https://metrumacoustics.com/mqa/ I think this explains it fairly well.
"The second unfold corrects the impulse of the digital to analog converter as well as extracting the fine temporal detail of the original recording. This process is tailored because every single DAC has its own behavior. This is most beneficial for oversampling DACs that use steep filtering. The DAC’s oversampling filters are always applied just “above” the music and have a negative impact on the impulse behavior. To solve this, one can use their own up-sampling technology to push the DAC’s steep filtering up to higher frequencies. As a result the DAC’s filter is being detached from the music and causes less disturbance to the music. Gentle up-sampling can also create artefacts, but these are generally considered less annoying than steep filtering.
"Now let’s talk Non Oversampling. Per definition a Non Oversampling digital to analog converter does NOT use any filtering and is entirely focused at the perception of the human ear. As NON Oversampling DAC do not use any filters, in combination with MQA Core Decoder (first unfold), a NON Oversampling output will only benefit, and this is the case with every single DAC in the Metrum Acoustics range."
I read that as the 2nd/3rd unfold for MQA is more beneficial to OS DAC’s.