What's stopping QRNG chips from being used for nonce generation in mining?

We should definitely be using Quantum Random Number Generators in mining rigs to get genuine randomness when creating nonces. I see that ID Quantique is selling QRNG chips, though I'm not sure how much they go for. So, why aren't these chips common in mining hardware yet? Is it just the price? Or maybe the tech isn’t fully developed yet? I’ve heard some discussions about quantum-resistant tech, which seems to render QRNG less relevant, but we can’t rely on them since we don’t know how they’ll hold up against future quantum computers. Nonces really should come from real randomness.

Mining hardware has no need of them as they are just making guesses at the required values to decrypt the hash. Now encrypting data into a hash - that's another story and would be a valid use. Thing is, unless there are size and power constraints as in mobile applications, conventional TRNG circuits work fine for most encryption applications as both produce totally random results.

I bet some companies are looking into it, you might see one used in coming next generation ASICs

Yes. guesses, which are created by algorithms and dictated by classical mechanics. Don't get me wrong, it is still technically impossible to predict those guesses, but not literally impossible. What happens when technology advances?

ASICs iterate through nonce range to find a share, there is nothing random about it (other than the actual nonce that produces a share is a random output based on the block header).

No you wont. Mining is simply: generate a header, then run the nonce from 0 to 2^32-1 Changing that nonce order will have no effect at all on luck - zero - none - absolutely no reason to do it. You also cannot selectively choose nonces that will give better results - since you have to hash the nonce to find out - which is what a miner does.

Anything more complicated than simply incrementing the nonce would slow down mining, and decrease efficiency. Each chip is changing the nonce around 500 billion times a second. There is no time to mess around -- especially when, as Kano says, there is no benefit. (Modern ASICs also roll bits in other fields, but same idea)