Milrem's UGV land-grab: one Estonian robot, three continents, six days

In 2013, a handful of Estonian engineers started building a tracked machine with no cockpit, no gun and nobody inside. Thirteen years later that machine, the THeMIS, is part of robotics programmes or in operational use in 21 countries.

This month Milrem Robotics did something more consequential than sell robots. Across six days it signed industrial agreements on two continents, converting a single Estonian factory product into a licence that other nations can build, integrate and sustain themselves.

On 7 October, at the Land Forces exposition in Perth, Hanwha Defence Australia bought a THeMIS and opened a framework to produce and sustain the vehicle locally for Australia and New Zealand. Five days later, at the US Army's AUSA 2026 meeting in Washington, Milrem signed a memorandum of cooperation with Chugach Alaska Corporation to explore building its unmanned systems inside the United States.

The robot was never the hard part

A THeMIS is not exotic hardware. It runs 1,815 kg, hauls a 750 kg payload, and stays out for 15 hours on a hybrid diesel-electric drivetrain. Top speed is 20 km/h. It tows at up to 80 km/h behind a truck, climbs a 60% grade, and crosses a 90 cm gap.

The interesting numbers are industrial. Milrem's value has always sat in the boring middle of the problem: a modular platform that militaries can bolt payloads onto without re-qualifying the base vehicle. Combat, cargo, casualty-evacuation and route-clearing versions share one chassis. Kongsberg weapon stations, EOD packages and clearance kits mount on the same frame.

That is what makes the platform easy to licence out. When the hard part is integration logic rather than a secret actuator, moving production abroad becomes a packaging decision.

Perth: the licence is the product

Hanwha's Asia-Pacific move is the clearer template. Hanwha Defence Australia bought a THeMIS outright. One platform goes to South Korean research and development; one stays in Australia. The purchase is the small print. The memorandum of understanding around it sets up local production, sales and sustainment for the wider region.

The clause that matters is MIFIK. Milrem's Intelligent Function Integration Kit adds autonomous navigation and mission execution to vehicles that were not born autonomous, and Hanwha holds an option to fit it to its Redback and Huntsman families. The work would run through Hanwha's Armoured Vehicle Centre of Excellence in regional Victoria.

The two firms already share history. Hanwha Aerospace signed a robotic combat vehicle memorandum with Milrem in February 2025, and the pair are chasing an unmanned ground vehicle programme in Romania with a localisation component. The Perth agreement reads less like a new relationship than an early one moving from paper to floor space.

Washington: a route around the export queue

The US agreement is looser but larger in ambition. Signed on 12 October, it names Chugach as Milrem's possible preferred American industrial partner for production, assembly, integration, sustainment and lifecycle support.

Two phrases carry the strategy. The first is Foreign Military Sales: systems built on US soil could reach allied and partner nations through established American acquisition channels. The second is the commercial dual-use market, where the same chassis serves non-defence customers.

Together with Hanwha, we are creating a strong foundation for expanding unmanned and autonomous capabilities across the Asia-Pacific region.— Kuldar Väärsi, Chief Executive Officer, Milrem Robotics

Stefan Behre, Milrem's chief sales officer, framed the American objective with similar restraint: establish the capability to produce, integrate and sustain the systems in the United States, rather than only sell into it.

Why localisation became the export

Both deals sit inside one shift. For a decade the export question in ground robotics was whether a foreign army could operate the vehicle. Now it is whether the buying country can build, repair and upgrade it without waiting on someone else's supply chain.

Milrem sells into that demand with an unusual asset: a product small enough to license and mature enough to quote. The company led the European iMUGS programme for modular manned-unmanned teams and now runs its €55 million successor, iMUGS2. ARCOS, its command-and-coordination layer for managing robotic forces, gives partners a reason to standardise on Milrem software even when they weld their own bodies.

The bet is that software travels further than steel. A country may never buy a finished THeMIS, but it can adopt the autonomy kit, the command layer, and the maintenance model underneath them.

Capacity is the real constraint

Milrem is not alone in the category. US startups build off-road autonomy stacks, and established contractors field their own tracked platforms. The differentiator is rarely software. It is whether a design can be manufactured on two continents without re-testing every bolt.

That is why the October agreements read as procurement news even though no large contract was signed. A defence buyer choosing a ground robot is choosing a supply chain. A vehicle that can be assembled in the buyer's own country, sustained by local crews and upgraded by local engineers is a lower-risk purchase than a better robot that must cross a border for every spare part.

The same logic shapes the payload catalogue. Weapon stations, casualty-evacuation kits, mine-clearance systems and reconnaissance masts all mount on one chassis. Each configuration is a reason for a partner to standardise; every standard is a form of lock-in.

The trade cuts both ways. A partner that learns to assemble the platform also learns how it is priced. Localisation transfers know-how as surely as it transfers jobs, and the partner, not Milrem, decides how much of the stack to keep buying once the line is running.

For now, the balance favours Milrem. The company holds the autonomy kit, the command layer and the design authority, and it is selling access to all three at the moment its platform is most in demand.

There is a cost to this model, and it is strategic, not technical. Every licence handed out trains a partner's engineers on Milrem's architecture, and a partner that can build a THeMIS can eventually build around it. The upside is reach: a US production line turns a foreign platform into an FMS-eligible one, and an Australian line turns a European product into a regional standard.

Does the licensing model travel?

Development scenarios

The ground-robot market is crowded with prototypes and thin on production capacity. Milrem's October moves are an attempt to convert an early lead in one into a claim on the other — and to sell the claim before a competitor can build it first.