@Luca
Jason Lowery describes the digitization of crypto-mining energy spent via its proof-of-work algorithm into bitcoin is analogous to the mathematical phenomenon known as Gabriel's horn. This is where the volume of the horn trends to a fixed value as the horn gets longer but the surface area of the horn continues to grow without a definitive end value.
He has a nice diagram in his thesis describing how this physical world (real energy spent) transfer into the digital world (as represented by the bitcoins in circulation).
The way I understand how this works is this:
Imagine if you own the only bitcoin miner in the world. It costs you a certain amount of electrical costs to mine bitcoins (ie receive block rewards). You keep all your BTC mined. If say @Dave86ch gets a miner too and starts mining as well. The total energy spent by the both of you is now greater but your proportion of total energy used is reduced by 50%. If say @rkbabang gets a miner, then both of you will have a smaller proportion of energy spent. As more miners join the party, the 3 of you will get a smaller and smaller proportion of energy spent (ie hash rate). This is similar to when companies issue more shares and existing shareholders get their ownership diluted in the physical world.
However, the total new energy (ie $ spent on electricity) spent by the 3 of you and everyone else is divided by the total number of new BTC mined (the marginal BTC minted) in the digital world. And because each BTC in the existing supply is fungible, this marginal production cost represents the "fair value" of BTC at that moment in time. The new successful miners will not likely part with their BTC unless it is well above the cost of their production. This drives the "rational" selling price upward. And lucky for you, the BTC that you originally mined at a much lower cost, gets the benefit of value appreciation from the new BTC mined by anyone else.
So despite the dilution of your electrical energy spent (ie $ of electricity used, hash rate), the value of the BTC that you earned earlier does not get diluted and in fact increases over time. With an eventual asymptote of 21 million caused by the BTC reward halving cycles ~ every 4 years, the increasing difficulty adjustment built into the proof-of-work protocol, it will likely cause more and more electrical energy needed to mine the next set of marginal BTC rewards.
This increasing energy spent has no specific upper limit (which is analogous to the infinite surface area of Gabriel's horn) but will trend towards a finite BTC supply (which is analogous to the finite volume of Gabriel's horn). So if people have now colonized our entire solar system and are mining BTC, this energy expenditure can still be represented by the fixed number of BTC.
The above scenario describes the situation if everyone that mines, holds onto their BTC and does not transact with it. If people start transacting with BTC, there exists a transaction fee which currently is about 2% of the block reward. If we leave aside why someone would want to transact in BTC, is the network capable of handling 8 billion people? 16 billion? 32 billion?. Because the block sizes are small, only so many transactions by individuals can be processed at any point in time. The average is 10 minutes now but it can be longer if the transaction fee that you are offering as an incentive to miners for confirming your BTC transaction is too low as you compete with others to get into a block. So with more and more people transacting at the base layer, the fees will move up as well.
At some point in time, the transaction fee will be greater than the actual satoshis/BTC being transacted. Which means that at some point in time, people will have increasing difficulty transacting at the base layer (due to competition for a block). This deliberate block size limit was placed to allow the smallest memory needed for the nodes in the system to store the blockchain ledger itself (to maximize decentralization) Therefore, although the # of BTC transacted in a block doesn't have an upper limit (Gabriel's horn's surface area), there is a limit on the # of individuals obtaining a block. This is where Layer 2 pooling of small transactions can take place and be confirmed later at the base layer in a single large transaction. (I'm not an expert with respect to the Lightning network, but this is my basic understanding of its purpose)
Hopefully this answers a bit of your question about scaling as humanity grows and its adoption increases.
Summary:
1) No upper limit on price (aka marginal cost of production/fair value)
2) No upper limit on the # of BTC transacted in each block
3) Therefore, scalable despite a fixed limit on the total # of BTC in circulation