Computing

Outlining a computing model for defense tech.

Intelligence and defense technology in the next 15 years will adopt a computing architecture, ushering in an era of interoperability, abstraction, and networking.

This is the premise of Software-Defined Warfare, a September 2022 report by Nand Mulchandani (CTO @ CIA).

Porter Smith and David Ulevitch at Andreessen Horowitz extended this in April 2023 in How the U.S. Can Rewire the Pentagon for a New Era, bridging the ideas introduced in Software-Defined Warfare and linking economic, organizational, and tactical incentives.

I’ve been thinking of a simplified version lately: a lens to analyze new intelligence and defense startups that is both:

• explainable in one minute and
• comprehensive enough to zero in on the right questions.

This is what I’ve got, my simplified defense tech computing architecture:

Simplified computational architecture for intelligence & defense tech.

Peripherals: stuff that plugs in with a common interface, eg: Sentry.

Interfaces: a common interface standard, eg: MOLLE/PALS webbing.

Processing: the body that does the thinking, eg: Colossus.

Logic: the thinking, eg: SAMWISE.

Networking: communications between systems, eg: ARPANET.

Yes, it’s over-simplified, and that’s the point. I use this to roughly and quickly consider how a new idea may fit into my vision for the future of intelligence & defense tech. For a particular company, this helps me see areas of strength, opportunity, or differentiation, as well as dependencies with what already exists.

Two notes about this lens:

1. You need to buy the idea that “all models are wrong but some are useful.” Without that belief in your bones, you will think your way out of every model out there, including this one.
2. You need to believe in computing architecture as a useful analog for the future of intelligence and defense technology. This model simply will not square with the historical “space, air, land, or sea” categorizations.

To give a few examples of how the model may work, let’s take six defense tech companies: Anduril, Helsing, Lambda Automata, Tune Insight, Labrys, and Origin.

Simplified defense tech computing architecture applied to a few companies.

This quickly illuminates strengths, opportunities, dependencies, and weaknesses in not only the companies themselves, but also in the overall market.

The model will morph a bit over time, but so far it’s proving useful.