5JUL

German AI drone passes US Army test

4 min read

10:21UTC

US soldiers from V Corps scored 15 kills across 17 engagements with Helsing's HX-2 at Pabrade, Lithuania, the first confirmed US Army operational test of a European autonomous strike drone.

← Back to UK's £5bn drone bet follows Healey's exit Jump to analysis ↓

TechnologyDeveloping

Key takeaway

A European AI drone cleared the US Army's own GPS-denied test, a scenario the US incumbent had not priced.

Soldiers from V Corps, the US Army's forward command in Europe, put Germany's Helsing and its HX-2 artificial-intelligence strike drone through a live trial at Pabrade, Lithuania on 9 June, scoring 15 kills and 2 near-misses across 17 engagements under Project Flytrap 5.0 ¹. Project Flytrap is V Corps's own fielding accelerator on NATO's eastern flank, not a vendor demonstration. The run jammed satellite navigation and disrupted communications, the exact conditions that defeat most off-the-shelf drones, which makes an 88% effective hit rate the figure that matters ².

Helsing's HX-2 had hit Russian targets in Ukraine before this trial , so the airframe arrived combat-tested. At Pabrade the watchers changed: an Army-run evaluation on an Army range, scored by US soldiers, marked the first confirmed US operational test of a European autonomous strike drone. GPS-denied autonomy, the ability to keep flying and striking when navigation signals are spoofed or blocked, is the capability boundary that separates drones usable in contested airspace from those limited to permissive skies. The HX-2 cleared it in front of the people who write requirements.

The competitive weight follows the money. Helsing raised at an $18bn valuation last month , so a European AI drone now carries both capital and US Army validation. The American autonomous-strike incumbent has built its case on being the default Western supplier; a foreign drone performing this well in front of US evaluators is the scenario that case did not obviously have to address. Caveats hold. One exercise is not a programme of record, the funded procurement line that turns a trial into orders, and a single contested run proves capability rather than scale. The result still puts a non-US strike drone on a US Army scoreboard for the first time.

Deep Analysis

In plain English

Helsing is a German company that makes a type of drone called a loitering munition: it flies to an area, waits for a target, and then strikes it. This kind of drone has been used extensively in the Ukraine war. The US Army tested Helsing's HX-2 drone at a training ground in Lithuania. They jammed GPS signals and communication systems to make conditions as difficult as possible, then scored how many targets the drone hit. It hit 15 out of 17. That result matters because it happened under official US Army conditions scored by US soldiers, rather than in a vendor-run demonstration. Clearing a US military test is a documented prerequisite for entry into the world's largest defence procurement pipeline.

Deep Analysis

Root Causes

Two structural gaps explain why the US Army evaluated a German drone rather than relying on domestic alternatives.

First, the US autonomous-strike sector has concentrated its investment above the loitering-munition tier. Anduril's Roadrunner interceptor and Fury combat aircraft both operate in the $100,000-$500,000 unit-cost band. The HX-2 sits below that at a price point designed to enable mass attrition tactics drawn from the Ukrainian theatre. No US programme of record has filled that band with an AI-autonomous system; Helsing entered a gap that the US industrial base had not yet addressed.

Second, V Corps' Project Flytrap exercise structure was specifically designed to accelerate adoption of non-US systems that have combat validation from Ukraine.

The exercise mandate to test in GPS-denied and electronic-warfare-contested conditions was written after the Lithuanian government lobbied SACEUR for NATO-eastern-flank trials to reflect the actual threat environment rather than the permissive conditions of exercises held inside the continental US. Helsing's combat record in Ukraine made it a natural candidate for the first non-US entry.

What could happen next?

Consequence
V Corps' Project Flytrap result gives Helsing a credible US evaluation credential that Nordic, Baltic, and Central European defence ministries will use as a procurement reference point, accelerating non-German NATO adoption without requiring a US programme of record.
Short term · Reported
Risk
US procurement would require Helsing to either navigate the Foreign Military Sales process or enter a co-production agreement with a US prime, both of which compress margins and dilute the revenue assumptions inside the $18bn valuation.
Medium term · Assessed
Opportunity
If Anduril's loitering-munition gap persists through 2027, Helsing could enter the US market as a sub-tier supplier to a US prime integrator rather than as a direct competitor, preserving its European revenue base while adding US volume.
Long term · Suggested

Source Landscape

Primary parallel: Israel's Rafael Advanced Defense Systems first received a US evaluation score for the Trophy active-protection system in 2011 at Aberdeen Proving Ground. The Army acknowledged the performance but did not buy Trophy for seven years, citing integration complexity with M1 Abrams and a requirement that foreign vendors partner with a US prime.

The HX-2 faces the same structural gate: a strong evaluation result opens the conversation but does not bypass the Foreign Military Sales process or the Buy American Act requirements that govern how a foreign system enters the US procurement pipeline.

Counter-parallel: South Korea's Hanwha AS21 Redback infantry fighting vehicle passed US Army evaluation at Fort Wainwright, Alaska in 2024 as part of the Optionally Manned Fighting Vehicle programme.

Hanwha submitted Redback as a direct programme competitor bidding against US platforms, a structurally different position from Helsing's HX-2, which entered a V Corps fielding-accelerator exercise rather than a programme-of-record competition. The Trophy path, where Rafael's system achieved acknowledged performance but spent seven years navigating a US-prime partnership requirement, is the structurally closer precedent for HX-2.

