Why people talk about "cost of production"? Blocks are 10 min average in the long run, no matter if 2 laptop CPUs, or 100k exahash does the computation, there is no real "cost element" for the protocol to work.

Mining isn't "production". Just because the network needed a way to gradually release new available units to incentivize SECURITY of itself through the Proof of Work consensus, the new units are not the goal of mining. Also, bitcoin units are not "used up" when transacted, every bitcoin introduced in a coinbase transaction is "immortal". Even "lost privatkey" coins, they will just sit in their address for eternity.

This whole concept of "production cost must equal/below product value" is fishy, failed labor theoryof value thinking, marxist bs, totally debunked and failing basic logic since 100+ years.

It does not matter from a utility point of view if there are max 21mil, or a single bitcoin, since it can be divided up to infinitely to smaller sub units, there is no real "production" - it is more like discovery, since all future bitcoins are already algorithmically guaranteed to exist, they just need to be "unlocked", if you must use a describing word.

It is a necessary human psychological trick in play to give out "new coins", to make mining happening, which gives security for the network, which gives decentralized trust to build up, which gives value to the units for users, which prompts more security through more miners going for the coins...you get the point. BUT the new coins will have the same value as old, cpu mined ones, no matter how much cost got sinked in asic farms!!!

In another, new viewpoint, and clarification:
The supply will not be smaller on the halvening at all (existing coins will still exist after it), the additional new unit generation TIME will increase by 2 (the same 25 new coins will be generated in ~20 min average instead of 10mins)! So given a same demand has to wait 2x the time to get access in the network (as UTXOs), or decide to pay the extra value for "frontrunning the queue". Depending on the time preference of the market actor, it is a self fulfilling prophecy which might trigger a run up in price.

Also, since the difficulty is dynamic and adapts to make coin release constant through time, it means that production can never be unprofitable in the long run, as the unprofitable miners leave the race, more new coins will be distributed for the remaining miners, hence mining profit% tends to be constant in the long run for every miner still in race, hence "production cost" can not have influence on coin value (and never had, as long as minimum security levels are reached).

Movement in hash rate is either a technical (hardware, electricity prices, other costs like cooling in northern ares, etc) sign, or fueled by bitcoin value increase, NOT the other way around!