The power sometimes is not in proportion to the speed gain as smaller technology versions of the same design have lower current (as internal capacitance is lower). Also newer technology usually HAS to run on a lower voltage - the HP49G+ could run on 3V rather than 4.5V? (I assume the HP49G+ runs from 3 x 1.5V batteries). I wonder what voltage is connected to the ARM processor.
Saturn design may have been more efficient as it wasn't trying to 'emulate' anything. The new design may have lower power 'per instruction' but the emulation may require 2 or 3 instructions each, so no net gain.
The stand by current of 10mA doesn't look too good, a large improvement can be gained by turning off or reducing the clock speed to the processor (not the display) when it's not doing anything. The down side is a tiny start up delay when you start pressing keys, but this should not be noticable.
Do you know the Off / Standby / Running current of other HP calcs using the Saturn processor (e.g. '28, '42, '32, '48)
If the speed is 4x and the running current is 4x then for a given set of tasks the batteries should last the same time. With most electronics the battery life is dominated by the standby current (10mA vs 7mA) so I would guess the batteries will only last about 70% of the time as the older calculator. From memory the AAA batteries used to last about 6 months of light use in the HP49G, so the HP49G+ should need new batteries every 4 months?
I presume there is no connector for an external power supply. HP could have done a 'docking station' including the data connection and power / charger for the internal batteries - oh well may be in the next series of calcs :-)