[ PERSPECTIVE ]17 Nov 20268 min
Amphibious Drones
Amphibious drones now have buyers in Ukraine and backers in the US and India. Water range limits the ground robots, the switch from air to water limits the hybrids, and no one has published how either survives a contested crossing.
In late September a ground robot from Ukraine's 93rd Separate Mechanized Brigade drove toward the front, reached water and kept going. It carried ammunition and food more than 40 kilometers over land and water to positions other ground robots couldn't reach. Ukraine's Defense Ministry called it the country's first logistics mission by an amphibious unmanned ground vehicle (UGV) [ REPORTED ]. A month later Brave1 Market, the defense-technology marketplace run by Ukraine's government-backed Brave1 cluster, opened a dedicated amphibious UGV category with three systems listed for military units to order [ DOCUMENTED ].
Amphibious drones now have buyers. The word covers two different machines, though, and the difference between them decides where the category goes.
01Two machines under one word
The first is a ground robot that floats: wheels or tracks on land, waterjets or paddling wheels in the water. Ukraine builds these, and the demand came before the product. The second is an aircraft that dives. It flies to a site, submerges, then surfaces and flies on. That branch is coming out of university labs and startup accelerators in the United States and India, and none has yet been used in a war.
A third pattern sits between them. In July the 123rd Separate Territorial Defense Brigade reported using an uncrewed boat to land an armed ground robot on Kinburn Spit, a Russian-held strip of coast in Ukraine's south. The brigade described the operation as the first of its kind [ REPORTED ]. Neither vehicle was amphibious. The boat handled the water and the robot handled the land.
02Why Ukraine got there first
Ukraine's front is cut by rivers, canals and marsh, and ground robots already run much of the last stretch of resupply. The ministry counts more than 100,000 UGV missions between January and mid-August 2026 [ DOCUMENTED ]. Every water obstacle on those routes meant a detour, a second vehicle, or a soldier. A dedicated funding track for amphibious UGVs opened in June, three months before the first mission [ DOCUMENTED ].
The 93rd Brigade's robot unit built its answer from two existing systems, one that could swim but not drive and one with the reverse problem, and taught each to do what it couldn't [ REPORTED ]. Its operator drove the vehicle entirely by hand, on separate land and water controls. The first attempt failed. Development to deployment took under three months. Defender Media identifies the vehicle as Triton, an eight-wheeled electric platform whose specifications have not been disclosed [ SINGLE-SOURCED ].
Manufacturers have followed. NODRICS, now part of Tencore, offers the MAT1LDA [ REPORTED ]. Mavka Technology showed the tracked BI34G in 2025 at a stated base price of $25,000 to $30,000 [ COMPANY-REPORTED ]. The Defense Ministry cleared UNEX for service in April 2025 [ DOCUMENTED ]. It floats, and its project lead says it can tow an artillery piece [ COMPANY-REPORTED ]. Toni, an eight-wheeled electric platform from a startup that hasn't disclosed its name, appeared in October [ REPORTED ]. Its developers say it passed trials at Iron Polygon, the test range run by the Defense Ministry and General Staff, and that it's the only amphibious platform to have done so [ COMPANY-REPORTED ] [ SINGLE-SOURCED ].
Russia, fighting across the same rivers, has aimed narrower. Its best-documented amphibious ground robot, the Chelnok, is a demining system built to carry a 725-kilogram explosive hose [ REPORTED ].
03Water range is the limit
On land this group performs respectably. In the water it thins out. Toni carries 380 kilograms on land at up to 13 km/h with up to 50 kilometers of range, and 280 kilograms on water at 2 to 4 km/h with 5 kilometers of range, the water figure stated at half battery [ COMPANY-REPORTED ]. MAT1LDA claims up to 160 kilometers on land and 50 on water [ COMPANY-REPORTED ].
A naval drone is a different order of machine. Ukraine's defense intelligence agency credits its Magura V5 surface drone with a range of up to 800 kilometers [ REPORTED ].

