Both sets of numbers are meaningful, for understanding different parts of the problem. The spread of an infectious disease like Covid-19 is not a virus moving through a relatively homogeneous population, but rather the overlay of a lot of small outbreaks. Early on, when the total number of cases is small, absolute numbers are most meaningful as a basis for comparison. [1] is a thread from Mar 29 (from the brilliant visualization expert John Burn-Murdoch) arguing that absolute numbers are more predictive of "death toll one week after 10th death" and that normalizing by population only makes smaller countries look worse.
But as the spread continues (a process which takes time), per-capita numbers make sense. Mr. Burn-Murdoch revised his position (see [2] for a podcast with transcript where he talks about his thinking), and now the FT graphs have a normalized option (though their focus has moved on to other metrics such as excess mortality).
Carl Bergstrom also has a great thread [3] where he talks about the choice, concluding that both are valid as long as you're careful about context.
HTH
[1]: https://twitter.com/jburnmurdoch/status/1244380095164420101
[2]: https://voices.media/transcript-john-burn-murdoch-senior-dat...
[3]: https://twitter.com/CT_Bergstrom/status/1249930293928030209