The logo of Japan’s electronic commerce and Internet company Rakuten is seen in Tokyo on March 18, 2014.
KAZUHIRO NOGI/AFP via Getty Images
Rakuten Group confirmed as Helsing’s broker on Monday, announcing it has become the official Japan representative for Munich-based defense AI startup Helsing — making the e-commerce and telecom giant the first civilian tech company to serve as an authorized broker under Japan’s newly reformed arms procurement rules, which the cabinet only enacted in April. A proof-of-concept field evaluation of Helsing’s HX-2 loitering munition is already underway with the Japan Ground Self-Defense Force (GSDF), with GSDF field tests through September results due by the end of September 2026. If the trial succeeds, Rakuten will propose Helsing’s systems formally to the Ministry of Defense — and the two companies will explore producing HX-2-class drones on Japanese soil.
The arrangement is notable not just as a business deal but as a structural test: Japan abolished the regulatory framework that confined arms exports and imports to five noncombat categories — rescue, transport, warning, surveillance, and minesweeping — only on April 21, 2026, under Prime Minister Sanae Takaichi. That reform five categories abolished April 2026 divided defense equipment into “weapons” (lethal systems such as warships, tanks, and missiles) and “nonweapons” for the first time, and it opened Japan’s defense procurement to a wider range of suppliers and intermediary arrangements. Rakuten, a company whose primary businesses are e-commerce, financial services, and mobile networks, would not have been in a position to shepherd a foreign loitering-munition maker into a GSDF evaluation under the prior rules. Now it is — and it is the first to do so.
What Rakuten Brings to Defense — and Why It Is Unusual
The decision to deploy Rakuten as a defense technology broker reflects both the company’s deliberate strategic pivot and Japan’s broader 2026 Defense White Paper civilian partnerships directive that the military establishment deepen ties with private-sector technology companies and startups. Rakuten’s Chief of Staff to the CEO, Hideaki Mukai, explained broker rationale to Reuters that startups “face significant structural barriers, including complex procurement systems, difficulties communicating with government agencies and challenging procurement cycles,” and that Rakuten aims to bridge those gaps by connecting defense technology with government needs.
The company’s path to this role was not sudden. In May 2025, Rakuten partnered with Brave1, Ukraine’s government-backed defense technology incubator, to help Ukrainian defense startups access the Japanese market with military and dual-use technologies — an initiative captured in the Rakuten-Brave1 defense startup partnership announcement that gave Rakuten its first sustained exposure to defense procurement logistics. Then, in early 2026, Rakuten Mobile played a central role in a biennial Japan-US joint military exercise, providing Open RAN joint military exercise cellular infrastructure for secure command, control, and communication across multiple US military installations in Japan in coordination with the Japan Self-Defense Forces.
Mukai had said in May 2025 that Rakuten would “not enter the defense industry right away.” Sixteen months later, Rakuten is brokering a loitering-munition trial for the Japanese army.
What Helsing Is Selling Japan — and How the HX-2 Actually Works
The centerpiece of the partnership is Helsing’s HX-2, an electrically propelled X-wing precision loitering munition with HX-2 unveiled December 2024 as its public debut. The drone weighs 12 kilograms (26.5 lbs), reaches speeds up to 220 km/h (137 mph), and carries a precision payload to targets at HX-2 range and specifications up to 100 km (62 miles). Its unit cost is approximately €17,500 (approximately $20,200) — roughly half the estimated production cost of Russia’s Lancet loitering munition, enabling a HX-2 unit cost versus Lancet radically different scale of deployment.
The HX-2’s defining technical feature is its ability to navigate, identify targets, and engage them with no dependency on GPS. Most drones in the Ukraine war have been vulnerable to Russian electronic warfare systems that suppress satellite positioning signals across most of the front line; a drone that cannot be jammed by that method operates in an entirely different threat environment. The HX-2 achieves this through a GPS-denied terrain-navigation architecture that combines three subsystems: stored terrain maps held in the drone’s flight computer, an onboard electro-optical camera that continuously captures live imagery, and a deep-learning AI system that compares those images to the stored maps in real time to maintain positional awareness. Inertial measurement units (IMUs) fill the gap between terrain fixes. The result is a drone that Russian satellite-jamming systems cannot defeat through their primary method — though the communications datalink between operator and drone remains a separate vulnerability that Russian electronic warfare can still target.
For terminal guidance, the system’s AI identifies and locks onto target classes — armored vehicles, artillery positions — as the drone enters the engagement zone. Human operators can abort the strike up to the final approach. Helsing describes this as a human-on-the-loop engagement model: the AI handles autonomous navigation and target tracking, while the human retains authority over the engagement decision. When operated through Helsing’s Altra software platform, multiple HX-2s can be coordinated by an Altra single-operator swarm coordination single operator — a swarm architecture Helsing calls “precision mass.”
