For the Air Force, the MVP is often something that can do a lot of damage. Some things require expensive upfront design. One cannot just iterate through two week sprints until the crew survives.

however, that MVP - used in a test environment - can lead to much more useful feedback from real people on the ground in days or weeks, vs years to build then deploy then getting ignored because it doesn't do what the users need.

There's a balance to be found, and it sounds like some of these skunkworks approaches are starting to get more formalized.

Oh most definitely. The tender process is awful from start to finish, regardless of market. If nothing else, as you say, it funnels huge amounts of resource away from where it needs to go (the product, developers and engineers) and into the pockets of salesmen, managers and shareholders. It just doesn't make sense for public money.

But MVPs aren't always themselves a viable development target... Especially here in a military scenario where the finished product tends to be the MVP. Safe enough and hopefully effective.