The Long Read: At Haraka Storm, failure is a ‘feature, not a bug’ for Stark
We spoke to over a dozen people familiar with the British Army training exercise in Kenya, following a news piece singling out Stark's Virtus drone perceived lacklustre performance
It’s a story making the rounds with all the right buzzwords. Drones. Startup. Peter Thiel. Crashed. Specifically, one of the hottest defence tech startups in Europe flamed out during an important military exercise using its tech. Militaries are rebuilding, and they want to tap into the latest defence tech to stay ahead of their adversaries, but being at the leading edge sometimes becomes being at the bleeding edge.
But what really happened at Haraka Storm?
At the British Army’s annual training exercise Haraka Storm in Kenya, among troop manoeuvres the military tests equipment ahead of buying and deploying it, including vehicles, weapons, and technology from third parties. This year, the list included three of the biggest startups currently operating in defence tech in Europe: Anduril, Helsing, and Stark. As it turned out, the results were not quite the slam-dunk everyone was led to believe it could be.
Stark — founded in 2024 — brought four drones to Haraka Storm and launched two. Both missed their targets “considerably,” according to one person who spoke to RM off the record. One apparently saw its battery burst into flames mid-air.
“A disaster” is how one person described it to the FT, in an article that has been widely shared and has triggered a lively debate across the sector.
Critics say this reflects badly not just on Stark but the whole startup ecosystem, which has been hoping that the military’s plan to upgrade and update could bring opportunities for them, too.
Others strike a different note, pointing out that failures have been an important part of the iteration process in defence. The key is to manage expectations and keep working.
Resilience Media spent the last few days talking to more than a dozen people directly involved with the Haraka Storm. We also talked to defence tech investors and industry insiders to help make sense of all this. Unless directly quoted, all sources requested anonymity due to sensitivities around the topic or because they were not authorised to speak on behalf of their organisations.
The background
In October, the British Army ran Haraka Storm, an annual training exercise in Kenya that has been running for several years. Drone makers Anduril, Helsing and its partner ARX Robotics, and Stark were among those invited to participate last month. Designed to be a test of both equipment and personnel, the participants faced 35 degree heat at the equator, torrential rain that created river-like roads, and flying termites. The equipment was supposed to face rigorous testing in the hands of military end users.
Why Kenya? Airspace in the UK and Europe is largely unavailable for testing due to the potential risk to nearby populations, so startups either have to be invited to demonstrate their capabilities by a Ministry of Defence that does have dedicated, remote test sites, or ship their tech to a battlefield to understand how it performs under pressure. These days, the battlefield used for testing by all major and minor defence tech startups is the frontline in Ukraine’s war with Russia, now into its third year.
A source tells us that the drone exercise at Haraka Storm was partially intended by the military organisers to inform decision makers in the British Army about the technologies they should be deploying on the Eastern Flank. Those who designed the drone exercise believe that loitering munitions (LM) should form a core part of the arsenal deployed on the Eastern Flank. The intention was to counter prevailing doctrine, which is still quite archaic compared to tactics in Ukraine, and argue for greater use of drones by demonstrating their potential.
Trying to effect a paradigm shift of this kind is no small effort. Although the UK has celebrated drones as “game-changing” technology, at the highest levels (and elsewhere) others have been significantly more skeptical and suspicious of hype.
The ups and downs of loitering munitions
For Stark, and the other startups in this space, developing European loitering munitions has been a journey. Helsing and Anduril are prime examples of the uphill battle startups face in bringing reliable technology to the battlefield. Most urgent today is the Ukrainian frontline, which requires mass-producing the devices to the tune of around 100,000 units per month.
Founded in 2021, Helsing’s first strike drone, the HF-1, had teething problems when it was deployed in Ukraine. But further testing, iteration, and a boatload of venture capital that valued Helsing at €13 billion led to Helsing releasing its second drone, the HX-2. The HX-2 is a product that militaries do want to use, and that has led to a number of wins. Helsing is now a strong contender in the British Army’s Project ASGARD procurement program; it supplies thousands of the drones to Ukrainian Defense Forces; and – alongside alongside Stark, Rhinemetall and others – has just been awarded a part of a €900 million defence contract from Germany’s land army, the Bundeswehr, to defend NATO’s Eastern Flank.
