Coby Adcock’s Scout AI raises $100 million to train its models for war. We visited its bootcamp.

Scout AI Secures $100M for Military Autonomy

Scout AI, a defense startup founded by Coby Adcock and Collin Otis, has raised a $100 million Series A round to develop autonomous systems for military applications. The funding, led by Align Ventures and Draper Associates, will accelerate the training of its core AI model, “Fury,” which is designed to operate military assets from logistical vehicles to autonomous weapons. The company is actively testing its technology on all-terrain vehicles at a U.S. military base, signaling a significant push to bring advanced AI directly into battlefield operations, moving beyond the structured environments where most autonomous vehicles currently operate.

Technical and Operational Details

At the heart of Scout's approach is a newer technology known as Vision Language Action (VLA) models, which are based on existing Large Language Models (LLMs). This allows the Fury model to translate general intelligence into physical actions, a capability CTO Collin Otis likens to how a human learns to control a drone by connecting prior knowledge to a new interface. The company's training operations, called Foundry, involve former soldiers driving ATVs on rugged terrain for eight-hour shifts to generate real-world data, which is then used to refine the model through reinforcement learning. This funding follows a $15 million seed round and complements $11 million in existing military development contracts.

• Company: Scout AI
• AI Model: "Fury" (Vision Language Action model)
• Series A Funding: $100 million
• Lead Investors: Align Ventures, Draper Associates
• Military Contracts: $11 million from customers including DARPA and the Army Applications Laboratory
• First Product: "Ox" command and control software

Ecosystem and Strategic Implications

Scout AI positions itself as a software-first company, aiming to build the intelligence layer for a wide range of military machines rather than manufacturing the hardware itself. Its first product, a command software called “Ox,” is intended to let soldiers orchestrate multiple drones and ground vehicles with simple commands. This places Scout in a growing but competitive field of off-road autonomy specialists, including Field AI and Overland AI, which also have roots in DARPA's RACER program. By openly embracing military applications, including one-way attack drones, Scout is also drawing a sharp contrast with some major AI labs like Google and Anthropic that have shown reluctance to engage directly in offensive military projects, positioning itself as a dedicated “frontier lab for defense.”

Scout AI's strategy isn't about perfecting an autonomous ATV; it's about using tangible hardware to train a universal, VLA-based command model. By focusing on a software intelligence layer ("Ox") designed to orchestrate diverse military assets, the company aims to become the essential operating system for robotic warfare, sidestepping the capital-intensive business of vehicle manufacturing.