Yeah, I downloaded both manuals of course and tried to compare the characteristics, but I didn't really find that big showstopper yet. I.e. the output voltage is identical and the (time) jitter of the DG4000 seems to be a even a little better. The harmonic distortion (sine) is not specified for the DG4000 line below 1MHz, but at 1MHz it's better than the DG2041A. Accuracy over temperature is the same, but only the DG2041A defines an accuracy over time (e.g. one year).
What's indeed a little strange is that the 60MHz DG4062 is limited to 25MHz square waves, while the 40MHz DG2041A reaches 40MHz for the square waves as well (especially taking into account the 500MSa/s vs. 100MSa/s sampling rate). Then again, for ramps, the DG4062 is better (1MHz) than the DG2041A (400kHz)...
Anyway, it's also intereseting that all DG4000 devices (60MHz, 100MHz, 160MHz) seem to share the 500MSa/s sampling rate. So obviously the lower devices are just crippled regarding the selectable waveform frequencies. Well, we kinda know that from the scopes of course.
BTW: on the lowest DG4000 (the DG4062) also the arbitrary frequency is limited to 15MHz. I can't imagine any reason for this apart from crippling the low end device. Anyway, again, the DG2041A is even worse (12MHz).