BitcoinWorld Scout AI Raises $100 Million to Train Autonomous Military Vehicles for War – Exclusive Bootcamp Visit Scout AI, a defense-focused startup, has raised $100 million in Series A funding to train its AI models for military operations. Bitcoin World visited its bootcamp at a US military base in central California. The company builds autonomous military vehicles using advanced AI. Its goal is to deploy these systems in conflict zones for logistics and combat support. Scout AI Raises $100 Million for Military AI Training Scout AI announced the $100 million Series A round on Wednesday. The funding is led by Align Ventures and Draper Associates. This follows a $15 million seed round in January 2025. The company calls itself a “frontier lab for defense.” It focuses on creating AI models that can operate military assets autonomously. The startup was founded in 2024 by Coby Adcock and Collin Otis. Adcock is also on the board of Figure AI, a humanoid robot company led by his brother Brett. Otis previously worked at autonomous trucking firm Kodiak. He says the limitations of existing autonomous systems in unpredictable environments drove him to start Scout. Scout AI’s core product is an AI model named “Fury.” This model controls and commands military vehicles. Initially, it will handle logistical support. However, the company plans to expand into autonomous weapons systems. The technology builds on existing large language models (LLMs). Inside Scout AI’s Bootcamp: Training AI for War Bitcoin World received an exclusive tour of Scout AI’s training operations. The company asked us not to name the military base. At the site, four-seater all-terrain vehicles (ATVs) roam hillside trails. These vehicles are part of a training exercise. The goal is to train AI models to navigate conflict zones. The training is led by former soldiers. They put the ATVs through simulated missions. Drivers work eight-hour shifts. They log moments when they had to take control from the AI. This data is used to improve the model through reinforcement learning. The company has been training for six weeks. Scout AI uses Vision Language Action models (VLAs). These are based on LLMs and control robots. Google DeepMind first released this technology in 2023. VLAs allow the AI to connect prior knowledge to new tasks quickly. Otis compares this to a human learning to fly a drone with a headset. It is not a big leap. Autonomous ATVs in Action I drove one of Scout’s ATVs on the rutty trails. The terrain was challenging: steep hills, loose sand, and confusing intersections. I am not an experienced driver but managed a fair attempt. This shows the kind of general intelligence the company wants in its models. I also rode in the ATV under autonomous control. The vehicle accelerated faster than a human might. It hugged the right on wider trails but stayed in the middle on narrow ones. When confused, it slowed down to think. This happened several times during a 6.5 km loop. The AI is still learning, but it shows promise. Military Contracts and Applications Scout AI has secured $11 million in military technology development contracts. Clients include DARPA, the Army Applications Laboratory, and other Department of Defense customers. It is one of 20 autonomy companies used by the US Army’s 1st Cavalry Division. The division uses these technologies during regular training cycles at Fort Hood in Texas. The expectation is that proven products will deploy with the unit in 2027. The first applications of ground autonomy will be automated resupply. This includes carrying water or ammunition to distant observation posts. In convoys, a crewed truck might be followed by six to ten autonomous vehicles. This saves human labor for more important tasks. Brian Mathwich, an active duty infantry officer, recalled a recent exercise in Alaska. He led a resupply convoy in total darkness and wished for autonomous vehicles. The Technology Behind Fury: VLAs and LLMs Scout AI’s approach differs from traditional autonomous systems. Most autonomous cars operate in structured environments with rules. Operating off-road on unmarked trails is a different challenge. Otis realized this when he saw that his system at Kodiak was not intelligent enough for a war zone. VLAs offer a solution. They are based on LLMs and used to control robots. Scout AI uses existing LLMs as a base but declined to name which ones. Otis says the company has agreements with “very well known hyperscalers.” It also uses deterministic systems and other AI flavors to round out capabilities. The company expects to build its own model from the ground up in the future. Much of the new capital will go into training and compute costs. Otis believes Scout could beat existing leaders to AGI because its model interacts with the real world constantly. Scout AI’s First Product: Ox Scout AI sees itself as a software company. It builds an intelligence layer