I think the confusion arises in that the marketing literature makes it sound like this partial multibit architecture is something new and unique when, in fact, most DS DACs have been like this for a long time. I can't immediately think of any commonly used DS DAC IC that uses a pure 1-bit (essentially what DSD is) conversion (Don't know about Sabre, they're super cryptic with their spec-sheet NDAs). On the other hand, it's a good thing that the marketing literature is accurately pointing this out, because many people may incorrectly assume that most or all DS DACs are purely 1-bit DACs. It's just not necessarily new or unique. I really liked Thorsten's explanation in his recent Head-Fi AMA where he explains that this hybrid architecture balances both high and low signal level information retrieval where pure multibit and pure 1-bit only excel at one while being weak in the other by design. The explanation does a good job of calming the Nervosa of whether these DS DAC chips are adequate compared to other solutions.
EDIT: These will probably go over most people' heads, but these technical papers really describe the Various DAC architectures in good detail. The third covers the segmented DAC, which is this technical name for this "hybrid" approach.
MT-014: Basic DAC Architectures I: String DACs and Thermometer (Fully Decoded) DACs (analog.com)
MT-015: Basic DAC Architectures II: Binary DACs (analog.com)
MT-016: Basic DAC Architectures III: Segmented DACs (analog.com)
Not to get too off topic from iFi, but since I know a lot of people get confused about Chord DACs, after respectfully prodding Rob Watts enough about a more technical explanation of the "Pulse Array" DAC, he described it as a "thermometer coded DAC with constant switching" which is how I stumbled on these papers trying to get a deeper understanding. I suspect his implementation differs significantly from the basic variations shown here.