Helsing's $18bn valuation rested heavily on three revenue pillars: the Bundestag's €1.46bn seven-year framework, potential NATO-wide adoption, and the prospect of US procurement. The V Corps result addresses the third pillar, but only as a first-stage signal.

US procurement of a foreign autonomous-strike system would require either a Foreign Military Sale (an intergovernmental transaction) or a domestic production agreement with a US prime, typically a co-production arrangement that divides intellectual property and manufacturing.

Each path compresses Helsing's margin. Under a co-production arrangement, the $18bn valuation would need to be repriced to reflect royalty-and-licence income rather than full production revenue, a multiple compression the market has not yet priced because the exercise result does not yet commit to either path.

For the European drone sector more broadly, a US Army validation carries market signalling value into non-US NATO procurement. Nordic, Baltic, and Central European defence ministries routinely use US evaluation outcomes as a proxy for technical credibility when their own test infrastructure is limited. A strong Flytrap score for an EU-manufactured drone reduces the informational cost of those ministries selecting a European supplier over a US alternative.

Consensus view: IISS's Franz-Stefan Gady, writing on NATO eastern-flank drone integration in April 2026, assessed that GPS-denied autonomous strike capability is the single hardest technical gate for non-US vendors to clear in American military evaluations because US Army testing doctrine (MIL-STD-882E) applies a systems-safety analysis that is structurally more demanding than any European national standard.

Clearing it in a US-run exercise on NATO soil carries a different signal weight than vendor-supplied performance claims.

Counter-view: The Moscow-based Institute of World Economy and International Relations (IMEMO) assessed in its May 2026 security review that the 88% hit rate should be read against the target set: exercise targets are stationary or predictable in motion, not actively manoeuvring or equipped with electronic countermeasures.

IMEMO's defence-technology analysts argue that contested-airspace performance against a structured exercise target systematically overstates operational lethality by 20-30% relative to combat conditions in Ukraine.

Key tension: Whether V Corps' formal scoring in a US-governed test environment constitutes the procurement-path endorsement that Helsing's $18bn valuation requires, or whether a programme of record requires a separate US acquisition process that the exercise result cannot shortcut.

Sources:Army Recognition

Mentions:Lithuania →Ukraine →Hanwha →US Army →Moscow →Alaska →IISS →Germany →NATO →Israel →South Korea →Pabrade →V Corps →Helsing →HX-2 →Project Flytrap →

First Reported In

Update #12 · Pentagon's drone buy lands a third short

Army Recognition· 15 Jun 2026

Read original →

Causes and effects

Caused by

Helsing HX-2 confirmed in Ukraine combat

HX-2's prior combat deployment in Ukraine (event 3745) established the combat-proven baseline that V Corps evaluated under Project Flytrap 5.0 in Lithuania.

Occurred 20 May 2026

Read story →

Helsing closes $18bn round, led by Dragoneer

Helsing's $18bn valuation raise (event 3175) contextualises the competitive stakes of HX-2 passing V Corps evaluation — a European AI drone now has both capital and US Army validation.

Occurred 9 May 2026

Read story →

This Event

German AI drone passes US Army test

A foreign AI strike drone performing in front of US Army evaluators puts a non-US supplier into a competition the American incumbent's valuation has not had to price.

Led to

Anduril's numbers defer profit to 2030

Helsing HX-2 passing V Corps evaluation (event 1 in this run) directly pressures the assumption that Anduril is the default autonomous-strike supplier underpinning its $61bn valuation.

Occurred 1 Jun 2026

Read story →

Different Perspectives

Procurement sceptics

Two identically sized IDIQs to different primes within seven weeks, and a five-nation pact where one partner moves weeks ahead of the rest, could just as easily read as an industrial base still improvising vendor mix as a deliberate hedging doctrine. Neither ceiling appears sized against a validated requirement yet.

Chinese component suppliers

FCC and Pentagon-level restrictions on Chinese-origin airframes and motors have progressively excluded DJI-linked suppliers from federal counter-drone catalogues, narrowing the field JIATF-401 and Gauntlet II can buy from to a short list of certified domestic bidders. Beijing reads the exclusions as protectionism dressed as security policy.

Quantum Systems

Blackstone, Airbus, Advent and Noteus reportedly priced the Munich drone maker at $8 billion, more than double its prior mark, on the strength of 19,000 Ukraine missions rather than confirmed revenue. It is betting production capacity, not current sales, is what buyout capital is now paying for.

United Kingdom

Britain released the first LEAP effector money three weeks after its defence secretary quit over the size of the drone budget, splitting £3.16 million across three small firms rather than one contractor. It expects the other four LEAP partners to follow its pace, not set their own.

JIATF-401

The task force handed AeroVironment a $500 million counter-drone ceiling identical to Perennial Autonomy's from seven weeks earlier, while its own Gauntlet II red team prepares to attack the drones the winners of that sprint will build. It expects to keep several qualified suppliers warm rather than certify one.

DroneShield

DroneShield appointed retired Rear Admiral Lee Goddard as an independent director from 1 July, its second board move since founder Oleg Vornik's April exit. The ASIC probe into November's disclosures and share sales stays open, so the admiral steadies the story without closing the file.