AINEWS
Taiwan's state shipbuilder CSBC and Rhode Island's Havoc AI are pairing autonomy software with locally built unmanned attack boats, sketching a template allied navies could copy.
Taiwan's navy plans to buy roughly 1,320 small, crewless attack boats for about $930 million (around NT$29.6 billion), a bet that mass and software can substitute for matching China's fleet hull-for-hull.
"American autonomy software, Taiwan-built platforms, Taiwan workforce," is how Havoc AI CEO Paul Lwin described the division of labor when his Rhode Island firm and Taiwan's state-backed shipbuilder CSBC announced a partnership in September to co-produce autonomous surface vessels. CSBC supplies the hulls, the shipyard capacity, and lifecycle support. Havoc supplies the software stack that lets the boats navigate, sense, and coordinate without crews. The agreement is for licensed co-production in Taiwan, not imported boats, and it routes through Taiwan's National Chung-Shan Institute of Science and Technology, the Defense Ministry's research arm that gates foreign defense technology onto local platforms.
The $930 million figure and 1,320-vessel target come from Nikkei Asia's reporting on the broader Taiwan sea-drone program, not from the partnership announcement itself, and they describe a planned program, not a contracted one. The CSBC-Havoc deal is the first named, concrete instance of U.S. autonomy software being married to Taiwanese shipbuilding at scale. The dollar figure and vessel count belong to the wider procurement plan that sits behind it.
Aggregator coverage has tended to lead with Chinese maritime pressure and then list U.S. vendors. The actually interesting move is the formula.
A growing list of American defense-technology firms, including Havoc for autonomy, Saronic and Anduril for additional platforms and integration, MARTAC and VATN Systems for high-speed unmanned craft, Shield AI for autonomous flight and maritime autonomy, Auterion for open-source flight software, and Albacore, is being threaded into Taiwan's existing shipbuilding base. Taiwan contributes hulls, electronics, and yards: CSBC for larger surface vessels, Jong Shyn and Lungteh for fast attack craft, plus established firms like Thunder Tiger and Myson Century moving in from the drone and semiconductor worlds. The model is layered and asymmetric, not a single-vendor sale.
The Strait context is real. China's navy is the largest in the world by hull count, and a blockade scenario is the planning case Taiwan's defense planners openly discuss. Unmanned surface vessels, small crewless boats often under 50 feet carrying sensors, communications gear, or weapons, are one of the few categories where a smaller navy can field credible mass quickly and cheaply, and where attrition matters less than the ability to replace losses. The bet is that 1,320 small boats, networked and software-driven, are a more usable deterrent than a handful of expensive crewed corvettes.
Smart Maritime Network's earlier coverage of the partnership, published September 11, framed the same split (software from Havoc, construction and sustainment from CSBC) and is useful neutral corroboration of the division of labor, though not of technical performance.
The release is a statement of intent, not a delivery schedule. No vessel cost, production rate, autonomy performance, communications resilience in a contested environment, or weapons integration has been independently verified. The deal is real. The deployed fleet is not.
The reusable template only matters if the first instance works. A September partnership announcement between a Rhode Island software firm and a Taiwanese state shipbuilder is not yet a deployed fleet, a tested autonomous stack, or a contested-communications demonstration. The next-quarter question is whether the CSBC-Havoc deal produces a hull on a launching ramp, a software drop that runs on it, and a test where both work under jamming, or whether it joins the long list of allied defense MOUs that look like capability on paper and deliver little.
If it works, the formula travels. The Philippines, Vietnam, and Japan all run into the same math problem Taiwan does, with smaller fleets next to larger ones. A template that puts American autonomy software on locally built, locally crewed, locally maintained hulls, rather than exporting finished U.S. platforms the partner cannot sustain, is the kind of industrial model that compounds. The next quarter's milestones, not the press release, will tell.