When transactions are created they are essentially a payment promise. They are submitted as a message to the network, where every node checks that the promised money exists and then forwards the transaction. Some of these nodes collect transactions in order to create a new block. These miners are essentially participating in a giant lottery where the winner gets to author a new block.
To that end, miners chose which transactions they are trying to confirm and add their own Coinbase transaction. The Coinbase transactions is the first in a block and gets to collect the transaction fees, creates the block subsidy, and contains the segwit commitment (in a segwit block). All the other transactions can be chosen at the leisure of the miner, although they pick the transactions with the highest fee rate, i.e. the transactions that generate the most revenue to them, by default.
The mining process is self-adjusting, in that the difficulty of blocks gets reset every 2016 blocks. Hereby, the difficulty can increase or decreaseee by up to a factor 4. Hence, the difficulty does not always have to get larger, in fact we've seen it decrease a few times this year. There is an argument to be made for it becoming slower, in that lately the demand for transaction confirmation has been much larger than the available blocksize which caused a lot of transactions to remain unconfirmed for weeks at times.
On the other hand, any new better currency would offer exactly the same service, except faster and cheaper.
The thing is: A new network actually cannot offer the same service. There is a huge demand for Bitcoin transactions. However, a new network can only transmit its own tokens. If you want USD, would you settle for Venezuelan Bolivars instead? Although there have been numerous clones, copies and forks, in the past eight years, just cheaper transactions don't seem to be a sufficient incentive for adoption.