I guess my thought process is that if these mining operations are getting cards "by the pallet load", then selling them for big profits and buying crypto with those profits, yet keeping some cards, then getting the next pallet load and repeating would just make more sense.
So you sell 100 cards at £500 profit each and end up with £50k profit total, buy £50k of Ethereum and then... Nothing
Or
You mine with 100 cards for a year, end up with £100k* worth of Ethereum = £50k profit total, and then.... Carry on mining, or sell the cards for a further £50k** (ending up with £100k profit). These figures are obviously simplified for ease of calculation, but are ballpark accurate.
Or
Some combination of the above, but then every card you sell for £500 profit is a card that isn't mining.
Obviously if you don't have the facilities for mining, can't be bothered with the hassle, want the money now, or think the risk is too high then selling is the better option, but longer term (to a point) mining will "probably" be more profitable. (of course no-one can predict the future, and in theory, eth could drop to $1 overnight, leaving you with £50k of cards that are now useless for mining and only worth £10k because the market is flooded by miners offloading)
* at current value, risk is it could have course go down, but equally it could go up
** again, at current prices, depending on the GPU situation in 12 months this may or may not be accurate