That's not last September, that's when the article was updated.
Cambridge, which is the main source quoted there stopped doing any kind of regional research at the end of 2021.
https://ccaf.io/cbnsi/cbeci/mining_mapThe 21% number is an approximation from December 2021!
To take control of bitcoin, you need to take control of 80% if not all the networks and even at that moment, you'd still need a lot of power and money (for the equipment) to be able to pull it off and China's lacking in money already
You see this is why I love the contradictions people keep falling for in this field
On one side we have the pitchforks lighting up when we talk about bitcoin energy consumption as we compare them with Christmas lights and then we claim it's impossible to make an attack because it requires a lot of power which no country has...so...which one?
First, there is a mistake in the percentages, they don't even need 50% of the current hashrate!
They could simply buy 3 billion worth of equipment (resell price) so way cheaper produced and with that 30% added they would force a ton of other miners in bankruptcy, taking over the hashrate pie, so rather than adding 600exahash to subdue the current one they could simply replace them by bankrupting and making mining less profitable for other miners, while still making some money.
Instead of 600 exa costing 12 billion in gear, it could theoretically be done with half of that, again at resell not producing cost, Beijing municipality has a 110 billion yearly budget!