The articles relate to tracking behaviour (IE watching TV, playing video games, cooking, showering, sleeping, out & about), not deciphering what your computer is doing.
Maybe they can tell if you are just browsing the internet, or if you are playing video games. But you could just as easily be training AI models, rendering some animations, 3d modelling or mining buttcoin.
The articles allude to "knowing it's a computer being used" because the noise of it's PSU is different than lightbulbs or fridges.

Data centers use UPS battery banks to provide persistent power during a power outage to cover the gap before backup generators start up.
Data centers use power smoothing systems to protect wiring and switch gear, as well as fixing the Power Factor - which will drastically reduce their electricity bill.

The power required to generate 00001111 Vs 11110000 is the same from an outside perspective. 4 bits on, 4 bits off. And computers have a word size of 64 bits. That's a lot of permutations that a miniscule ripple in current might indicate.
And all ICs in a computer will have local decoupling caps (because the same power noise that "can be decoded by a power meter" interferes with the ICs, so it is filtered).
And the the SMPS will have some chonky smoothing caps, and the MOSFETs chopping that at 100kHz or whatever. Lots of smoothing, and lots of more significant noise sources.
Nothing upstream of the PSU is going to be able to read what's going on inside a PC, other than "it's drawing a few watts more".

Now, maybe there are security minded companies that take things to the extreme.
Having a central DC power system that powers 100s of servers has the benefit of mixing all that potential noise together into an even more incomprehensible noise.
So, I can see some companies spending tens of thousands more just in case someone manages to break physics. Because 20k on an install could save millions later.
But if they are paying that much attention to the data security of power, I hope it's an air gapped system.
Or they've been sold some snake oil.

Seriously, noise on the power is not a data security issue.
It is a privacy issue if power companies are tracking it that closely, because they can model your behaviour.
Which is what the articles are talking about

same, I'm preplexed, when the other thread is literally proving what I'm warning about.

Electronic magnetic interference and light bending tracking is something we've known for awhile in physics.