Cracking the worlds toughest job: Finance Head of a crypto company
Kudos on the amazing in depth article
A few important corrections and precisions on MPC.
In MPC the private key is not "split" because it is *never* generated or constructed in the first place. Independant shards are generated in different places and never meet and are never in the same place. The wallet example you gave (Cypherock) does not use MPC but Shamir secret which actually does split a private key a represents a weaker model.
I would argue that MPC wallets are more Self Custodial that Custodial because they are on chain and transaction can only be made by the user although there is some elements of trust but so is the case of any software based self custodial wallet where you need to trust the code of the app will work as advertised and that the transaction will hit the node as expected.
Finally there are open source MPC libraries. We (ZenGo) operate one of the most popular one on github and it is used by other parties too. The article of 1kX you refer to has a complete and accurate description
Last reflection : the wallet market should not be organized by hot/cold but according to security primitive on critical failures 1. seed based wallets (critical failure: loss of the seed whether you re a software or hardware) 2. MPC wallets (no seed phrase - off chain risk) 3. smart contract wallets .(contract failure). This would be more representative of options and choices of the key decisions a CFO can and should make