Helsing’s Sovereign Software Bet

On May 28, 2025, a Saab Gripen E fighter flew over the Baltic Sea and handed long-range maneuvering to an AI system built by a four-year-old German startup.

A human pilot was still in the cockpit, ready to intervene. But during the test, Helsing’s Centaur agent took control of beyond-visual-range (BVR) maneuvering, recommended missile shots against a Gripen training aircraft, and avoided flight paths that would put the aircraft at a disadvantage. 

According to the companies responsible, this was the first time an AI application had been “in charge of real-world maneuvering” in that way.

The symbolism of the flight was hard to miss.

A European fighter jet, built by one of Europe’s major defense primes, being flown by software from Helsing, one of Europe’s most prominent defense start-ups, over the Baltic, a region whose defense is critical to European security against threats from the east.

It was a glimpse of the future Helsing says Europe needs to build: software-defined, AI-enabled, and sovereign.

Warfighters operate inside a fog of sensor feeds.

War is getting faster, more saturated, and too data-rich for human-only systems.

The problem is no longer a lack of information; it is a superabundance of it. More than humans can absorb, interpret and distribute at operational speed.

Niklas Köhler, Helsing’s chief product officer, described the current state bluntly to Wired: “Now, all of this is done manually: phone calls, reading things, drawing stuff on maps.”

Understanding how many systems are present, what they are doing and what their intent might be, he argued, is “an AI problem.”

That is one half of the shift. The other is autonomy.

Drones, autonomous aircraft, and unmanned maritime systems are moving from the margins to the center. Ukraine has made that transition stark. 

Modern forces need systems that can sense, classify, navigate, and act at machine speed. And they need them in numbers large enough to survive attrition: capable, cheap enough to lose, fast to replace. 

They also need those systems in numbers large enough to survive attrition. The future is not only exquisite platforms at the edge of physics. It is also systems that are capable, inexpensive enough to lose, and fast to replace.

For Europe, there is a third problem: sovereignty. 

The United States has produced software-first defense companies such as Palantir, Anduril, and Shield AI. Europe has long had formidable defense primes, but it has lacked a visible AI-defense pure-play operating at scale. 

Software is the core of the defense industrial base. If Europe can’t build an AI layer into its military, it risks dependency on others for the systems that determine what it can see, decide, and do.

Helsing’s pitch is that democratic militaries need an information edge: software that can process overwhelming sensor inputs, support decisions and execute on the battlefield, including inside autonomous systems.

Helsing was founded in Munich in 2021 by three figures with very different but complementary backgrounds.

Torsten Reil, a video game technology founder whose experience shows in Helsing’s focus on intuitive operator interfaces. 

Gundbert Scherf, former adviser to Germany’s defense ministry with direct knowledge of European procurement.

Niklas Köhler, a machine learning specialist who came from medical imaging and later told Wired that detecting drones is, methodologically, not so different from finding cancer in CT scans.

Their timing was almost uncomfortably precise: 

Before then, Helsing’s early backers had drawn criticism. Spotify founder Daniel Ek became Helsing’s chairman, and his fund Prima Materia led the company’s Series A in 2021.

This triggered backlash from some users who objected to money connected to their subscriptions flowing into defense. 

After February 2022, Ek’s position looked less controversial and more like an early read on Europe’s security environment.

More capital followed.

By 2026, Helsing had raised roughly $2.8 billion across multiple rounds, including a €600 million round in 2025 at a valuation of about €12 billion and a reported 2026 round that will value the company around $18 billion.

Helsing now has offices in Munich, London, Paris and Berlin, and more than 900 employees. Saab is both a shareholder and a partner.

In June 2025, Helsing acquired Grob Aircraft, a German aircraft maker, bringing more manufacturing capacity for its CA-1 Europa aircraft program.

Still, Helsing’s identity is not that of a traditional hardware prime. It is self-consciously software-first.

The hardware matters, but mostly because software needs vehicles, sensors, factories and battlefield feedback loops before it becomes military capability.

The easiest way to misunderstand Helsing is to list its hardware programs and conclude that it builds drones, underwater gliders, aircraft and robots.

It does build or back those things. But the hierarchy matters. Helsing’s center of gravity is software: applications that can support combat and manage information more effectively than humans.

The hardware is the casing around that software.

