Defense startups trapped by services
The biggest mistake defense startups make
This is the trap that separates a scalable defense product company from a well paid engineering shop. To win early contracts, startups often agree to custom work tied to one program, one customer, or one vehicle, and that pulls engineers into bespoke integrations, support, and compliance work that does not repeat. Anduril’s edge was funding product upfront, then selling a working system through commercial style procurement, while Forterra had to shut down 17 legacy contracts to get back to one repeatable autonomy stack.
-
In practice, services creep starts with small contracts. A startup takes a $100K to $2M project to prove traction, then gets contractually locked into customer specific features, demos, and field support. That can tear engineering away from the core product and make scaling impossible.
-
The money model is the real issue. Cost plus and time materials contracts scale by adding headcount and usually produce prime like margins around 8% to 10%. Product companies self fund R&D, set a fixed price, and can build much higher margin recurring revenue if the same system ships again and again.
-
The best countermeasure is one product across several buyers. Forterra’s autonomy kit can go on military trucks or commercial yard trucks, and Anduril reused the same sensor fusion software across towers and counter drone systems. That shared core is what keeps custom work from becoming the whole company.
Going forward, the winners in defense tech will be the companies that use early contracts as proof points, not as their operating model. The market is moving toward fixed price, repeatable systems with recurring software and upgrade revenue, and away from bespoke project work that grows only by hiring more people.