There's a pretty detailed proposal of a privacy-preserving L2 solution for Zcash, called BOLT (https://z.cash/blog/bolt-private-payment-channels/). The original paper is at https://eprint.iacr.org/2016/701 , and there's a prototype implementation by J. Ayo Akinyele at https://github.com/ZcashFoundation/libbolt . Further development of that implementation is being funded by the Zcash Foundation (https://github.com/ZcashFoundation/GrantProposals-2018Q2/iss...). The Blossom upgrade scheduled for October 2019 is intended to include consensus changes needed for BOLT (https://forum.zcashcommunity.com/t/announcing-zcash-blossom-... and https://github.com/zcash/zcash/issues/3676).

There are also realistic longer-term possibilities for L1 scaling of Zcash, including the possibility of doing a Coda-like succinct blockchain (https://codaprotocol.com/). I've been working on optimizing recursive validation (https://github.com/zcash/zcash/issues/3425) and researching how to find more efficient pairing-friendly curves in order to make this practical. Networking optimizations and other consensus layer upgrades along the lines of SPECTRE/PHANTOM (https://medium.com/@drstone/an-overview-of-spectre-a-blockda... and https://medium.com/@drstone/an-overview-of-phantom-a-blockda...), etc. are also quite feasible.

It's correct that we're not running anywhere near capacity at the moment, and that gives us time to develop these solutions.

-- Daira Hopwood (Zcash developer)