Like Helsing, Anduril, the US-based defence startup that was founded in 2016, also experienced early stumbles in Ukraine, specifically facing challenges арround electronic warfare. It has, however, returned to military effectiveness with the launch of its Altius drone, and Anduril currently has a contract with the UK Ministry of Defence.
Loitering munitions companies currently performing well on the Ukrainian frontline include Ukrainian company Iron Belly, which recently signed a partnership with ShieldAI, and Buvala by Deviro, which is among one of the biggest companies in Ukraine.
Failure is a feature, not a bug
Notably, the individuals we spoke to, including Stark’s competitors, defended Stark’s display at the test in Kenya. This is not a surprise: it’s still a young market, so the success, or failure, of any single drone company has ramifications for all of them.
Furthermore, failure, as many have noted, is a very important part of product development.
“If you are not experiencing failures then you are likely not pushing innovation hard enough,” one source from the UK Ministry of Defence told RM. “It’s refreshing to see actual range tests compared to claims of success that are not always easily verified.”
Militaries need to be comfortable with this facet of technology development, said one founder. They “need to get to the point where [government] contracts get appointed on risk,” continued the founder, who spoke to RM on the condition of anonymity. “The end customer has to start taking risks.”
Ukrainians are more open to working with a higher degree of risk because of the climate they operate in.
“Nothing [that arrives on the frontlines] works,” Alex Kinash, European partnerships director at Vermeer said on stage at Resilience Conference 2025. He said this is because the pace of innovation is so fast that even if the technology is not good enough on arrival, it is adapted rapidly in battle.
“We want these companies to be pushing the limit,” said Samuel Burrell, a partner at Expeditions, a VC fund (which does not back Stark). “If you’re not failing, you’re not going fast enough.”
Most startups do not want to replace the incumbent major defence contractors, so-called “Primes.” But founders reasonably worry that if defence ministries are not comfortable with ‘lean startup,’ iterative approaches, then Primes might become the only game in town for technology development. This potentially means longer product lead times and bigger bills for our militaries.
Not everyone is concerned about this risk to startups, where primes are concerned. One person close to the drone industry and the UK Ministry of Defence described the story about Haraka Storm as “sensationalised” and added that it won’t present a serious threat to the acquisition of startup drone technology by the British Army in the longer term.
If startups are participating in these exercises, they are bringing a prototype and expecting that things will go wrong, the source continued. Plus, the source added, being a startup is immaterial to the outcome: at this level of development, a Prime might experience the same issues, though these might only be seen – and fixed – within the privacy of its own testing centre.
“No single company will solve Europe’s security challenges alone,” Phil Lockwood, Stark’s managing director, said in response to the Haraka Storm test. “It takes an ecosystem — operators, governments and industry working side by side, testing in real conditions, learning quickly and strengthening one another. In the spirit of NATO’s principles of collective defence and interoperability, we collaborate widely, working with partners across programmes, joint experimentation and shared development. We need to all roll our sleeves up, make mistakes and get messy on the testing and training grounds, or we will not be prepared to defend ourselves.”
Forget the hype. Just manage expectations
A source with expertise in the drone industry told us that he thought Stark’s failures in Kenya won’t necessarily push away government contracts, because this is a sector where people are already comfortable working with risk. However, both the military and the tech companies may need to align their expectations more constructively. Whereas Primes are expected to deliver finished products after a completed R&D and development cycle, startups typically field betas and MVPs to gain feedback for an iterative innovation process. These are two different worlds, but only one is familiar to the military.
Startups hyping their products is normal in the tech sector, especially when they are VC backed. This source suggested that hype may not be such a good fit in defence, where the stakes are infinitely higher than in other industries.
This concern about over-selling new technologies was raised by several investors, who spoke off the record. Some backers are driving huge rounds and valuations, encouraging founders to overpromise on capabilities but then risk underdelivering. Where founders are concerned, “aggressive vision is great, but startups need to concentrate on how they can deliver,” said Nicholas Nelson, a general partner at Archangel.
