Not 100% backed

Thus, a practical CBDC really will likely be limited to a wide array of non-bank financial institutions, who then handle the consumer-facing details, for a small fee. But having stated it that way, we are essentially rediscovering narrow banks: financial institutions that take deposits and 100% back those deposits with reserves at the central bank, and provide high-speed electronic transactions services.

John Cochrane on digital currency.

He posits that we all have Fed S/L accounts, an exchange only takes milliseconds.  But the traders, in this case boys, will still crowd the gate for an external shock.  Having high speed does not solve the congestion problem when traders have the same speed as the Fed. Bots get congested just like humans.

Digital currencies are great but they will be delegated to fiat banks running digital exchanges with a market risk account, a buffer.  Aggregation by economies of scale still rule as no one want's to get bottled up at a single access point and the banks still perform their role in congestion management.  

This is true of all the so called 100% backed theories.  They can all be hit with congestion on an external shock and they all will thus have discounted reserves due to hyperbolic discounting.  That is, under external shock, traders prefer immediate access at a higher congestion fee as they have immediate losses from the shock, it is a disequilibrium. Traders will prefer to aggregate their deposits to an intermediate bank acting as insurance.