Global threats are accelerating. Defence engineering capacity is not. That is the uncomfortable truth behind every defence-readiness conversation right now. Industry needs more capability, delivered faster, in domains that barely existed five years ago, and the industrial base is stretched to breaking point.
A gap you cannot hire your way out of
I ran engineering businesses for over a decade, and I have watched this play out four times in ten years. You cannot recruit fast enough. You train your best people, someone outbids you, and the knowledge walks out the door. Defence lives this at scale, with constraints most sectors never face: classified environments, security clearances measured in months, technology moving faster than procurement can follow.
The European defence industrial base is short an estimated 150,000 to 200,000 skilled workers. At Barrow, the submarine workforce fell from 14,000 to under 3,000 between programme gaps. That kind of knowledge takes a generation to rebuild. This is not a problem you can hire your way out of. Not at this speed.
Make every engineer you already have dramatically more productive
So the smartest question is not “where do we find more engineers.” It is “how do we make every engineer we have dramatically more productive.” Think of AI as adding capacity to your team, not replacing anyone, but taking on the admin burden that buries your best people so they can focus on what actually matters: developing IP, winning sales, understanding why a distributor keeps discounting, managing supply chains, pricing and compliance.
Sovereignty is not the barrier people assume
The technology is ready. The sector that needs it most is the least equipped to adopt it, because deploying AI demands the same scarce talent defence already cannot find. And no, data sovereignty is not the blocker: JSP 936 and DEFCON 658 are technology-agnostic. They care about how data is protected, not what tools process it, which is precisely what a self-hosted, air-gappable system is built to satisfy. It is the same architecture behind Cortex, designed to run entirely inside your own walls.
What is the single biggest bottleneck stopping your team from delivering more with the people they already have?