The use of unmanned combat systems has evolved in the past 2 decades from a one-off, expensive system launched from expensive platforms, to very low cost autonomous drones ($25-50K), UUVs (Unmanned Underwater Vehicles), autonomous ships and ground vehicles. This has occurred against a backdrop of active conflicts in Ukraine and the the Middle East, as well as tensions between superpowers in Europe and the Pacific, and a fight to control natural resources and rare earth materials (the Artic is an area where there are increasing tensions, South America and Africa are others). The race to deploy increasingly sophisticated systems (offense and defense) in outer space and oceans is on, primarily a contest between the United States, Russia and China.
Advances in physical AI and AoT™ (Autonomy of Things) in commercial areas like construction, agriculture, infrastructure monitoring, transportation and industrial robotics can be leveraged to provide low cost autonomy solutions in the defense and war sector – such assets can be sent into contested areas, minimizing loss of human lives and achieving strategic and tactical goals in warfare. It also reduces the risks of losing expensive assets like ships, submarines, and expensive aerial assets. The advances in AoT™ in the commercial area has been driven by increased sophistication of sensors for perception and localization (GPS is used, although GPS denial needs to be dealt with in a contested environment), as well as path planning and control. Efficient, low power consumption edge computing to learn and implement intelligent decision making has evolved substantially. These advances can be leveraged for defense as well, and can create a force multiplier in conflict environments.
While autonomous assets like drones and ISR (Intelligence, Surveillance and Reconnaissance) have been implemented over the past decade, they have generally been launched as a single asset or multiple assets which worked independently. Collaborative autonomy is the next frontier – multiple networked, unmanned systems (UAVs, robots, surface vessels) operating together as a team to execute complex missions, improving survivability and efficiency. These systems, often powered by AI, share data in real-time to make collective decisions based on a human commander’s intent, acting as a force multiplier in contested or dangerous environments. This is where Havoc.ai comes in – “inventing how autonomous systems team with each other to accomplish critical missions“.
HavocAI – AI Assisted Operation of Multiple Dynamic Autonomous Assets
HavocAI (yes, that is its actual name, coined to instill fear into it’s adversaries) was founded 18 months ago by Paul Lwin and Joe Turner, both U.S. Navy veterans and AI experts. Headquartered in Rhode Island, it recently raised $85M to expand operations (total investment to date is $100M) and has 108 people on staff. The raise was led by B Capital and participation from In-Q-Tel venture arm of the CIA), Lockheed Martin Ventures, Hanwha, Taiwania (essentially a venture fund supported by the Taiwanese government), Vanderbilt University, Up Partners, Island Green Capital, and Zero Infinity Partners. Early investors Scout Ventures and Outlander Ventures also participated. The investment from Taiwania is especially significant given the tensions in the Taiwan Strait.
The company was founded to provide the U.S. Department of War an unfair advantage over its rivals, using physical AI advances to launch collaborative autonomous assets in maritime, terrestrial and aerial environments. It recently executed the world’s first successful Air-Sea autonomy mission trial in a GPS-denied environment. Conducted in Troia, Portugal, the trial showcased real-time coordination between autonomous surface vessels (ASVs) and unmanned aerial vehicles (UAVs), all managed through a single, unified command interface. Multiple runs were conducted under international signal degradation, and all platforms maintained mission performance through alternate navigation and timing solutions. The demonstration also showcased fused, real-time sensor inputs across both air and surface platforms (Maritime Domain Awareness), identification and tracking of a designated vessel of interest (Automatic Target Recognition), and autonomous execution of the full kill chain (Target Detection, Classification & Engagement). The focus of this trial was to demonstrate capabilities to the Ukranian defense force, which desperately needs these solutions to fight and defend against a much more powerfully equipped and manned Russian army (Figure 1):
The motivation for HavocAI’s founders was the realization that nations facing contested littoral and open-sea environments—including Ukraine and other frontline partners (like Taiwan), need solutions that leverage advances in physical AI and sensing to create collaborative autonomy between various asset modalities ( air, sea, and ground), without persistent connectivity and communications. Given a mission objective, the goal is sense the operating environment, detect and engage targets at machine speed, with or without HITL (human in the loop).
HavocAI’s maritime autonomy fleet currently consists of 3 types of vessels, all commercially procured and outfitted with HavocAI’s sensor and compute suites:
- Rampage: 16 ft. length, all electric, solar panels, 100 nautical mile (115 miles) range, 15 knot max. speed (18 mph), 300 lb. payload. Primary role is to sense the environment (Maritime Domain Awareness). Other tasks are contested logistics (deliver cargo in dangerous environments), electronic warfare, and kinetic strike (including launch and retrieval of drone assets).
- Kaikoa: in sizes of 20, 30, 46 and 42 ft., these are diesel powered boats, with ranges from 150 – 500 nautical miles, 35 knot max. speed, and up to 4000 lb. payload. Kaikoa primary function is to to operate in conjunction with a fleet of Rampage ASVs, as an autonomous fleet support vessel.
- Atlas: 100 ft. length, diesel powered, 7400 nautical mile range, 25 knots max. speed, 36,000 lb. cargo capacity. Optimized for long endurance missions with containerized payload integration, including electronic warfare, fires, C-C5ISRT (Counter-Command, Control, Computing, Communications, Cyber, Intelligence, Surveillance, Reconnaissance, and Targeting) and contested logistics. C-C5ISRT disrupts, degrades, or destroys an adversary’s ability to command their forces, or gather intelligence. It also executes targeting, particularly in contested environments. Atlas is built in strategic collaboration with Lockheed Martin.
Figure 2 shows the three current ASV’s supported and sold by HavocAI.
