The German company that built the Leopard tank just became a contractor to the U.S. Army's last mile. On July 31, American Rheinmetall won an 18-month award under Project Sustainment to deliver hybrid-powered autonomous ground vehicles for company-sized resupply missions. Rheinmetall's name is not the surprise. The surprise is who it brought along: an electric truck chassis maker from California, an autonomy startup, and a natural-language interface company.
Harbinger's hybrid chassis and Primordial Labs' Anura voice interface turn a logistics vehicle into a system a single soldier can task.
The award is a prototype contract, not a production decision. The winner is whoever proves out the contested-environment resupply run.
The award that reopens the last mile
Project Sustainment runs through the National Advanced Mobility Consortium (NAMC), a Department of War acquisition body that uses Other Transaction Authority to bypass conventional procurement. Five companies got in: AM General, American Rheinmetall, Carnegie Robotics, HDT Robotics (now BLADE), and Stratom. Rheinmetall primes the team; Harbinger supplies the dual-use hybrid chassis; Forterra provides the autonomy stack; Primordial Labs contributes Anura, a voice interface that turns spoken commander intent into robot commands.
Autonomous resupply is not a new idea. The Army has tested unmanned ground vehicles for logistics for a decade, and the early platforms proved the concept while failing the economics. A tele-operated truck requires a dedicated operator per vehicle, which cancels the whole point of removing a soldier from the delivery run. The shift under Project Sustainment is architectural: one operator tasks a fleet, the vehicles handle the rest.
That is the difference Anura is built for. Primordial Labs claims its interface scales the soldier-to-robot ratio, so a single soldier commands multiple systems with heads-up, hands-free control. Whether it works at the unit level is the open question the award is meant to answer.
Project Sustainment contract term
American Rheinmetall leads an industrial team on an 18-month prototype contract with potential follow-on orders, per the company announcement. · Rheinmetall, 2026
Three partners, one chassis
Harbinger is the quiet piece of this deal and the most important one. The company builds medium-duty commercial electric vehicle platforms, chassis-first, designed to be up-fit for different missions. A commercial chassis means the Army is no longer paying a defense prime to design a bespoke truck and then amortize that cost over a few hundred units. It is buying a platform that already exists in the market and telling its engineers to bolt on the autonomy.
That inverts the usual defense supply chain. For years, ground robotics programs started with a military requirement, contracted a prime, and waited six years for a prototype. The commercial-first approach starts with a product that is already on the road and moves the defense work to the software layer.
This award expands American Rheinmetall's growing portfolio of uncrewed ground systems within the U.S. market and reinforces our position as a leader in ground robotics and autonomous technologies.Jim Schirmer, VP of Sales & Marketing, American Rheinmetall
The competing BLADE platform shows the same logic from the other direction. Its Dire WOLF is a six-wheeled diesel-electric hybrid with non-pneumatic Michelin Tweels and an onboard battery pack that can export power at the forward edge. The vehicle is built to survive terrain that flattens commercial tires and to double as a power source. Two different hybrid answers to the same problem: keep the last mile moving without a driver.
What Project Sustainment actually is
What to watch: whether the program converts from prototype to production, and which of the five architectures wins the down-select.
The economics of the driverless supply run
The business case for autonomous logistics is unusually clean compared to other defense robotics. A resupply convoy is a target; the U.S. Army has lost drivers and vehicles on predictable resupply routes for decades. Every vehicle that removes a driver from that run is an operational asset and a cost saving at once. There is no adoption barrier of "nice to have" , the requirement is written in doctrine.
The commercial parallel exists too. Warehouse automation, mining haul trucks, and port terminal tractors have already proven that autonomous ground vehicles work in structured, high-traffic environments. Defense logistics is the same problem with worse terrain and active adversaries. The companies that solve the contested-environment case will own a market that has no civilian equivalent because no civilian customer faces the same threat model.
Investors should price the difference between a prototype award and a production contract. The 18-month term funds prototypes and field testing. The prize, if it comes, is the follow-on order. Rheinmetall, BLADE, and the other three are racing for that, and the contest is what makes the program worth tracking.
Whether the Army down-selects one platform or keeps parallel vehicles at order level
Anura's real-world performance at field tests , does one soldier actually task a fleet?
Whether the Harbinger commercial chassis survives contested-terrain testing
Follow-on order announcements under the NAMC vehicle through 2027
What the commercial chassis changes for defense buyers
BLADE's Dire WOLF and Rheinmetall's Harbinger-based UGV both prove the same procurement shift. The Army is no longer selecting a single interceptor or a single truck. It is selecting a capability area, then letting five vendors compete at the order level under a standing vehicle. That is how the counter-drone market organized itself, and the Army says the counter-drone market is red hot.
Autonomous logistics is the second capability area to get that treatment. The structural consequence is a wave of mid-size defense suppliers holding seats on vehicles they can sell against on every task order. For investors, that is a different risk profile than betting on one prime. The winners are the ones who can execute at the order level, not the ones who win the headline.
Can the Army make autonomous resupply work in the next two years?
Probability: 70% , the prototype funding is committed, the commercial chassis already exists, and the Army's stated priority for autonomous logistics removes the political risk that sank earlier programs.
✅ Arguments for
NAMC's Other Transaction Authority lets the Army move faster than the traditional acquisition pipeline.
Five vendors competing at order level means the Army hedges its bet rather than betting on one design.
Confirmation criteria: a down-select or follow-on production order from one of the five by Q4 2027.
❌ Arguments against
Prototype programs often stall between award and production as budgets shift.
Five vendors on one vehicle is a crowded field; consolidation may kill weaker architectures before trials.
Disconfirmation criteria: a lapse in the NAMC vehicle, or a shift back to a single-prime procurement model.
Development scenarios
🟢 Optimistic scenario (25%)
Implications: commercial-chassis suppliers like Harbinger become structural defense suppliers, and the dual-use playbook spreads to other Army capability areas.
🟡 Base-case scenario (55%)
Implications: the five vendors hold seats and fight at order level; the capability improves steadily but never breaks out as a headline program.
🔴 Pessimistic scenario (20%)
Implications: autonomous logistics returns to pilot purgatory, and the commercial-chassis experiment is shelved in favor of a single prime.
The last-mile delivery market has been through this exact cycle before. As we wrote in July about Gallatin AI's agentic logistics navigator, the Army is pushing autonomy into sustainment planning ahead of field hardware. Project Sustainment is the hardware half of that bet, and procurement reform, which we covered in the $54.6 billion defense robotics context last month, is what makes the five-vendor structure possible.
The name Rheinmetall on the award gets the attention. The chassis and the voice interface are the parts worth watching.