“When you underperform in defence… we should be a lot less forgiving about that,” he continued. “Because of the critical nature of this… the idea of ‘fake it till you make it’ only goes so far and actually can be quite dangerous.”
In Ukraine, soldiers on the frontline might use tech they know works to 80% effectiveness (a percentage that Ukrainian military people have shared with us), or they might test something new in hope that it’s more effective. These are never simple trade-offs and choices: our Ukrainian military contacts have reminded us that if new tech does not work as well as the existing option, people can be killed.
Messiness is a catalyst for opportunity
The key appears to be not whether or not a technology fails, but how those failures inform innovation. Does it manage expectations, and can it identify and fix what isn’t working? What might appear like a catastrophic failure can catapult a startup into a wildly unexpected success.
A recent example of that is Quantum Systems. When the company deployed its Vector reconnaissance drone in Ukraine, it had teething problems: pilot feedback loops, data-flow reliability, and electronic-warfare vulnerability. The company publicly acknowledged these, and the early failures were what led Quantum Systems to fix these issues and build further features that actually worked well. The company went on to become one of the top international performers in Ukraine. Now Quantum Systems is one of the three defence tech unicorns in Europe.
It’s notable that Stark has a history with Quantum Systems. The two companies share a founder and Stark was originally built using some of Quantum Systems’ technology before it was created as a separate company. Some see the connection as encouraging for Stark. One defence insider predicts it will follow the same playbook as Quantum Systems did after their initial technical challenges.
“Now the Stark team is going to be aggressively focussed on Ukraine and fixing [Virtus’ performance],” the source said. “They know how to do that.”
Indeed, one word that came up a lot with regard to Stark’s performance in Haraka Storm was “mess.” But Phil Budden, a senior lecturer in Technological Innovation, Entrepreneurship, and Strategic Management at MIT, believes that a certain amount of messiness is to be expected with startups.
Budden continued: “Not everything is going to succeed the first time. And it reminds me, given how dependent we are on Starlink, for example, the early failures that Elon Musk had with SpaceX getting the Starlink satellites up there.
“And now people look around and say, oh, my goodness, it’s terrible how dependent we are. But they forget about the early stages of the story where people said, well you can’t even get a satellite up.”
What now?
On the surface, Haraka Storm has been a discussion about money, investors, and highly capitalised startups. Perhaps it should mainly be a discussion about the urgency of war.
Europe needs startups like Stark, Helsing, and other defence tech companies to succeed because Europe faces an imminent threat from Russia. If there is an attack against NATO, as the recent drone incursion in Poland showed, Europe is completely unable to cope with a Russian drone swarm attack. If Russia fired hundreds or even thousands of small, cheap drones into a European country we’d be unable to counter it at all.
So if Stark’s credibility ends up damaged, that potentially weakens the wider startup ecosystem. That could discourage more private capital to get invested in the most innovative builders.
Critically, Stark’s backers seem to know the score on this – or at least for now they are willing to take on some teething problems in the spirit of growth.
“Failures are a feature of innovation,” said John Ridge, chief adoption officer, at the NATO Innovation Fund. “We should expect situations where innovative start-ups testing novel products run into problems. The litmus test is whether they can then learn from these situations to build a better version. Given the security challenges we face today, innovating safely and at the pace of conflict is a necessity - not a choice.”
Reflecting the extent to which the failure of two drones in a field test in Ukraine has triggered a storm of comment, Linkedin posts, and sensational media coverage, Sam Burrell, pointed out that “it takes time for MoD and journalists to understand how [the industry] works, the teething problems we experience as we move into a new era of product-defined defence tech companies iterating fast and sometimes failing [is not fatal].”
It remains to be seen now whether Stark — one of Europe’s most capitalised defence tech companies — can turn this failure into the sort of learning exercise that has informed the innovation process of companies like Helsing and Quantum Systems. Perhaps this recent storm will also challenge how startups, and their investors, working in defence approach marketing their capabilities to adapt to a more complex market with far higher stakes.
Now we watch.
 | A guest post by
|