I disagree. GNSS systems used to be very complex and hard to design.

Times have changed though, and nearly all the components are off-the-shelf now. If you already have the capability to launch a satellite, adding GNSS capabilities is a small extra step of adding an atomic clock and an antenna. All the signal generation can be software defined. All the complexity can live in models on the ground.

Sure, some of the 'high power anti-jam directional antenna' bits require more hardware, but the basics do not.

>small extra step of adding [...] an antenna

That's where the ducktape comes in I guess? (Just kidding)

I mean yeah GPS is ancient tech. But actually getting the required 24 or whatever satellites into orbit (or retrofitting) and getting the receiver tech into enough hands to get any kind of traction is still a sizable challenge.

Couldn't they pay SpaceX or a similar company to do it for them?

What would the difficulty level be of incorporating GNSS into SpaceX’s StarLink satellites?

All the active station keeping required for small satellites in low orbits makes me wonder how much precision would suffer. Perhaps doable but requiring so much constant self-calibration that quality breaks down in areas without a network of active reference base stations? It's difficult to guesstimate the impact with magnitudes so far outside human scale.

Low... I wouldn't be super surprised to discover they already have capable hardware onboard. Although ground receivers would need to use different frequency ranges and therefore it would probably be a non-starter.

Good point actually. That could be a nice value add