> This method has been used countless times to waste the current defense budget on unproven tech that often goes nowhere.
It's time to rework how fundamental / "moonshot" military R&D should get done, then. And that's not just true of the US, it's also a hot issue for the German military.
We all need to separate "keeping the stuff we have up and running, and get new versions every so often to incorporate technical developments that happened in the meantime" from "let's try and see if we can hide a B2 bomber not just from radars but also eyes and ears" or "maybe we can get a Star Wars style repulsorlift antigrav system working".
The basic stuff should be done by buying whatever is available on the market. No matter if it's Abrams, Leopard, <insert whatever South Korea makes in tanks here>, and ideally the US, EU, NATO and their major allies (Israel, South Korea, Ukraine) should pool their IP and manufacturing resources for a common pool where everyone can draw from, and everyone can contribute to, alone to keep geographic redundancy.
Unfortunately for moonshot projects though, all the "easy" stuff has been done decades ago, and while the startups may have fancy ideas (especially the offshoots of university research), they lack the funding to get them to real-world stage, or they get bought out by one of the big experienced government milkers where the project fades into oblivion. And the US isn't particularly happy to share their stuff, not even with their close-knit FVEY circle.