To date, the company has built 80 ASVs and delivered 33 to customers (28 Rampage, 4 Kalkoa, 1 Atlas). The strategy is to procure these craft from commercial manufacturers and outfit them with HavocAI’s customized perception sensor suite (cameras, radar, LiDAR, ultrasonics, radar), navigation capability (IMUs, Starlink connectivity) and radios (for local communication). The software stack includes:
- Neural network-based perception and localization stacks
- Path planning and velocity control algorithms based on vehicle dynamic models. These problems are well-studied and proven, and HavocAI builds on these proven control, guidance, and hydrodynamic models that account for resistance, drag, thrust limits, and stability across operating envelopes, including at higher speeds.
- Tactical simulation (war games) and decision making software. This focuses on mission reasoning, tasking, coordination, and supervision across multiple platforms and domains.
To date, HavocAI has collected ~100B data points across these dimensions to train and test their systems. Rampage ASVs sell in the $100K range. CEO Paul Lwin indicated that costs scale exponentially with size and payload capacity, so that a 42 ft. Kaikoa is in the ~$1M range and an Atlas in the multi-M$ range. These are still much cheaper than other U.S Navy assets and therefore more expendable and deployable in large volumes.
An emerging focus for HavocAI is to design and build an ASV ground up, with purely autonomous operation in mind (Rampage, Kaikoa and Atlas are commercial vehicles designed for human operation, and retrofitted with HavocAI’s autonomy stack). Just like Zoox and Nuro designed AVs ground up with autonomous operation in mind, HavocAI is doing the same for ASVs – the belief being that this enables more optimized designs (weight, power, capacity, costs and performance).
HavocAI is working with Hanwha Defense USA, a unit of Hanwha, a Korean shipbuilder who already has operations in the United States (Hanwha Philly Shipyard in Philadelphia). Hanwha is the sixth largest shipbuilder in the world, and an investor in the current round. The goal of the collaboration with HavocAI is to build a 200 ft. ASV (as a comparison, the Statue of Liberty is ~300 ft. in length and a U.S. football field is 360 ft. in length). The motivation behind larger autonomous ships is logistics capacity – think of the the ability to ship 2 TEU (twenty foot equivalent unit) containers with autonomous assets like drones, quadcopters, missiles and autonomous terrestrial vehicles like ATVs and ground rovers) to the shores of contested mission areas, and then operating them as part of HavocAI’s collaborative autonomy stack. As opposed to manned large ships which take 6 years to build, are expensive (a frigate measuring 300-500 ft., costs ~$1B to build) and need expensive docking structures, this offers faster scaling and more cost effective capacity (estimated to be in the $10-50M range).
HavocAI’s autonomy stack works equally well on aerial (drones, quadcopters, missiles) and terrestrial (trucks, ATVs and rovers) autonomy assets, as evidenced by the demonstrations in Troia, Portugal (discussed above) and Geoje (discussed below). Per Mr. Lwin, “HavocAI is a collaborative autonomy company, and our software is running on aerial, ground and maritime vehicles. When I see a boat, I see it as a moving thing, which HavocAI enables collaboration with”.
Hanwha Defense USA – Advanced Ground and Sea Combat Systems
Hanwha Defense USA’s focus is on advancing the capabilities of ground combat forces through innovative and mission-oriented solutions. The company specializes in cutting-edge manned and unmanned combat systems, artillery, and advanced ammunition to enhance survivability and operational efficiency on the battlefield. It is a division of the Hanwha Group of South Korea, a seasoned builder of maritime vehicles for commercial and defense applications. The rationale for the partnership with HavocAI is to combine Hanwha’s shipbuilding and manufacturing capabilities with HavocAI’s collaborative autonomy technology, with potential large-scale production of autonomous surface vessels for defense applications. A rendering of the 200 ft. ASV is shown in Figure 3:
Hanwha acquired the Philly Shipyard location in December 2024 for $100M, the original goal being to leverage Hanwha’s maritime expertise to support production of nuclear powered submarines for the U.S. Navy. (Figure 4):
According to Hanwha Defense CEO, Michael Coulter, the Hanwha’s investment in the United States goes beyond expanding nuclear submarine capability for the U.S. Navy. The company’s focus on efficient and autonomous ship manufacturing and mature supply chain enables high volume, low cost delivery of commercial and defense maritime vehicles, in surface and sub-surface environments (currently 95% of commercial ship-building occurs in Asia). The goal is to bring this experience to bear in the United States.
The collaboration between Hanwha and HavocAI builds on a joint technology demonstration in October 2025 from Hanwha Ocean’s Geoje shipyard in Korea where Havoc AI conducted an autonomous force protection mission (while in Geoje) remotely off the coast of Hawaii, with beyond-line-of-sight command and control. The partnership with HavocAI on designing bottoms-up for autonomous operations builds on Hanwha’s experience in delivering ships with minimal human presence or control. Autonomy enables efficient use of space (facilities to support human presence on-board can be removed or eliminated), leading to leaner or more efficient designs. Autonomy for the U.S. Navy follows a “quarterback model” where a manned frigate or destroyer is accompanied by forward deployment of a large number of distributed autonomous asset. The Hanwha- HavocAI collaboration on the 200 ft, ASV enables this.
Per CEO Michael Coulter: “By forging a partnership between an allied defense company with advanced manufacturing scale in Hanwha with a software-first defense technology company in HavocAI, we will deliver state-of-the art ASVs at scale for American service members. This agreement will deliver for the American warfighter and create much-needed competition for the Department of War acquisition process”.
HavocAI and Hanwha are changing the dynamics of autonomy – from individual to collaborative, across multiple moving Things, with and without human participation. Collaborative autonomy is elevating AoT™ to new levels from an individual Thing to an Ecosystem of Things.