Stated water range in kilometers: Toni 5 (at half battery), MAT1LDA 50, Magura V5 surface drone 800. Drawn to scale.
Source: manufacturer statements for Toni and MAT1LDA; Ukrainian defense intelligence figure for Magura V5. None independently tested.
These are crossing vehicles. A river on Ukraine's front is crossed in hundreds of meters, and the crossing is when the robot is slowest and most exposed. Even on land the robots don't last long: near Pokrovsk, a logistics robot survived an average of two sorties [ SINGLE-SOURCED ]. At 2 to 4 km/h, a 300-meter crossing takes four and a half to nine minutes [ ASSESSED ]. That's the window that matters, and no manufacturer publishes data on how its platform survives it.
The ceiling is physical. A hull built to drive is a poor hull to swim, and a battery sized for land range buys little in water, where drag climbs steeply with speed. Taking water range toward naval-drone scale means building a boat that also drives, which is a different design [ ASSESSED ]. At Kinburn, Ukraine got around the problem by putting the robot on a boat.
04The aircraft that dives
The second branch starts from the other end. Kairox Systems, from Y Combinator's Fall 2026 batch, says it combines the speed of aerial drones and the stealth of underwater drones in one vehicle [ COMPANY-REPORTED ]. The pitch is to look for mines in a strait, fly out of the water to inspect a coastline, then engage maritime threats. Its three founders met at MIT, and it says it has engaged the Marine Corps [ COMPANY-REPORTED ]. In Bengaluru, AquaAirX's AVATAAR flies as a quadcopter, lands on water, dives and resurfaces. Its developers say the prototype has reached technology readiness level 6, the stage at which a working prototype is tested in conditions close to real operations [ COMPANY-REPORTED ]. As of March, open-water trials were still under way [ REPORTED ]. The company has been selected under iDEX, India's Defence Ministry program for funding defense startups [ REPORTED ].
The branch is older than these startups. The U.S. Office of Naval Research has funded Rutgers University's Naviator, a drone that flies and swims, since 2015. In November 2023 it awarded the spin-off SubUAS a contract worth about $3.7 million to have an uncrewed submarine launch and recover the drone without a human in the loop [ DOCUMENTED ].
The engineering problem is the transition. Water is nearly 800 times denser than air [ DOCUMENTED ]. A propeller or wing sized for flight is therefore badly matched underwater, and every design pays for the crossing in weight, efficiency or complexity [ ASSESSED ]. The missions on offer are real: mine-hunting, hull and coastline inspection, short-range strike in shallow water. What the branch lacks is what Ukraine's ground robots had from the start, a unit with a route it can't run today.
Cost per successful mission is the test both branches will face. Ukraine's land-and-water robots are priced to be lost, at $25,000 to $30,000 for Mavka's base model and about $45,000 for Toni [ COMPANY-REPORTED ]. Kairox pitches the air-and-water branch as cheap enough for one-way swarms [ COMPANY-REPORTED ]. No developer has yet shown a production price.
05What to watch
Water-leg test data. Survival and speed on a contested crossing, measured independently by a test range or a unit. A unit's published crossing record or loss count would be the first public benchmark. Until that data exists, every water-range figure in this note is a claim.
Armed variants. The commander of the 93rd Brigade's robot unit expects amphibious UGVs to follow ground robots: logistics first, then casualty evacuation, then strike and engineering [ REPORTED ]. The signal is the first reported combat use of an armed land-and-water robot.
A U.S. buyer. A Marine Corps or Navy purchase of an air-and-water hybrid, as distinct from the research funding the Navy already gives the Naviator, or a U.S. evaluation of a Ukrainian land-and-water platform. Either would show whether the American branch has found its mission.
06Judgment
We assess that amphibious drones will develop as two separate categories [ ASSESSED ]. Land-and-water robots will stay a crossing tool through 2027, bought in volume wherever rivers sit across supply routes, and their water range will keep them out of the naval-drone mission. Air-and-water hybrids will reach users first in inspection and mine-hunting, where the transition earns its cost.
Both branches share one gap: no public test data on what happens in the water under fire. The first platform to close it will set the price for the category.