The Performance Gap Japan’s GSDF Will Be Evaluating
Any Japanese defense official assessing the HX-2 this September is working with two sets of performance data that are not easy to reconcile.
In controlled exercises, the HX-2 has performed impressively. During the US Army’s Project Flytrap 5.0 hit rate exercise in Lithuania — held in late April through May 2026 — 15 of 17 HX-2 drones successfully hit their designated targets in a simulated electronic warfare environment, an 88% hit rate. US Army Chief Technology Officer Alex Miller described soldiers using the HX-2 as both a reconnaissance platform and a loitering munition because it was able to find and track targets with onboard computer vision and fly even under jamming conditions, per the Army CTO Alex Miller assessment.
In early combat deployment in Ukraine, the picture was more complicated. In January 2026, reporting citing an internal German Defense Ministry presentation documented a significantly lower launch success rate during trials with Ukraine’s 14th Separate Regiment of Unmanned Systems Forces, with some units missing AI-enabled terminal guidance and visual target acquisition components. Separately, a German investigative report cited a Ukrainian field document reporting that HX-2 units hit their targets in approximately 5 of 14 documented Donbas engagements — a rate near 35%. Both reports are discussed in depth in Bloomberg lower Ukraine success rate coverage from The Defense Post. Helsing disputed both reports and provided military test data from Germany, the United Kingdom, and Kenya documenting hit rates near 100%. Germany proceeded with its procurement, with the Bundestag approving an initial Bundestag EUR 269 million contract for HX-2 systems in February 2026.
The divergence between controlled-exercise and early combat performance is a documented pattern in military drone development — active adversaries adapting electronic warfare countermeasures in real time create conditions that controlled trials cannot replicate. The GSDF’s September evaluation, if conducted under realistic electronic warfare conditions, will generate a data point that is independent of the Ukraine context entirely.
Japan’s Defense Landscape: Why This Trial Is Happening Now
The Rakuten-Helsing deal lands against a backdrop of the most significant transformation of Japan’s military posture in a generation. Tokyo has committed to spending 2% of GDP on defense, roughly doubling its prior budget, and has paired that increase with sweeping reforms affecting what it can buy, from whom, and through what channels.
The April 21, 2026 arms export reform eliminated the Five Categories framework that had confined Japan’s defense equipment transfers to non-combat purposes since the 1960s. Under the two-category lethal-weapons framework, Japan divides defense equipment into “weapons” and “nonweapons” by lethal capability, with weapon exports permitted to the 17 countries with which Japan has signed defense transfer agreements — a list that includes Germany, the US, the UK, Australia, France, and other NATO allies. The reform passed through a cabinet decision rather than parliamentary vote, drawing opposition protests arms export reform from opposition parties, which argued that parliamentary approval was legally required and that the change risked turning Japan into what one critic called “a merchant of death.” More than 30,000 protesters gathered outside the National Diet building in Tokyo on April 19, 2026.
Within the GSDF itself, the structural transformation is equally visible. On April 13, 2026, the Ministry of Defense formally established GSDF unmanned offices — the Unmanned Defense Capability Promotion Office and the Unmanned Systems Office — to institutionalize the use of unmanned systems. In February 2026, the GSDF selected Drone40 via Marubeni Aerospace — the Australian-designed Drone40 loitering munition — for close-range infantry strikes, delivered through Japanese trading firm Marubeni Aerospace. The HX-2, with its 100 km (62 mile) range and GPS-denied autonomous navigation, targets a fundamentally different mission set: medium-to-long-range precision strikes against armored vehicles and surface vessels — exactly the capability Japan’s FY2026 wide-area UAV budget earmarks ¥11.1 billion (approximately $70 million) to acquire in the form of five “wide-area UAVs.” The Bayraktar TB2S and IAI Heron Mk II are the leading wide-area candidates; the HX-2 fills a more attritable strike role in the same portfolio.
Japan’s August 2026 Defense White Paper, approved by the cabinet on August 4, made explicit what had been implicit in the budget: Japan intends to make civilian technology companies and startups integral to its defense ecosystem. “Speed is key,” Defense Minister Shinjiro Koizumi said in early July, describing his ministry’s intent to accelerate integration of civilian innovation. Japan’s five-year defense buildup plan allocates approximately ¥1 trillion (approximately $6.3 billion) to procure unmanned systems across the Ground, Maritime, and Air Self-Defense Forces through fiscal 2027, per five-year trillion-yen drone plan analysis from The Diplomat.
