It’s been a busy, beefy week for the Pentagon and a certain AI giant, but a startup out of El Segundo has a solution to those tensions: Just build an AI lab purpose-built for national security. And investors are on board.
Yesterday, Smack Technologies announced a combo Seed and Series A round to build what they call the “first frontier AI lab for national security.” Smack’s initial seed round was led by Point72 Ventures, and the Series A was led by Geodesic Capital and Costanoa Ventures.
Talking smack: Right now, decision-making at both the command and operational level is pretty antiquated (lots of whiteboards and PowerPoint) and siloed between time horizons and warfighting functions, like fires, intel, logistics, and force protection.
Smack, founded in 2024 by former MARSOC operators Andy Markoff (CEO) and Clint Alanis (COO), aims to break that down and help bring decision-making up to the pace of operations.
It’s all Greek to me: Smack’s warfighting domain-specific AI reasoning models, Alpha and Omega, are designed to evaluate millions of scenarios across different time horizons and warfighting functions to make planning recommendations based on encoded human domain expertise and military training.
- Omega: Smack’s command-level AI stack that combines multimodal data fusion, analysis, and reasoning to speed up the Orient and Decide stages of the Observe, Orient, Decide, Act (OODA) Loop, according to Markoff. It’s built to run on desktops in operations centers and is moving from the prototype stage to production.
- Alpha: Their edge-level AI model designed to run on tablets and edge devices to offer immediate decision-making support in the field. It’s still in the early stages, but they’re “actively pursuing prototyping contracts for Alpha for fires and maneuver at the edge,” he added.
“The amount of purposely injected misinformation in a modern peer-level conflict, along with the cognitive overload for human decision-makers up and down the chain, is just too high,” Markoff told Tectonic. “You’re going to see a bottleneck at the Orient and Decide phase, and if you can speed up that part of the loop, our argument is that that side will win.”
Beefy business: In case you missed the whole Pentagon-Anthropic saga last week, Secretary Hegseth and Pentagon CTO Emil Michael are pretty hell-bent on turning the US military into an “AI-first warfighting force.”
Smack’s argument is that the frontier AI models, like Anthropic’s Claude and OpenAI’s ChatGPT, are “tuned to write LinkedIn posts and fill grocery lists, [not] fight World War III,” Markoff said. “We’re building domain-specific frontier models that have one job: to deter a peer-level conflict.”
So far, Smack’s idea of a warfighting-specific AI lab has been well-received in the AI-frenzied Pentagon. They’re currently under contract with the “Air Force, Marine Corps, and the Navy,” and “two of those contracts are actively moving from prototype to production contracts,” Markoff said.
In the lab: With $32M in fresh capital, Smack is focused on bringing in more “AI research and engineering horsepower to expand and release new versions of the model faster, develop the Alpha edge variant of of the model, and [build] a smaller, lightweight derivative of Omega with hardware built around it to enable it to connect into the physical world at the edge.”
That all sounds like music to Mr. Michael’s ears.