I don't buy this argument, for a couple of reasons.
First, for large classes of media, the energy required to consume the media is much larger than the mental cost of even an explicit microtransaction - e.g. the energy required to decide whether or not I want to pay 10c to watch a 10 minute educational video is dwarfed by the energy spent watching the video itself, and probably comparable to the cost of watching 10 seconds of ads beforehand.
Second, for many of the media for which the opposite is true (e.g. short TikTok videos), you can make an argument that the value of those media is very low anyway.
Third, for short media that is still valuable, the cost can be amortized through either buying chunks all at once (e.g. the algorithm pre-computes a feed of 30 minutes of video and tells you the cost, and you can decide whether to watch or not), or through other technical solutions like setting a "payment limit" for the day - there are lots of ways that you can get creative with this.
Fourth, as for "I have to believe it would have been solved" - there are many, many problems that could be solved through technically superior solutions that aren't, because of external forces (monopolies on a market, manufacturing difficulties, lack of a market, economic dis-incentives for either sellers or buyers, band-wagoning/legacy baggage that people are emotionally attached to, etc).
I think that the answer is much simpler - people don't want to pay for content. People want it to be free. And, currently, most people (at least based on my interpersonal experiences) either don't know how much of their data is being harvested in order to pay for their zero-dollar-cost services, don't care, or do care but not enough to push for a micropayment solution.
EDIT: As an entirely separate point from what I wrote above, I think that the mental transaction costs would actually be useful (even if people hate them), because they'd disincentivize tiny chunks of content (due to the overhead of the decision being much higher relative to the length of the content), and incentivize long-form, thoughtful content, which would have a number of beneficial effects.