Is it racism to be concerned that our military is using computer parts that can't (or won't) be produced at home?
A major factor in designing any system is taking careful analysis of the risks and considering the cost of avoidance/mitigation with the cost of risk realization.
Analyze the cost of the risk occuring, the likelihood of that occurance, and the overall expected cost of the risk.
Now consider how expensive it would be to mitigate that risk. If the cost of mitigation is greater than the expected cost of the risk, you are better off putting your money elsewhere. (This assumes that you have fully analyzed the risk and considered ALL costs in your evaluation.)
When it comes to computer parts, sometimes you go down the route of 'trusted foundries'. However, that is an EXPENSIVE route. You will often find that you can plan around the risk, or you might be faced with the fact that your system is just too dangerous to operate given your original CONOPS.
Rough Example:
You need a data store, but you can't be completely sure that the hardware in that data store doesn't have a backdoor that would allow remote access to the data stored in that system.
Options:
1. Build the hardware using a trusted foundry. This is expensive, slow, and often behind the latest tech.
2. Rebuild the industrial capability in your country to manufacture the hardware. This will take a while. Also, why should you trust it just because it's in your country? The only thing this helps is to ensure that you can build replacement parts. (until the factory is bombed)
BUT WAIT!
Why not reevaluate your design and see if you can mitigate this risk with some design changes?
Encrypt the data before it crosses into the domain of the suspect system rather than encrypting it in place after it enters into the domain of the system. Now you don't have to care about the potential for that aspect of the backdoor, and it cost you a hell of a lot less than sourcing from a trusted foundry.
Obviously there is a lot more that most backdoors will allow other than just pulling data. The point is that once you start getting to the point that you are strongly considering using a trusted foundry, it is critically important that you evaluate your design because there are often ways to render the threat moot in design, rather than trying to completely trust your supply chain all the way to the end user.
I once had to design a system that needed to support the same radios for 20 years. Did I enter into a contract with motorola to keep a manufacturing line up during those 20 years? Nope. I analyzed/tested to see how they would handle sitting on a shelf. Pre-purchased enough radios to handle the expected DoA/Spares/need for 20 years. Since I only needed 100 of them, that was an option. If it were 1,000,000? Well, then Motorola might have considered keeping that line open on their own dime.
The point is, understand your design, and try not to paint yourself into a corner where you NEED to care about such a situation.