I/The next war is offline
For three decades, American technology has built intelligence the same way. That architecture is about to fail.
A model lives in a data center. A soldier carries a phone. A fiber line connects them. This works in San Francisco. It does not work over the Taiwan Strait, or the Black Sea, or the South China Sea, or any meaningful distance from a happy fiber backbone.
The next war will be fought against an adversary who can sever the link, and will. We have spent twenty years building a force whose intelligence is hostage to its own infrastructure. That bill is about to come due.
II/Cloud intelligence is a single point of failure
Every modern AI system shares the same brittleness. It needs a conversation.
The model is somewhere else. The reasoning is somewhere else. The memory is somewhere else. When the link goes down, the system degrades to a calculator. Against a peer adversary, the link goes down on day one.
The drone forgets what it was looking at. The radio forgets how to translate. The watchstander loses access to the mind he was supposed to share. None of the systems we currently rely on were designed for the environment they will actually operate in.
III/Intelligence belongs on the platform
The fix is not bigger satellites or harder waveforms. The fix is to put the mind on the thing.
Inside the drone. Inside the radio. Inside the buoy. Inside the helmet. A system that perceives, decides, and adapts on its own substrate, with its own memory, in real time, with no expectation that the network will be there.
This is what we mean by on-device intelligence. It is not a smaller version of cloud AI. It is a different kind of system entirely, designed from the first principle that the link is gone.
IV/We don’t build models. We build something that holds.
The hardest engineering problem in this space is not throughput or accuracy. It is continuity.
A system that degrades gracefully. That keeps reasoning when half its inputs disappear. That survives a long deployment in a sealed enclosure with no firmware update. That recovers from things it was never trained on. This is a different discipline than training a model. It looks more like building a mind than fitting a curve.
We have been working on it for years. Most of the work is not yet public, and will not be public for some time.
V/Place matters
Formenos is on the Gulf Coast, between the largest shipbuilder in America and the schoolhouse of naval aviation.
Our neighbors are the people who actually take this hardware to sea. Distance from Silicon Valley is a feature. We are closer to the operators than to the investors, closer to the salt than to the screen. The work benefits.
Defense technology built in coastal industrial cities by people who can drive to a hangar in twenty minutes is going to look different from defense technology built in office parks two thousand miles inland. We think it should.
VI/Quiet is a design choice
Most of what we build will not be announced.
There will not be product launches. There will not be press releases for individual contracts. We will not tell you the size of the team, or the architecture of the system, or the names of the customers. This is not modesty. It is a posture.
The systems that matter most do not draw attention to themselves, and neither do the people who build them. If you need to know what we do, you will already know how to find us.
Founder · Formenos
Gulf Coast · 04.07.2026
Return to Operations