Centaur is the clearest example. Helsing describes it as an autonomous air-combat AI, trained through large-scale simulator self-play. Before the Gripen flight, the system had reportedly accumulated the equivalent of 50 years of human pilot experience in simulation. 

In the Baltic test, Centaur took control of the aircraft and demonstrated just how effective that software could be. 

Air combat tests both human and AI pilots in an unforgiving environment: limited time, incomplete information, complex adversary behavior and high consequences.

Helsing’s claim is not that humans will disappear. It is that AI systems will, more and more, perform parts of the decision and control loop once reserved for humans.

Altra brings Helsing’s software to land warfare. It connects ISR drones, spotters, artillery, strike drones and other effectors into a common targeting picture, using edge AI and resilient networking to move faster from detection to engagement. 

The next step is hardware, not as the center of Helsing’s identity, but as the means of getting its software into the fight at scale.

This starts with the HX-2 Strike Drone. With a 100 kilometer range, its onboard AI can search for, identify, and engage targets without a continuous datalink. This means it can operate in GPS-denied or jammed environments.

Ukraine has ordered roughly 10,000 HX-2s, while Germany has approved a €268 million procurement package with options that could rise to €1 billion. 

For Helsing, HX-2 is also a manufacturing argument. Shifting technical complexity into the software means hardware manufacturing can be simpler, allowing production in larger numbers at lower cost. 

CA-1 Europa is the most ambitious expression of the model. Unveiled in September 2025 at Grob’s facility in Bavaria, the autonomous combat aircraft is Helsing’s entry into the collaborative combat aircraft market.

Intended to be produced affordably, and to operate alone or in swarms, CA-1 has an operational target around 2029.

Helsing’s Resilience Factory model follows from that logic, with RF-1 in southern Germany intended to produce about 1,000 HX-2s per month and a UK facility in Plymouth established for underwater glider production.

Other programs, including a collaborative satellite constellation project, and the Area 9 robotics research division, extend the vision even further.

The range can look sprawling. Even bewildering.

Helsing’s argument is that the through-line is software. Whether it flies, swims, orbits or drives, the company is building AI systems that help militaries understand the battlefield and act on that understanding faster.

Helsing’s main distinction is not that it uses AI. Defense or otherwise, these days every company says that.

Its distinction is that it was built on the principle that software will be the basis of military power. That puts it in a different position from traditional primes, which begin with platforms and integrate software around them.

Helsing starts with the software problem and then builds, buys or partners for the hardware needed to make that software operational.

Wired’s 2023 visit to Helsing’s offices described a battlefield display that felt almost like a video game: a yellow dot moves across a mountain range, zooms-in on a wide-winged military drone, the dot turns red. 

The point was not aesthetic. When software is supposed to help operators make decisions under pressure, usability is functionality.

It also shows up in Helsing’s emphasis on edge AI. Systems such as HX-2 are intended to operate when datalinks are disrupted and GPS is unreliable. In the electromagnetic spectrum, the same logic applies.

Western militaries face adversaries that have invested heavily in software-defined electronic warfare. AI that can help classify signals, identify usable frequencies and support jamming decisions is a necessity, not a nice-to-have.

Helsing is selling European governments more than products. It is selling the idea that Europe can build its own AI-enabled defense stack: software, autonomous platforms, manufacturing and, increasingly, space-based sensing. 9

That message lands in a Europe trying to rearm while also asking how dependent it can afford to remain on U.S. technology and U.S. political will.

Next is ethics, or at least ethical positioning. Helsing says it serves democracies — and Reil is specific about where that line gets drawn:

Reil has also emphasized that operators using Helsing products should not be nominally “in the loop,” but actively involved, with deliberate pauses before following AI recommendations. 

That positioning is part of the company’s appeal. It is also part of the tension. Helsing’s software sits very close to lethal decision-making, including in systems deployed in Ukraine. 

The company argues that democratic control, human engagement, and responsible design can make such systems compatible with Western values.

Whether governments, publics and international law settle on the same answer remains open.

Helsing’s rise is inseparable from Europe’s security moment.

Before Russia’s full-scale invasion of Ukraine, a large European defense AI startup was a harder political sell.

After it, the argument became easier: Europe had under-invested, the threat was no longer theoretical and the continent needed more of its own capability. 

