Scout AI
Scout AI builds Fury, a vision-language-action foundation model for unmanned warfare, and raised a $100 million Series A to put one autonomy layer across ground vehicles from multiple manufacturers.
Background
Scout AI was founded in 2024 by Colby Adcock, chief executive, and Collin Otis, chief technology officer, and is headquartered in Sunnyvale, California. It describes itself as a frontier AI lab for US defence.
The company emerged from stealth in April 2025 with a $15 million seed round and the unveiling of Fury, which it presents as the first vision-language-action foundation model built for defence robotics.
A VLA model connects visual perception and natural-language instruction to physical action, and is trained on human imitation and simulation data rather than hand-written rules for each platform — the approach that produced general-purpose commercial robotics models, applied to military vehicles.
In February 2026 Scout introduced the Fury Autonomous Vehicle Orchestrator, extending the model from controlling a vehicle to coordinating several while keeping a human at the centre.
The company's stated ambition is one-to-many human-to-robot coordination across land, air, sea and space.
Financial footing
- $15M seed (Apr 2025)
- $100M Series A (Apr 2026) co-led by Align Ventures and Draper Associates
- ~$11M in DARPA, Army Applications Laboratory and other DoD contracts
- Army xTechOverWatch winner
- Selected with Forterra and Overland AI for Army ISV autonomy
- Backers: Align Ventures, Draper Associates, Booz Allen Ventures, Decisive Point
Scout AI raised a $15 million oversubscribed seed round in April 2025 led by Align Ventures and Booz Allen Ventures, the corporate venture arm of Booz Allen Hamilton.
On 29 April 2026 it announced an oversubscribed $100 million Series A co-led by Align Ventures and Draper Associates, with participation from Booz Allen Ventures, Decisive Point, BVVC, Neman Ventures, Evolution VC Partners, Heraclitus Capital Management, Sigmas Group, Disruptive Founders Fund and Vaughn Capital Partners. Total disclosed funding is therefore about $115 million.
No valuation has been published. Government revenue is small relative to the raise: roughly $11 million in development contracts from DARPA, the Army Applications Laboratory and other Department of Defense customers. The Series A is priced on the model, not the contract book.
Major customers & contracts
Scout AI's disclosed customers are US government research and acquisition organisations: DARPA, the Army Applications Laboratory and other Department of Defense customers, totalling approximately $11 million in defence technology development contracts.
The company said it had landed two Department of Defense contracts at the time it left stealth.
It was named one of twenty winners of the Army's xTechOverWatch competition. In late August 2025 the Army selected Scout AI together with Forterra and Overland AI — both tracked separately in this directory — to receive $15.5 million in total to rapidly integrate their commercial autonomy onto existing Infantry Squad Vehicles and deliver prototypes for soldier demonstration. No production contract has been announced.
Leadership & contact surface
Leadership: Colby Adcock
Official site: https://scoutco.ai
Competitive position
Scout AI competes directly with Forterra and Overland AI for Army ground autonomy, and with Applied Intuition and Scale AI on the machine-learning layer of the Robotic Combat Vehicle effort.
Its differentiation is architectural: rather than a perception-and-planning stack tuned per platform, Fury is a single foundation model intended to generalise across vehicle types, which if it works collapses the integration cost that has made military autonomy slow and vehicle-specific.
Against Anduril's Lattice and Palantir's software, Scout is a layer lower — it drives the machine rather than orchestrating the battle picture — and does not carry their incumbency.
Against Kodiak AI, it lacks a commercial business subsidising the R&D but is purpose-built for off-road and contested conditions rather than adapted to them.
The competitive risk is that VLA models are the current consensus approach in robotics generally, so Scout's method is not proprietary; what would be defensible is the defence-specific training data, and that accrues to whoever fields the most vehicles.
Scout has raised $115 million against competitors with fielded systems and, in Applied Intuition's case, far deeper resources.
Battlefield relevance
No confirmed combat use. Scout AI's work is at prototype and demonstration stage, including autonomous military ATVs operated at its own training events and the Army Infantry Squad Vehicle integration effort delivering prototypes for soldier demonstration.
Sources & verification
- https://techcrunch.com/2026/04/29/coby-adcocks-scout-ai-raises-100-million-to-train-models-for-war-we-visited-its-bootcamp/
- https://www.prnewswire.com/news-releases/scout-ai-raises-100m-series-a-to-build-the-ai-brain-for-unmanned-warfare-302756871.html
- https://www.washingtontechnology.com/companies/2026/04/scout-ai-fetches-100m-series-capital/413233/
- https://www.axios.com/2025/04/16/scout-ai-military-autonomous-fury
- https://www.prnewswire.com/news-releases/scout-ai-introduces-fury-autonomous-vehicle-orchestrator-302691787.html
- https://breakingdefense.com/2025/09/army-shaking-up-new-autonomy-initiative-pivoting-away-from-isvs-sources/
- https://nominalinc.substack.com/p/mission-brief-scout-ai
- https://scoutco.ai/