My guess is that turnout will decline over the next several cycles, for "regression to the mean" reasons if nothing else (2020 had the highest turnout since 1900), so it might take a while. Contrary to pbrower, I wouldn't be surprised if 2024 is a comfortable victory for one side or the other but the winner still receives fewer raw votes than Biden '20.
The Millennial generation is likely to vote in heavier numbers as it matures, especially as its candidates run for High Office. There will be more Congressional representatives, and certainly more Senators (Jon Ossoff is only the first, and there will surely be more) and Governors to well fit Millennial sensibilities.
Here's a catch: Joe Biden beat Donald Trump by 4.26% in the popular election, but in an even shift of 1.74% of the popular vote, Trump would have won Wisconsin and the vote in the Electoral College. A margin of 4.26% should be comfortable, which is possible unless the winner is racking up huge vote totals in not-quite-enough states and failing to seal the deal in the others. Obama beat Romney by 3.86% in 2012, and practically nobody thought that Romney was going to win without doing decidedly better than he did.
So what is "comfortable'? Except for 2012, which is the only Presidential election near the mean result, Presidential elections over the last 120 years have resulted in the winners getting 57.1% (Truman 1948) and 66.5% (Taft, 1908) of the electoral vote. At 61.7% of the electoral vote, Obama in 2012 is in a sort of 'null zone' that electoral results seem to avoid. I'm guessing that elections that are close through late October stay close (Truman, 1948, Kennedy 1960, Carter 1976, Trump 2016, Biden 2020) stay close and that elections that look like blow-outs early (we know those) stay so. A nominee staring at 330 electoral votes for the opponent must do a desperate triage, taking risks of losing some states to win something necessary, as did John McCain in 2008. 330 electoral votes is not close, but it might seem possible to pick off 65 or so electoral votes if everything goes right for one (or wrong for the Other Side).