for military machines. It does not intend to make the vehicles themselves. Its first product is called “Ox.” This is a command and control software bundled on hardened computer hardware. It includes GPUs, communications, and cameras. Ox allows individual soldiers to orchestrate multiple drones and autonomous ground vehicles. Commands are prompt-like: “Go to this waypoint and watch for enemy forces.” This makes it easy for soldiers to use the technology in the field. Autonomous Weapons: A Controversial Path Scout AI is also working on autonomous weapons. It is testing drones for reconnaissance and as weapons. The drones use vision language models for intelligence. One system involves groups of munition drones flying with a larger “quarterback” platform. This platform provides more compute resources to command them. In one mission, drones would search for hidden enemy tanks and attack them. This could happen without human intervention. Otis argues that this is more precise than indirect artillery fire. Autonomous weapons are a flash point in defense tech politics. Experts note that the concept is old: heat-seeking missiles and mines have been in use for decades. Jay Adams, a retired US Army Captain who leads Scout’s operations team, says the key is how weapons are controlled. The company’s munition drones can be programmed to attack only threats in a specific geographic area. They can also require human confirmation. Adams says autonomous weapons are unlikely to fire because they are scared, unlike a young soldier. VLAs for Better Targeting VLAs offer promise for better targeting. Scout says its models are pretrained on military data. This prepares them for scenarios like running into an enemy tank during a resupply mission. Lt. Col Nick Rinaldi, who supervises Scout’s work for the Army Applications Laboratory, says automated targeting is hard. It is unlikely to be used outside constrained environments in the near term. However, the potential of VLAs to reason about threats makes them a promising technology to investigate. Scout AI’s Mission and Challenges Scout AI wears its mission on its sleeve. Executives criticize companies that are reluctant to work with the government. Google reportedly pulled out of a Pentagon contest for autonomous drone swarm control. Scout is working on similar capabilities. Otis says, “The AI people don’t want to work with the military.” He references Anthropic’s spat with the Pentagon over terms of service. He adds, “None of them are open to running agents on one-way attack drones.” Despite this, Scout uses existing LLMs as a base. It declined to say if it uses open-weight models from Chinese companies. The company expects to address this by building its own model. The founders say much of the capital will go into training and compute costs. Otis wonders if Scout will beat existing leaders to AGI because its model constantly interacts with the real world. Conclusion Scout AI’s $100 million raise marks a significant step in military AI development. The company is training its Fury model to operate autonomous vehicles in conflict zones. Its bootcamp at a US military base shows the practical challenges of this work. The technology uses VLAs and LLMs to provide general intelligence for military assets. The first applications will focus on logistics and resupply. However, the company is also working on autonomous weapons. This raises ethical and practical questions. Scout AI’s approach is controversial but driven by a belief that autonomous systems are necessary for future warfare. The company’s success will depend on its ability to train models that are reliable, safe, and effective in unpredictable environments. FAQs Q1: What is Scout AI? Scout AI is a defense startup that builds AI models for autonomous military vehicles. It was founded in 2024 by Coby Adcock and Collin Otis. Q2: How much funding has Scout AI raised? Scout AI raised a $100 million Series A round in 2025, led by Align Ventures and Draper Associates. It also raised a $15 million seed round earlier in the year. Q3: What is the Fury AI model? Fury is Scout AI’s AI model that operates and commands military assets. It is built on Vision Language Action models and large language models. Q4: What are the first applications of Scout AI’s technology? The first applications are automated resupply missions. This includes carrying water or ammunition to distant observation posts. The technology will also be used in convoys with autonomous vehicles. Q5: Is Scout AI working on autonomous weapons? Yes, Scout AI is developing autonomous weapons systems. It is testing drones for reconnaissance and attacks. The company says these systems can be programmed to require human confirmation or operate within specific geographic areas. This post Scout AI Raises $100 Million to Train Autonomous Military Vehicles for War – Exclusive Bootcamp Visit first appeared on BitcoinWorld .
