If the point is just saving space, you can already do that by running a node in pruned mode, in which case old block data is deleted after being validated, and just the UTXO set is kept. It still downloads and validates all blocks though.
A next step that is being worked on (in Bitcoin Core) is the “assumeutxo” project, where you can download and bootstrap from a UTXO snapshot, and instantly have a functioning node at that point. For security, assumeutxo does involve a second validation process, which re-validates everything from genesis to the snapshot point.
The concerns about scalability are generally not about disk storage size for the blockchain, or the bandwidth needed to help bootstrap new users. Things like the cost to keep up with the chain (which involves network topology, network latency, bandwidth for transmission, CPU time for validation, database I/O needed for validation, …) within times that are negligible compared to the block time (and the centralization pressure from not being able to do that) are far more important concerns when talking about increasing scale.