What Comes Next — and What Could Still Fail
The September timeline is strict. A successful GSDF evaluation would likely accelerate a formal Defense Ministry procurement proposal for the HX-2 and open discussions on domestic Japanese production — which would align with Tokyo’s stated goal of building sovereign defense industrial capability and reducing reliance on foreign supply chains for critical military hardware.
The risks are real. Japan’s defense procurement process is notoriously deliberate, and a proof-of-concept trial is not a contract. The performance gap documented in Ukraine — whatever its precise cause — is a data point GSDF officials will need to assess against the Flytrap 5.0 results. And Rakuten’s role as a civilian broker, while now legally permissible, is still unprecedented in practice: there is no established playbook for how a company whose primary products are e-commerce and mobile networks navigates the Ministry of Defense’s procurement bureaucracy on behalf of a foreign munitions startup.
If it works, however, the model matters beyond this specific deal. European defense companies competing for access to Japan’s newly opened market have been watching for exactly this kind of civilian-brokered pathway. Helsing’s successful navigation of it would validate a go-to-market strategy that dozens of other non-traditional defense companies — from Germany, from the UK, from Scandinavia — could replicate with Japanese civilian tech partners who have government relationships but no defense primes to lean on.
That is what Rakuten Chief of Staff Hideaki Mukai described as the Mukai broker value proposition: not a weapons manufacturer taking on a new product line, but a company with institutional access, procurement experience, and government trust acting as the connective tissue between a foreign defense AI innovator and a military customer that is, for the first time in decades, ready to buy.
Currency conversions in this article are approximate, based on exchange rates as of August 17, 2026.
Frequently Asked QuestionsHow does the Helsing HX-2 navigate and strike targets without GPS?
The HX-2 uses a navigation system that does not depend on satellite signals at all. Before a mission, operators load georeferenced terrain maps into the drone’s onboard flight computer. During flight, the drone’s electro-optical camera continuously captures live imagery and compares it against those stored maps using a deep-learning AI system — the same basic principle as the terrain contour matching used in Tomahawk cruise missiles, but implemented with modern machine vision. This allows the drone to maintain precise positional awareness even when Russian electronic warfare systems suppress satellite signals completely. For target acquisition in the terminal phase, the AI recognizes and locks onto target types — armored vehicles, for example — while the human operator retains the ability to abort the strike up to the final approach. One vulnerability the GPS independence does not solve: the radio datalink between operator and drone can still be targeted by electronic jamming separately.
Why is Rakuten, an e-commerce company, involved in Japan’s military procurement?
Two things converged to make Rakuten’s role possible. First, Japan’s April 2026 arms export reform eliminated the regulatory framework that had previously confined defense equipment transfers to five non-lethal categories, opening the procurement process to a broader range of intermediary arrangements. Second, Rakuten had been systematically building defense-adjacent credibility: in May 2025 it brokered Ukrainian defense startup access to the Japanese market through a partnership with the Ukrainian government’s Brave1 incubator, and in early 2026 its Open RAN mobile network infrastructure played a direct role in a biennial US-Japan joint military exercise. Those two tracks — regulatory opening plus demonstrated relevance — created the conditions for Rakuten to serve as Helsing’s official Japan representative.
What is Japan’s arms export reform, and what changed on April 21, 2026?
Japan had operated under the “Three Principles on Transfer of Defense Equipment and Technology” since the 1960s, with an implementation guideline that confined equipment transfers to five non-combat categories: rescue, transport, warning, surveillance, and minesweeping. This framework essentially prevented Japan from exporting or importing lethal weapons through civilian channels. On April 21, 2026, Prime Minister Sanae Takaichi’s cabinet abolished that five-category restriction, replacing it with a two-category system based on whether equipment has “lethal or destructive capability.” Non-weapons remain largely freely exportable; weapons may be exported only to the 17 countries with which Japan has signed defense transfer agreements. The change passed by cabinet decision rather than parliamentary vote, drawing significant domestic opposition, including protests of more than 30,000 people near the Diet building days before the decision.
What happens if the September GSDF trial fails?
A failed or inconclusive trial would most likely delay rather than end Helsing’s Japan ambitions. Japan’s defense procurement timeline is typically measured in years, and a single proof-of-concept evaluation rarely produces a final yes or no. What a failed trial would test is Rakuten’s broker model itself: if the company cannot successfully shepherd Helsing through the evaluation process to a positive outcome, it signals that civilian tech companies without defense-prime DNA face structural challenges as procurement intermediaries that go beyond government access and procurement relationships. Either outcome — success or failure — will generate information about whether Japan’s newly reformed procurement model is as open in practice as it is on paper.