The Pentagon will spend $93 billion on autonomous weapons and supporting infrastructure over the next three fiscal years, according to budget documents submitted to Congress last month. Of that total, $54 billion flows directly to autonomous and remotely operated systems, while $39 billion targets the sensor arrays, communications networks, and edge computing hardware required to field them at scale. The figures represent actual procurement and research dollars, not aspirational planning numbers, and they eclipse the combined market capitalization of every pure-play defense robotics company traded publicly in North America.
Defense contractors saw the shift coming well before these budget documents landed. Anduril Industries raised $1.5 billion in August 2024 at an $8.5 billion valuation, with nearly half that capital earmarked for expanding production capacity for its Ghost and Altius autonomous platforms. Shield AI closed a $500 million Series F in October, bringing its total raise to just over $1 billion. Both companies are burning through that capital to build manufacturing lines capable of delivering thousands of units annually, not dozens. Traditional primes are reconfiguring as well. Lockheed Martin announced in January it would consolidate its unmanned systems divisions under a single business unit reporting directly to the CEO, a structural change that typically signals a permanent reallocation of engineering talent and capital expenditure. Northrop Grumman followed two weeks later with a similar reorganization.
The budget breakdown reveals where the Defense Department expects autonomous systems to deliver operational advantage first. Unmanned aerial systems account for $22 billion of the total, split between intelligence-surveillance-reconnaissance platforms and loitering munitions. Ground-based systems, including robotic combat vehicles and autonomous logistics convoys, receive $14 billion. Maritime domains get $18 billion, divided between unmanned surface vessels for mine countermeasures and extra-large unmanned undersea vehicles designed for anti-submarine warfare. The remaining allocation funds counter-drone systems, many of which rely on artificial intelligence for target identification in GPS-denied environments. Each category includes line items for both the platforms themselves and the software integration work required to connect them to existing command and control infrastructure. That second bucket matters more than most budget watchers realize. An autonomous drone is only as useful as the network that feeds it targeting data and receives its sensor feeds in return.
Investors tracking defense technology have already begun repositioning portfolios around these allocations. Venture capital flowing into dual-use robotics companies hit $4.2 billion in 2024, up from $2.1 billion the year prior, according to PitchBook data. That money is chasing companies with existing Pentagon contracts and clear paths to production scale, not science projects. The calculus is straightforward. Defense budgets move slowly, but once procurement priorities lock in, they generate revenue for a decade or more. A company that secures a position on an approved vendor list for unmanned ground vehicles today can expect follow-on orders through the early 2030s. The software side looks even more durable. Once a perception stack or autonomy engine integrates into a platform, replacing it requires extensive regression testing and recertification. Switching costs are high enough that initial design wins often become de facto standards.
What to Watch: The Army's Robotic Combat Vehicle program is scheduled to select two vendors for its Medium variant by June 2025, with a planned procurement of 500 units starting in fiscal 2028. The Navy will release final requirements for its Large Unmanned Surface Vehicle program in May, which should clarify whether autonomous maritime platforms need human oversight for weapon release decisions or can operate fully independently. Anduril's production facility in Ohio is expected to begin Ghost-X deliveries in Q3 2025, providing the first real-world data on whether venture-backed startups can manufacture complex autonomous systems at DoD scale and cost targets.
