Scout AI Secures $100 Million Series A to Train Autonomous Military AI Models — Inside Its Bootcamp

At a U.S. military base in central California, four-seat all-terrAIn vehicles navigate rugged hillside trails as part of a training exercise — not for the human passengers, but for the Artificial Intelligence models designed to operate in conflict zones. These autonomous military ATVs are developed by Scout AI, a defense-focused startup founded in 2024 by Colby Adcock and Collin Otis, which describes itself as a "frontier lab for defense." The company announced Wednesday that it has raised $100millioninaSeriesAroundledbyAlignVenturesandDraperAssociates,followinga$15 million seed round in January 2025.

Scout AI invited TechCrunch for an exclusive look at its training Operations on a military base it requested not be named.

The company is building an AI model called "Fury" to operate and command military assets, initially for logistical support and soon for autonomous weapons systems. CTO Collin Otis likens the APProach — building on existing large language models — to training soldiers. "Soldiers start when they're 18 years old, and sometimes they even start after college, so you want to start with that base level of Intelligence," Otis said. "It's useful to start with someone who's already made an Investment and then say, 'What do I have to do to teach this thing to be an incredible military AGI, versus just being a broadly intelligent AGI?'"

Scout has secured military Technology development contrACTs totaling $11 million from organizations including DARPA, the Army Applications Laboratory, and other Department of Defense customers. The company is among 20 autonomy firms whose technology is being tested by the U.S. Army's 1st Cavalry Division during its regular training cycle at Fort Hood, Texas, with the expectation that proven products will be deployed with the unit in 2027.

For internal testing, Scout's operations team — led by former soldiers — puts the vehicles through simulated missions on the base's challenging terrain. While autonomous vehicles are increasingly appearing in structured urban environments, navigating unmarked trails or off-road conditions presents an entirely different challenge. Otis, who previously worked at autonomous trucking company Kodiak, said he was motivated to found Scout after realizing the system he helped build wasn't intelligent enough for unpredictable war zone conditions.

A New Approach to Autonomy

Scout is adopting a newer autonomy technology: Vision Language Action models, or VLAs, which are built on Large Language Models and used to control robots. First released by Google DeepMind in 2023, the technology has seeded robotics startups such as Physical Intelligence and Figure.AI, the humanoid robot company led by Adcock's brother, Brett.

Colby Adcock serves on Figure's board and says that experience convinced him of the opportunity to bring broader intelligence to the military's growing fleet of autonomous vehicles. His brother introduced him to Otis, who was advising Figure, and they set about applying the latest AI advances to military solutions.

"If I handed you the controller of a drone right now and I strapped a headset on you, you could learn to fly that thing in minutes," Otis said. "You're actually just learning how to connect your prior knowledge to these couple little joysticks. It's not a big leap. That's the way to think about VLAs and why they're such an unlock."

During the visit, a TechCrunch reporter drove one of Scout's ATVs along rutted trails featuring steep hills, loose sand, disappearing tracks, and confusing intersections — challenging terrain even for experienced drivers. That kind of General intelligence is what the company seeks in its models, which have been training on these ATVs for just six weeks after starting with civilian vehicles.

Riding in the ATV under autonomous control revealed distinct differences — the vehicle accelerates faster than a human driver mindful of passenger comfort. The operations team noted how the vehicles hug the right side on wider trails but stay centered on nARRow ones, mirroring their human trainers. When uncertain, the ATV would suddenly slow to process its next move, which occurred several times during a 6.5-kilometer loop before returning to base.

Though VLAs are new enough that no company has yet deployed them operationally, "the technology is good enough to be doing that experimentation in the field with soldiers to figure out how to most effectively support U.S. forces," said Stuart Young, a former DARPA program manager who worked on ground vehicle autonomy. Scout's full technology stack also includes deterministic systems and other AI approaches to round out its Agents' capabilities. Young left DARPA this month to join Field AI after managing the RACER program, which challenged companies to create high-speed autonomous off-road vehicles — helping seed this space much as DARPA's Grand Challenge boosted self-driving cars.

