Scout AI, a defense-focused startup co-founded by Coby Adcock and Collin Otis in 2024, has successfully raised $100 million in a Series A funding round. This considerable investment, led by Align Ventures and Draper Associates, is earmarked for advancing the company's AI models designed for military applications. The funding comes on the heels of a $15 million seed round secured in January 2025, marking a significant leap in financial backing for the burgeoning company.
Training AI for the Battlefield
The core of Scout AI's mission is to develop AI technologies that can be seamlessly integrated into military operations. At a confidential U.S. military base in central California, the company is rigorously training its AI models using autonomous all-terrain vehicles (ATVs) on challenging hillside trails. This initiative is not aimed at training human operators but instead focuses on enhancing the capabilities of AI systems to function effectively in conflict zones.
The central project, known as “Fury,” is a sophisticated AI model intended to operate and manage military assets. Initially designed for logistical tasks, the long-term goal is to extend its capabilities to autonomous weaponry. Collin Otis, Scout AI's CTO, likens the process to training soldiers, emphasizing the importance of building on existing knowledge to develop highly specialized military AI.
Collaborations and Contracts
Scout AI has already secured substantial military technology development contracts, totaling $11 million, from key organizations such as DARPA and the Army Applications Laboratory. These collaborations are pivotal as they integrate Scout's technology into the U.S. Army's operational framework. The 1st Cavalry Division at Fort Hood, Texas, has incorporated Scout AI's technology into its training cycles, with plans to deploy proven systems by 2027.
The testing ground at the unnamed base provides a realistic environment for Scout AI's operations team, composed of former military personnel, to simulate missions and refine their AI models. This hands-on approach is crucial for developing AI systems that can navigate unstructured and unpredictable terrains — a stark contrast to the more controlled environments where autonomous vehicles typically operate.
Innovative AI Technologies
At the heart of Scout AI's technological advancements is the use of Vision Language Action models (VLAs). These models, which emerged from Google's DeepMind in 2023, offer a cutting-edge approach to controlling robotic systems. VLAs have already shown promise in the robotics sector, inspiring startups like Physical Intelligence and Figure.AI, the latter led by Coby Adcock’s brother, Brett.
The application of VLAs in military contexts represents a significant leap forward. These models enable Scout AI to imbue military vehicles with a level of general intelligence akin to that required for navigating complex terrains, as demonstrated during a recent ATV training session. The vehicles, capable of autonomous control, exhibit behaviors such as adjusting speed and position based on the terrain, mimicking human drivers' cautious yet strategic decision-making processes.
Future Applications and Implications
Scout AI's vision extends well beyond the development of autonomous vehicles. The company aims to create a comprehensive intelligence layer for military machines. Its upcoming product, “Ox,” will bundle command and control software with robust hardware, enabling soldiers to coordinate multiple drones and ground vehicles through intuitive prompts.
This strategic focus on software over hardware allows Scout AI to remain agile and adaptable, enhancing existing military platforms rather than developing new ones. The potential applications are vast, from automated resupply missions to advanced reconnaissance and defense systems using intelligent drones.
Challenges and Opportunities
Despite the promising outlook, the deployment of VLAs in operational settings remains a frontier yet to be fully explored. According to Stuart Young, a former DARPA program manager, the technology is ripe for field experimentation but requires further refinement to meet the rigorous demands of military operations.
Scout AI's approach involves using real-world data gathered from extensive testing at their training facility, known as Foundry. Here, drivers work in shifts to push the ATVs to their limits, collecting valuable data to enhance the AI models through reinforcement learning. This iterative process is crucial for overcoming the current limitations of autonomous systems, particularly in off-road and unpredictable environments.
Moreover, Scout AI is venturing into drone technology, experimenting with vision language models to endow drones with intelligence for reconnaissance and defense tasks. The development of a system where groups of munition drones operate under a larger command platform illustrates the potential for AI-driven military strategies.
Looking Ahead
As Scout AI continues to refine its technologies, the implications for military operations are profound. The successful integration of their AI models could revolutionize how the military conducts logistics, reconnaissance, and combat operations, potentially reducing the risk to human soldiers and increasing operational efficiency.
The next steps for Scout AI involve scaling their technologies for broader adoption within the military. With significant funding secured and robust partnerships established, the company is well-positioned to lead the charge in AI-driven military innovation. As the landscape of warfare evolves, Scout AI's advancements could play a pivotal role in shaping the future of defense technology.