State of AI outtakes #1: scaling in defense
The State of AI Report aims to provide a comprehensive overview of everything you need to know across AI research, industry, politics, and safety. To ensure the report remains at a manageable length, lots of material doesn’t make it into the final version. We’re bringing Air Street Press readers some of the research that didn’t make the original cut, subsequent follow-up work, and our reflections.
Introduction
Exactly a week after the State of AI Report landed, we were in Madrid for the European Defense Tech Summit, organized by our friend Eric Slesinger of 201 Ventures. Alex kicked off the day, hosting a panel diving into the challenges founders and investors face in building and scaling defense tech companies in Europe, where he was joined by Jonathan Turner, a partner at Accel (which invested in Helsing’s Series C) and Lorenz Meier, co-founder and CEO of drone autonomy company Auterion.
The panel touched on a number of the defense-specific themes that we brought out in this year’s report, including the pains of defense contracting, the need for courage among investors to bet ahead of government contracts landing, as well as the importance of spending time on the frontline to understand your end user’s needs.
The path to scale
In last year’s report, we covered the record level of investment being pumped into US defense start-ups. Anduril, the clear leader amongst new defense companies, has gone from strength to strength since then. The company has begun to win serious contracts in the US, including $250M to deliver advanced air defense capabilities for the Pentagon.
In August, Anduril delivered the first Tactical Intelligence Targeting Access Node (TITAN) to the US Army, as part of a $178M contract, in partnership with Palantir. Anduril also delivered its first Ghost Submarine prototype to the Australian military on budget and a year early - unheard of in the world of defense procurement.
But Anduril with its $3.7B warchest from investors is unusual. It’s too early to judge whether the ecosystem is likely to broaden. There are potential bright spots, like Saronic, a US-based manufacturer of unmanned surface vehicles. But the company, as far as we can tell, is yet to move beyond small pilots to larger contracts.
These scaling challenges were a primary theme of Alex’s panel in Madrid. Jonathan from Accel emphasized that one of the reasons Helsing had raised so much so quickly was that the next generation of winners couldn’t simply wait for government contracting cycles to work themselves out. Only a small handful of winners were likely to prevail and this would require courage from investors.
He was also bearish on the prospect of M&A coming to early-stage defense companies anytime soon, with national governments wary of foreign acquisitions and the pool of potential acquirers in most European countries being small. This is a dynamic we’re already seeing play out in France. Preligens, for example, was bought by Safran only after the French government announced that it would bar an overseas buyer. We see M&A as part of the solution, but also believe that defense primes will only spend money on it when there’s an incentive for them to do so. While procurement remains broken and competition lackluster, they can continue to focus on dividends and buybacks at the expense of R&D.
There’s a stark contrast between this sluggishness and the speed with which Ukraine is moving to procure technology for use on the battlefield. Alex Bornyakov, from the country’s Ministry of Digital Transformation, talked about how contracting cycles had fallen from 2-3 years before the war to 21 days today, while the gap between a manufacturer presenting its equipment for testing to being certified is 3-4 months.
This is crucial when rapidly-evolving electronic warfare means that equipment usually has a shelf life of about six months on the battlefield.
Western start-ups in Ukraine
Our other main point around defense concerned the fate of US start-ups, whose enthusiasm for the Ukrainian cause did not always translate into convincing results on the battlefield.
It appears we’re now moving into a new phase. The focus has shifted from attempting to equip the Ukrainian military with drones made overseas, in favor of boosting the domestic Ukrainian industry. At the time of writing, the US is thought to be preparing an $800M aid package.
This makes sense for a few reasons.
Ukrainian teams have a combination of proximity to the frontline and technical ability. Wahid Nawabi, the CEO of US-based Aerovironment, has acknowledged that “Ukrainian drone companies in many different domains are going to be a global, legitimate player”.
However, this technical talent often struggles to attract international VC investment; according to Brave1, foreign investment in Ukrainian defense tech start-ups over the past year amounts to a mere $9M. VCs aren’t used to investing in companies whose facilities can be blown up, especially when the start-up concerned can’t secure any wartime insurance. The other blocker is the Ukraine Central Bank’s capital controls, which make transferring money out of the country a logistical nightmare.
Finally, US-based start-ups simply aren’t able to match Ukrainian start-ups on price. As well as lower operating costs, Ukrainian companies are able to create ‘good enough’ products tailored to the needs of end users.
Anduril is charging Taiwan close to $1M a unit for its Altius 600 drones, which are propeller-powered and weigh 45 lbs. The costs are also stiff for more established contractors. The Switchblade 600, which the US is purchasing via the Replicator program, comes in at over $100k a unit. While these costs are low in the context of US military expenditure, they’re unworkable for a cash-strapped Ukrainian government.
Software-only procurement may not be the norm for many countries, but AI-first software companies are gaining some traction in Ukraine by partnering with local manufacturers. While many of these companies don’t need help building hardware, they often have less in-house AI expertise.
In Madrid, Lorenz spoke about the team Auterion has built in Ukraine, as well as the speed with which they’d been able to navigate Ukrainian procurement versus other European countries. This local team, whose software helps drones lock onto targets to counter the effects of electronic warfare, has grown to over 20 people. Similarly, the company has scored a number of contract wins in the US with the DoD, while remaining stuck in negotiations in Germany, their home market.
Helsing also appears to be partnering with local manufacturers, but as ever, their work is shrouded in a degree of mystery (potentially on purpose).
Closing thoughts
While the defense world looks like a maze to outsiders, the underlying principles are straightforward. It’s possible to circumvent a broken market, but only if you have the balance sheet to outlast it. Similarly, it's possible to charge the US or European militaries bloated margins if you’re part of an oligopoly, but don’t be surprised when a cash-strapped wartime government would rather turn to local innovation to get the job done.