European nations soon began to ramp up defense spending. In 2024, the European Defence Agency reported record spending of €343 billion, a 23% increase from 2023.

This more aggressive spending got further encouragement with the return of the Trump administration, which has repeatedly called for Europe to “to step up in a big way” when it comes to its own defense.

Helsing did not create this shift, but it was well positioned to capitalize on it.

In 2021, the idea of the Spotify guy backing a defense company triggered more than a little disdain. A few years later Helsing is raising capital at valuations that would have been difficult to imagine before the war.

He has also compared the situation to the car industry’s “Tesla moment”: if legacy defense companies don’t adapt, they risk being disrupted by companies that do.

This is the strategic bet beneath Helsing’s product announcements. Europe does not just need more platforms. It needs a sovereign software layer for modern war, plus enough industrial capacity to turn that software into fielded systems.

Sovereignty is not nostalgia for national defense industries. It is a practical requirement for a continent that may have to defend itself with less certainty about outside support.

The clearest risk is that Helsing’s ambition outruns execution.

The company is trying to deliver drones to Ukraine, scale Resilience Factories, develop autonomous combat aircraft, partner on a satellite constellation, expand into undersea sensing and explore robotics.

Breadth can be a sign of vision. It can also become sprawl.

HX-2 has already shown how harsh the gap between concept and battlefield reality can be. In January 2026, Testing issues and front-line jamming seemed to cool demand for further HX-2 orders from Ukraine and Germany. 

The Defense Post reported that orders had been paused or suspended, not cancelled. Helsing disputed this.

Germany’s later approval of a major HX-2 procurement package suggests the controversy did not derail the program, but it remains a useful warning: autonomy that works in controlled conditions still has to survive mud, jamming, logistics, and operator trust.

Valuation is another risk. An $18 billion valuation gives Helsing enormous credibility and resources, but it also creates expectations.

Defense procurement is slow, political and often locked into legacy programs. Even when governments want to move quickly, budgets and contracts may not. This can turn great expectations into great disappointments.

Then there is the autonomy question. Helsing insists it is building responsible systems for democracies, with humans engaged in lethal decisions.

As AI systems take the reins in combat, legal and moral pressure will increase. One high-profile mistake could shift the public debate quickly.

Helsing’s brand leans heavily on European autonomy, but advanced electronics supply chains are global. European sovereign will depend not only on code and factories, but on chips, sensors, components, cloud infrastructure, export controls, and political alignment.

Helsing is one of the clearest expressions of Europe’s new defense technology moment: software-first, autonomy-focused, well capitalized and explicit about sovereignty.

Its bet is that the future of defense will be shaped by companies that can turn overwhelming information into decisions, put autonomy into attritable systems, and manufacture those systems fast enough to matter.

The question is whether Helsing’s breadth becomes its advantage or its burden. Drones, fighter autonomy, underwater gliders, satellites, robotics, and factories all point toward an integrated vision.

It will be on the battlefield where Europe will find out if that vision can become a reality.

Helsing: The German Startup Building AI for Military Defense — Morgan Meaker, WIRED, July 2023

Saab and Helsing Let Gripen Fighter Fly With AI in Charge — Defense News, June 2025

Germany's Helsing Unveils AI-Enabled CA-1 Europa UCAV, Targets 2029 Entry to Service — Breaking Defense, September 2025

Helsing's CA-1 Drone Is an MQ-28 Ghost Bat Lookalike — The War Zone, September 2025

Ukraine Pauses Helsing HX-2 Orders — The Defense Post, January 2026

Helsing Disputes Reports Ukraine Paused HX-2 Orders — The Defense Post, January 2026

Ukraine Orders 6,000 Loitering Munitions From Germany's Helsing — Breaking Defense, February 2025

Once-Reluctant Germany Goes Big on One-Way Attack Drones — Defense News, February 2026

AI Company Helsing Unveils Swarming Underwater Surveillance Drones — Defense News, May 2025

Helsing Launches AI-Enabled KIRK Space Targeting Initiative — The Defense Post, May 2026

Helsing Nears $1.2B Raise at $18B Valuation — TechCrunch, May 2026

Raising $828M to Build the Defence Champion of Europe, 20VC, December 2024

Helsing's Reil on AI's Role in Defense Innovation, Bloomberg, October 2025