Sources and confidence. Most specifications in this note are manufacturer-stated and unverified. The Defense Ministry has not named the vehicle used in the 93rd Brigade mission; Defender Media's identification of it as Triton rests on a single source. Toni's developers have not named their company, and Iron Polygon results are not public. Sources are linked on the claims they support. Figures as of 16 November 2026.
Legal disclaimers
General Disclaimer. This page, and Palisade Bergschrund Corporation’s (“Pal-Berg’s”) research, analysis, validation, testing, and advisory materials, are provided for general informational and evaluation purposes only, are subject to change, and do not constitute legal, regulatory, tax, accounting, investment banking, broker-dealer, placement agent, investment advisory, procurement, technical, operational, or military advice. This page is not a legal opinion, regulatory assessment, procurement advice, financial advice, or recommendation regarding any particular market-entry, transaction, teaming, licensing, investment, export-control, CFIUS, FOCI, facility-clearance, funding, or government-contracting strategy. Pal-Berg does not guarantee the accuracy, completeness, timeliness, future applicability, operational effectiveness, regulatory approval, procurement approval, financing outcome, investment outcome, transaction outcome, product performance, fitness for a particular purpose, or commercial success of any technology, product, company, market, or opportunity discussed. Pal-Berg makes no warranties, express or implied, regarding the accuracy, completeness, reliability, or fitness for a particular purpose of its research. Pal-Berg shall not be held liable for any direct, indirect, incidental, consequential, special, or punitive damages arising from the use or reliance upon this Research, including but not limited to loss of life, personal injury, property damage, or mission failure. Any reliance on Pal-Berg materials is at the recipient’s own risk and should be made only in connection with the applicable agreement, statement of work, and the recipient’s own independent technical, legal, regulatory, operational, and business judgment and advice from qualified counsel and other professional advisors.
Legal, Regulatory, Procurement, and Export-Control Matters. Statements regarding U.S. defense procurement, domestic-preference requirements, FOCI, CFIUS, export controls, cybersecurity requirements, funding programs, facility clearances, teaming, subcontracting, licensing, or other regulatory matters are provided for general informational purposes only. These areas are fact-specific, change frequently, and may depend on the recipient’s jurisdiction, ownership, product classification, customer, transaction structure, end use, end user, funding source, and other facts. Recipients should consult qualified legal, regulatory, export-control, procurement, tax, accounting, and other professional advisors before taking action or making decisions based on this page.
No Product Warranty; Third-Party Products. Pal-Berg does not manufacture, design, own, sell, supply, maintain, warrant, or guarantee any third-party or customer products, systems, hardware, software, equipment, components, or technologies reviewed, tested, validated, researched, or discussed. Pal-Berg disclaims any warranty, express or implied, regarding such products, systems, hardware, software, equipment, components, or technologies, including any implied warranty of workmanship, merchantability, fitness for a particular purpose, safety, operational suitability, durability, or performance. Product-level claims, specifications, warranties, defects, failures, and compliance obligations remain solely the responsibility of the applicable manufacturer, supplier, owner, or customer, and Pal-Berg shall not be responsible for any losses, claims, damages, injuries, failures, or liabilities arising from or relating to such products, systems, hardware, software, equipment, components, or technologies, except to the extent expressly agreed in a written agreement signed by Pal-Berg.
Operational Decisions. Pal-Berg’s services shall not be construed as operational, tactical, or strategic military advice. All operational decisions remain the sole responsibility of the commanding authority and designated decision-makers. Pal-Berg’s research, analysis, validation, testing, and advisory materials are few inputs among many that should inform, but not solely determine, operational decisions.
No Government or Third-Party Endorsement. This page is not prepared, approved, endorsed, or sponsored by the U.S. Army, the U.S. Department of Defense, or any other government agency, military organization, prime contractor, or third-party organization referenced herein. References to third parties are for informational purposes only.
pal-berg.com · ascend@pal-berg.com
© 2026 Palisade Bergschrund Corporation. All rights reserved.

