That seems to be rather nonsensical - "new posts to Passmark" does not equal "market share". It simply means "new posts to Passmark".
And, of course, since AMD has some new CPUs and Intel not so much, you'd expect more new AMD posts - which has nothing to do with market share.
It is an indicator of the installed base, not sales.
Of course you expect it to be biased towards new builds. I also think that Ryzen might be motivating more people with older AMD processors to check how slow their systems are.
The share for individual models is different from that of userbenchmark. This one has Ryzen as the most popular AMD models, which would seem an indicator of more bias towards new systems and/or advanced users.
If new releases weighed so much the results, you would expect a less gradual curve previously.
Also, this graph only shows Windows.
Even if it were only sales, it still shows Intel at around 70%, which is consistent with what
@koyoot was saying.
In Steam the full AMD share keeps dropping, now at 19%. AMD is still weak in notebooks.