Andrew Gabriel wrote:

[...] and no one produces a 4-quadrant analogue multiplier
chip with differential inputs, which is what you really need for
this task.

Analog Devices still do - AD633 etc.

I suspect most true power meters nowadays work buy
sampling the voltage and current and multiplying digitally.

Which must be a better way of doing it, given enough resolution on the
A/D converters (especially the current, which has quite a wide dynamic
range). The analogue multiplier approach needs great care with zeroing
and offsets if you want to be able to accumulate an accurate kWh count.
You've got to handle current up to 100 A RMS, with some allowance for
crest factor - but most of the time it will be measuring down in the mA
to 5 A sort of range.

--
Andy