military startup companies. Every single one of the five was either funded by the VC firm Andreessen Horowitz or otherwise affiliated with the firm.
At the US Armed Services Committee hearing, Palantir’s Chief Technology Officer, Shyam Sankar, gave evidence advocating for “more crazy” and for “letting chaos reign” in the military acquisition and procurement process, so that the necessary incentives can be fostered for innovation through inter-departmental competition.
Regulatory limitations, he thinks, “constrains you to oversight” and he “would gladly accept more failure if it meant that we had more catastrophic success”. What kind of success this might be, or what the implications are for failure, remains unaddressed but it is clear that Palantir’s CTO speaks with a venture capital logic in mind. And, according to a recent US Defense Innovation Board report, it seems the government is ready to embrace more risk and provide top cover for such “mavericks”.
The ‘crisis’ narrative
Besides cultivating startups with high potential, there are a number of ways to bend the defence sector to the needs of Silicon Valley contractors and their VC backers. The power of narratives goes a long way here too. Venture capital managers and their startups often pen high-profile op-eds in which the poor state of (US) defence is lamented, in which the need for accelerated innovation is emphasised, and in which the possibility that the US might “very likely” become embroiled in “a three-front war with China, Russia and Iran” is conjured up. In short, a story of urgency is spun which helps valorise those companies that are seen to address the imminent crisis.
A second pillar in the structural overhaul of defence is to employ an intricate network of former government employees who serve either as lobbyists or as advisers with close links to government.
Former Republican Congressman Mike Gallagher, for example, became Palantir’s Head of Defense operations in August 2024 and former National Security Advisor H.R. McMaster is now Senior Advisor to Shield Capital. There are many more such “revolving door” moments in which credible experts lend their authority to the new startups. The military tech startup scene, like most Silicon Valley creations, holds a certain reputational cache and the money is attractive too.
Anduril, having learned from Palantir, hired a slew of lobbyists in the first week, spending more money on “lawyers and lobbyists than engineers” as Luckey noted in a recent interview with The Economist.
With this, Anduril adopts a relatively traditional way of shaping the defence landscape, which is also employed by prime defence contractors which have, as the Anduril acknowledges in a 2022 blog post, an incentive “to spend heavily on teams of lawyers and lobbyists to shape program requirements in line with the company’s existing technology”.
Anduril, and its backers, are now doing the very same, tailored to their own suite of technologies. The lawyers are often employed not only to oversee mergers, acquisitions and partnerships, but also as a way to use the law as an instrument to force reform.
The primary goal of the SpaceX and Palantir lawsuits against the US army and airforce which I mentioned earlier was not necessarily to win (Space X lawsuit was not successful, Palantir’s was), but to pry open space for acquisitions overhaul and both lawsuits achieved just that.
A strategy of creating a sense of urgency, doubling down with lobbyists and creating the structural possibility for a defence overhaul is now well underway. To be clear, I am not arguing that the defence sector would not benefit from modernisation or restructuring. Nor am I arguing that all military startup products are irrelevant or unsustainable. I am also not seeking to pit the primes against the new venture capital dynamics and their focus on growth.
But what I think is worth examining is the dynamics at play with these new companies and their implicit priorities and interests, because they will shape practices and priorities. And where disruption is at work, some level of breakage is to be expected. And this takes on a different tone in matters of life and death.
Disruption debris
The defence sector disruption is well underway and efforts to shape it in the image of Silicon Valley have borne fruit in recent years with a number of concrete results. The JADC2 programme mentioned earlier is one. Others are evident in programmes like the US Department of Defense’s Replicator Initiative, which incorporates the aims, timelines and products that Silicon Valley military startups have to offer.
High-level defence officials are repeating the talking points of the venture capital industry and various acquisition programmes have adjusted to accommodate the speed and scale needed. These companies have the ear of policymakers and the demands for a quasi-spiritual “Defense Reformation” are finding a growing audience.
So what are the possible consequences?
When Uber disrupted the private transport industry, it left in its wake a raft of eroded labour laws, worker’s rights and healthcare provisions for drivers. When AirBnB shook up the accommodation industry, it resulted in increased rental prices in popular tourist destinations. When you try to create a monopoly, there are always social and political consequences. Often these consequences are foreseeable, sometimes they are not.
Disrupting the defence acquisitions process comes, at the very minimum, at the expense of greater oversight to the acquisitions process. The technology sector is not known for its appreciation of regulatory boundaries. Quite the contrary. Some of the most prominent funders of the new military startup landscape are most vocally opposed to any type of regulation.
VC heavyweight Marc Andreessen, for example, famously penned a Techno-Optimist manifesto in which he names risk management, trust and safety measures and the precautionary principles as “the enemy”.
Less regulation means less oversight and accountability, not only in spending but also in how and where certain technologies are used and with what effects. This much is evident.
But there are many other highly plausible consequences we might foresee with the accelerated acquisition and rollout of military technologies for battle. One is the refocusing on risk and experimentation.
The current crop of military startup technologies, like AI-enabled drones and AI decision support systems, are being tested and improved live and during ongoing conflicts, such as in the Russia Ukraine war, but also in Gaza. This is a form of prototyping which is becoming increasingly prominent and which needs an active battlefield for effective testing, iteration and optimising of the technologies.
This also means that it is possible that technologies will be fielded that are not fit for purpose, only to test them and improve as you go along. It normalises, if not promotes, the launch and sale of flawed and possibly inadequate AI products, which will inevitably cause harm to innocent civilians caught in the crosshairs of conflict.
We can observe this right now with the push by technology companies to sell their large language models to military organisations. Scale AI, for example, has teamed up with Meta to sell an LLM product, Defense Llama, for defence purposes. The company says that human involvement is “absolutely necessary” for the system.
But given the well-known fact that LLMs are prone to what are known as hallucinations, the chances that such technologies will work exactly as advertised are slim for a context so complex and dynamic as warfare. This could cause harm to those caught in the middle of this experimentation, fine-tuning and live testing.
It is a key concern that the technology might not be suitable for the unexpected, for the less calculable or less foreseeable elements in warfare. That includes potential emerging terror threats or moves by those states that are often considered as irrational; like North Korea, for example.
Anduril CEO, Luckey, admitted as much in the interview I opened with. He acknowledged that the logic on which his weapons are built falls apart with potential enemies that eschew the game’s theoretical approach on which much of the AI logic for defence rests: “It’s very hard to engage in game theory with people who pursue the non-game theory optimal strategy…It’s like playing monopoly with the person who is going to drop out and give all their money to somebody else.” A serious limitation for something so riddled with chance as warfare.
There are also second and third order effects that emanate from this shift toward venture capital logics. By conjuring up an imminent threat, the broader global risk and security landscape might change; by prioritising weapons technologies, funding for other ways to address conflict might be curtailed, by dedicating an increasing amount to technologies that remain untested and that may not have permanence, significant amounts of money which would be better allocated elsewhere might be wasted.
But this is a land of make-believe and unicorns, where such considerations are as speculative as the much-hyped promises of AI weapons as the defenders of democracy.
In Silicon Valley the “move fast and break things” motto implies that problems that arise in the roll out of the tech can always be addressed and solved later. In the world of defence and war, the harm produced by this kind of risk-taking cannot so easily be undone.
The Conversation put the points raised in this article to the tech and venture capital firms named. They did not respond to our request for comment.
For you: more from our Insights series:
To hear about new Insights articles, join the hundreds of thousands of people who value The Conversation’s evidence-based news. Subscribe to our newsletter.
Source link
lol
