Smack Technologies

Market position

Smack is a comparatively new entrant in a defense-technology funding boom, described by one outlet as a firm most of Silicon Valley had not heard of a year earlier. It differentiates itself from vendors that fine-tune commercial LLMs and from platform-level autonomy providers, positioning around campaign-level rather than platform-level autonomy. Its rise is linked in reporting to Pentagon demand for a broader set of AI vendors following the March 2026 designation of Anthropic as a supply-chain risk.

Smack argues that defense AI cannot be produced by fine-tuning commercial large language models, and instead builds small, physics-based domain-specific models trained in a purpose-built environment that encodes tacit operator knowledge. It frames its scope as campaign-level "Intelligent Autonomy" — coherent action by a distributed force under degraded comms, resource constraints and an adapting adversary — rather than platform-level autonomy such as drone navigation or swarm coordination. Additional stated differentiators include operation in low-compute, disconnected environments, API-first/MOSA-based vendor-agnostic integration, and a founding and staff base of military operators alongside researchers and engineers.

Technology

Smack trains deep reinforcement learning models in a proprietary simulation environment intended to reflect the physics, decision spaces, constraints and human expertise of war. Because no usable real-world dataset exists for large-scale peer conflict, the environment itself generates millions of synthetic engagements to teach multi-agent behavior across space and time, with high-fidelity expert-defined trajectories from human domain experts and real-world data used to refine the synthetic manifold. The model architecture is described as proprietary, combining graph networks, diffusion models and transformers for tactical inference. A model-agnostic human-machine interface layer uses commercially available LLMs to power user-defined agentic workflows, grounded in a proprietary knowledge graph of entities, relationships and context, which calls Smack's decision models to generate and evaluate courses of action. Lightweight models plus proprietary hardware in the Alpha suite fuse local sensor data with campaign-level context for use without persistent cloud connectivity.

Go-to-market

Direct engagement with U.S. military customers via prototype agreements and prototyping contracts with organizations including the Joint Fires Network, the Marine Corps Warfighting Lab and the Navy. The company markets through its website (including demo booking), a journal/blog, and recruitment of military domain experts and leaders from across the services; it emphasizes vendor-agnostic, API-first, MOSA-based integration with existing Joint Force systems.

U.S. Department of War/Department of Defense organizations and the Joint Force, including the Marine Corps (Marine Corps Warfighting Lab), the Navy, and joint organizations such as the Joint Fires Network, as well as U.S. allies and partners.

Geography

Headquartered in Texas — press releases and coverage identify Austin, Texas as the headquarters — with offices in Washington, D.C. and El Segundo, California. The company says a dedicated Austin office supports its wearable hardware effort.

History

Smack Technologies was founded in 2024 by two Marine Special Operations (MARSOC) commanders. Press reporting indicates the Defense Department's March 2026 designation of Anthropic as a supply-chain risk and its removal from classified work led the Marine Corps and Navy to engage smaller vendors, driving spring 2026 discussions with Smack. In July 2026 the company won two prototyping contracts (Joint Fires Network and Marine Corps Warfighting Lab) and hired John Falcone to lead RLHF. On 18 August 2026 it announced a $61 million Series B, its second raise of 2026, bringing total capital raised to over $90 million. Headcount grew from 19 employees in April 2026 to 51 by August 2026, with a target of approximately 85 by the end of the year.

Risks & controversies

Reporting notes ethical and oversight concerns around software that recommends targets at machine speed, including the argument from ethicists and some lawmakers that faster targeting loops reduce the scope for meaningful human intervention. Smack states that a human remains in the loop, while skeptics argue such oversight can degrade into rubber-stamping rapidly produced outputs. Accountability for erroneous software outputs is described as an unresolved question across the sector. The company also faces concentration risk in U.S. defense demand and rapid-hiring execution risk, and it did not disclose the Series B valuation.