Imagine how much fun it'd be trying to get that to build. Multiply the terror that inspires by (at least) 100x.
I very much doubt that building Windows is a 10 ACE problem, or a 1 ACE problem, or a 0.1 ACE problem.
Or if you're lazy you just Google up this Wikipedia entry:
https://en.wikipedia.org/wiki/List_of_aircraft_carriers_of_t...
The USA has ten active Nimitz-class aircraft carriers. Wikipedia claims that the most recent, the USS George H.W. Bush, was completed in 2009 at a cost of $6.2 billion. A bargain compared to the projected cost of the USS Gerald R. Ford, now under construction with a budget of $15.5 billion.
These numbers are ridiculously overlarge in this context, of course, which is the point. Access to the Windows build toolchain is not even a $10M project. It's not a technical problem. You start by just offering the money to the company. A few million dollars, plus a history of enforcing your TOP SECRET security clearances, plus credible assurances that you're not going to launch a competing product, is probably enough to convince the company to just give you what you want. Quietly, of course. No need to spook the other customers.
If that doesn't work, there's always espionage.
One problem is that you not only have to succeed in building and running, and compatibility-testing Windows from a snapshot of the source code (which, as a snapshot, might not correspond to any release), you have to keep up with patches and new releases. So if that ever had been done, the utility would be limited. Perhaps some modules would match bit-for-bit. What would that tell you?
That's still very unlike being able to audit source code, audit updates, and build from audited code using an identical toolchain, also built from audited code, as an upstream release.
I also think your "ACE" metric turns out to be larger than you think, once you add in an air wing and escort and support ships. Launching a ship that is an aircraft carrier is a bit like booting a kernel with no userland.