Collaborative autonomy: Why software will define the future of defense

It is a departure from today's model, in which a single operator controls a single vehicle, an approach that breaks down when communications or GPS are denied. By making the systems truly autonomous, Havoc removes that dependency.

A live demonstration, 2,000 miles away

Rather than describe the technology, Lwin showed it. His chief of staff, working from the stage in Deer Valley, Utah, commanded five sea vessels stationed in Rhode Island, roughly 2,000 miles away. Selecting a formation and two points on a map, she moved the fleet into a new blockade position in under 15 seconds, with no one physically in the loop. She then drew a box on the screen and instructed the vessels to patrol the area; they divided the task among themselves and coordinated directly with one another.

The point was not the boats. It was that the intelligence guiding them lives entirely in software, allowing commercially available hardware to operate as a system in concert. As Lwin put it, the systems are decentralized, with no single leader to target: each vessel receives the intent and negotiates its own role.

Watch a short recap of Paul Lwin’s demonstration at the Technology Leadership Forum

The economics of software-defined hardware

The problem is not hardware. The strategic implication is cost. Because the differentiation sits in software, Havoc pairs its proprietary systems with commercial off-the-shelf components, keeping platforms reliable and inexpensive.

The economics are central to Havoc's approach. Its vessels sell for roughly $200,000, priced more like munitions than traditional ships. Lwin noted that a Hellfire missile, a precision-guided weapon commonly used by the U.S. military, costs approximately $175,000 each, underscoring his belief that autonomous systems must be affordable enough to deploy at scale. When an attendee raised the threat of an electromagnetic-pulse weapon, Lwin pointed to the cost asymmetry: if an adversary spends $10 million on an EMP to disable a fleet of $200,000 boats, Havoc still wins the exchange.

That same discipline shows up in construction. Where competitors have raised hundreds of millions to build shipyards, Havoc partners with existing shipbuilders – he discussed Havoc’s partnership with PacMar Technologies in Hawaii as an example, and how, with a team of five engineers and less than $1 million, the companies put a 100-foot ship in the water before rivals who had spent far more. The company has even 3D-printed a 40-foot vessel in about a month. The model borrows the economics of software, high margins and rapid iteration, and applies them to a category long defined by heavy capital intensity and slow procurement.

Proof of security, not just certification

Credibility in defense requires more than a security checkmark. Lwin recounted placing Havoc's entire technology stack in a sandbox at Black Hat, the major hacking conference, and inviting world-class penetration teams to break in. For four days (an eternity for the best hackers), they could not. On the final day, even after being allowed to hardwire directly into the boats, they could only monitor traffic; they could never take control. It was, he suggested, the difference between claiming to be secure and proving it.

A founder shaped by service

The mission behind the company is personal. An immigrant who arrived in the United States at age 10, Lwin went on to graduate from the U.S. Naval Academy, flew a 40-year-old aircraft over Iraq and Afghanistan, and became a Navy test pilot developing next-generation systems. He founded Havoc in 2024 with his co-founder Joe, also a Navy veteran, after watching existing players focus on building hardware while the harder problem, coordinating thousands of systems, went unsolved.

He frames the purpose around deterrence rather than conflict, describing a goal of fielding enough autonomous capability in the Pacific to change an adversary's calculus.