Adding Intelligence to the Army's Motor Pool

The first applications of ground autonomy, according to Scout executives and military technologists, will be automated resupply: carrying water or ammunition to distant observation posts, or enabling convoys where a single crewed truck leads six to ten autonomous vehicles, saving human labor for more critical tasks. Brian Mathwich, an active-duty infantry officer serving as a military fellow at Scout, recalled a recent training exercise in Alaska where he led a resupply convoy in total darkness and wished for autonomous vehicle support.

Scout views itself primarily as a software company building an intelligence layer for military machines, rather than manufacturing the vehicles themselves. Adcock expects the startup's first widely adopted product to be "Ox," a command and control software bundled with hardened computer hardware including GPUs, communications systems, and cameras. It is designed to allow indiVidual soldiers to orchestrate multiple drones and autonomous ground vehicles using Prompts such as, "Go to this waypoint and watch for enemy forces."

Making that software work requires real-vehicle training, which is why Scout established "Foundry," its training range at the military base. Drivers spend eight-hour shifts operating the ATVs, then work through a reinforcement learning system to log moments requiring human intervention, data used to improve the model. The base commander has even requested Scout's ATV assist with security patrols.

One hypothesis Scout is testing: whether VLAs can deliver a fully capable driving agent using this relatively limited dataset combined with simulation training. While the vehicle navigates trails comfortably, it is not yet ready for fully off-road operation.

Scout is also practicing with drones for reconnaissance and defense, equipping them with intelligence through vision language models. The startup is developing a system in which munition drones fly alongside a larger "quarterback" platform providing enhanced computing resources to command them. For example, drones could search a geographic area for hidden enemy tanks and attack them, potentially without human intervention. Otis argues the alternative approach in such scenarios — indirect artillery fire — lacks the precision of drone strikes.

While autonomous weapons remain a political flashpoint in defense technology, experts note the concept is decades old: heat-seeking missiles and mines have been used in warfare for years. The key question for technologists is how weapons are controlled, according to Jay Adams, a retired U.S. Army captain who leads Scout's operations team. Adams notes the company's munitions drones can be programmed to engage threats only within specific geographic areas, or only following human confirmation. He adds that autonomous weapons platforms are unlikely to fire out of fear, unlike an 18-year-old soldier.

VLAs also show promise for improving targeting. Scout says its models are pre-trained on specific military data to handle scenarios like encountering an enemy tank during a resupply mission. Lt. Col. Nick Rinaldi, who oversees Scout's work for the Army Applications Laboratory, says that while automated targeting remains challenging and unlikely to be deployed outside constrained environments in the near term, the potential of VLAs to reason about threats makes them a promising area of investigation.

Adams contends that drones capable of identifying their own targets will be critical to future warfare. While Russia's invasion of Ukraine has generated intense interest in drone combat, he believes the human operation of individual UAVs won't scale sufficiently if U.S. forces face large numbers of low-cost unmanned systems.

A Mission to Counter Anti-Military Sentiment in Tech

Like many defense startups, Scout openly embraces its mission, and its executives freely criticize companies reluctant to supply technology to the government. Google, for instance, reportedly withdrew from a Pentagon competition to develop control systems for autonomous drone swarms — a capability Scout is also pursuing.

"The AI people don't want to work with the military," Otis said, referencing anthropic's dispute with the Pentagon over terms of service. "None of them are open to running agents on one-way attack drones, or running agents on missile systems."

Nevertheless, Scout uses existing large language models as the foundation for its agents, though it deCLIned to specify which ones. Otis says the company has agreements with "very well-known hyperscalers" to provide pre-trained intelligence for Scout's Foundation Model. He also declined to say whether Scout uses open-weight models, such as those offered by Chinese companies. Many AI-dependent firms build on open-weight models because they are more affordable than offerings from frontier labs like Anthropic or OpenAI.

Scout plans to build its own model from the ground up in the coming years, and the founders say much of its capital will go toward training and computing costs. Otis even wonders whether Scout might beat existing leaders to artificial general intelligence because its model will continuously interact with the real world. "There's an argument in the AGI community that you can only get so intelligent by reading the internet, and most intelligence comes from interacting in the world," he said.

Does this mean Colby Adcock is competing with his brother's army of humanoid robots at Figure? No, Otis says: "We can get to scale much faster because our customer has assets," referring to the Pentagon.