Bringing Air-Gapped AI to the
Tactical Edge for Decision Dominance
By ANNE WAINSCOAT-SARGENT
Today’s modern militaries are experiencing exponential growth of data collected on the battlefield. Analysis by Deloitte’s Center for Government Insights finds that the next generation of wide-area motion imagery sensors is projected to be able to collect 2.2 petabytes of data each day, or more data than if someone recorded high-resolution video 24 hours a day, seven days a week for nearly seven years.
This data explosion, combined with the need for real-time analysis using techniques like artificial intelligence (AI), has outpaced traditional storage and cloud solutions, especially when the need involves analysis at the edge. That’s starkly proven in Ukraine, the site of the world’s first drone war: Since the war with Russia began, Ukraine has collected two million hours, or the equivalent of 228 years’ worth of battlefield drone video footage, to enhance battlefield decision-making and feed to AI models, reports Reuters.
“The amount of data being collected in the battlefield is fast outpacing and bandwidth that is physically possible in deployed locations,” said Raj Iyer, president of Public Sector Markets for Tsecond, a deep-tech startup offering compute and storage capabilities.
Iyer previously served as the U.S. Army’s chief information officer (CIO), where he set the strategic direction and oversaw policy and programs for the Army’s digital transformation. He noted the collection of massive data sets creates two challenges for military decision-makers: the ability to store so much data at the edge and the need to transfer the data back to the cloud for analysis and AI modeling, which requires connectivity.
“Connectivity is very hard to come by in constrained environments, even with the DOD leveraging Starlink and lower-Earth orbit (LEO) satellites for greater connectivity to the battlefield,” Iyer said.
Maj. Gen. Patrick Ellis, commander, 4th Infantry Division, outlined the imperative of getting more computing and storage capabilities to the tactical edge, in the defense publication, War on the Rocks. “The next conversation and experimentation focus will be about how much computing power and storage will be needed to process data at the edge: Instead of a prebuilt system or command post vehicle, the Next Generation Command and Control prototype components will be modular ‘building blocks’ that enable commanders at echelon to take apart the command post and put it in a vehicle, move it into a building, or disperse it in a tree line based on their mission and threat environments” Ellis wrote.
This new warfighting doctrine will require technology that can take data, compute and AI “on the go,” something the military has not had to be concerned about in the past with fixed command posts serving as hub for all compute. However, these fixed command posts are highly vulnerable against a near-peer adversary and validated most recently in the conflict in Ukraine.
Reimagining the Tactical Edge
Fortunately, this modular vision for command and control is feasible. Advances in data storage and processing can now deliver “air-gapped AI,” or analytics at the tactical edge with no need for network connectivity.
San Jose, California-based Tsecond specializes in innovation for edge infrastructure. With the company’s flagship product, BRYCK, a high-performance, mobile data solution, users can perform large data capture, data transport and AI in disadvantaged, harsh, forward, tactical, mobile and flying edge environments. In essence, it enables AI at the edge.
Petabyte-Level Storage and ML/AI Modeling
The device stores up to a petabyte of data, allowing users to collect, process and analyze massive amounts of sensor and intelligence data on-site, without relying on network connectivity. The BRYCK also comes with more than 120 machine learning models and integrated large-language models, enabling advanced analytics, object recognition and language translation on the device itself.
The BRYCK can process sensor data, recognize patterns and make split-second decisions without needing to send data back to a centralized ground network for processing, Iyer said. The solution not only increases operational speed and efficiency, but enables drones to operate effectively in contested or disconnected environments, supporting a wide range of autonomous missions.
“Because the data is on the device, you can run Gen AI on it and query it like you’re talking to ChatGPT,” he explained. “With a capability like this, users can completely reimagine what the tactical edge means.”
Driving Smarter, More Capable Autonomous Systems
BRYCK’s capabilities could enable U.S. and allied forces to enhance the capabilities of autonomous systems. Today, drones, whether deployed on the ground, in the air or as submersibles under the sea, rely on a camera feed that goes to a human operator, who makes decisions.
“That may be OK when you have one operator and one or five drones. But when you’re swarming and essentially have 100 drones together, all needing to work together, you’ll have to rely more on autonomous decision-making,” Iyer said.
Because BRYCK can easily plug into modern drones, it enables more intelligence directly on the platform. “Depending on the situation, the operator doesn’t have to be in the loop because of the ability to bring AI directly to the drone,” Iyer says.
Adding Agility in the Indo-Pacific Region
Tsecond’s technology is already being used in commercial and defense applications, with ongoing experimentation and deployment in critical regions like the Indo-Pacific, seen as a major threat for potential future conflict given China’s growing influence and military capabilities.
“This is going to make U.S. units much more expeditionary: units can bring their agility and rapid movement to the battlefield in any theater,” Iyer predicted. This capability directly supports decision dominance—faster insights to commanders with real-time AI thereby establishing stronger deterrence against near-peer adversaries.
Disclaimer: The link will be redirected to the Signal Magazine’s Premium Paid version and might require existing membership.