One of the reasons old people vote more is because people are strongly habit-forming creatures, and are much more likely to vote if they've done so before.
A very basic model that can demonstrate how useful this effect is: suppose you are 40% likely to vote if you didn't vote in the previous election, and 70% likely to vote if you did vote in the previous election. If you tried forecasting this forward...
In the first election, you are obviously 40% likely to vote.
In the second election, you are 40% likely to have a 70% voting chance and 60% likely to have a 40% voting chance, for a 52% voting chance.
In the third election, you are 52% likely to have a 70% voting chance and 48% likely to have a 40% voting chance, for a 55.6% voting chance.
In the fourth election, you are 55.6% likely to have a 70% voting chance and 44.4% likely to have a 40% voting chance, for a 56.8% voting chance...
and so on. I've made up the numbers, but you can see that as a cohort ages, they become more likely to vote, because they're more likely to have formed the voting habit (eventually, the example above will become asymptotic with a ~57.1% voting chance; it's not a very strong model but tbf I made the numbers up on the spot. Real models also have lag over extended periods and so on).
Getting young people out to vote is an investment. It won't pay off in the upcoming election, because it's hard to get young people to vote because they've not done it before and in the short-term, your political effort would have had more effect focusing on other demographics. It does pay off in about twenty years' time, though, because the average person's political opinions barely change throughout their lifetime - genuine converts are quite rare; and when it does pay off, it has a pretty big effect.