The news about Michael Bloomberg’s enormous opening ad buy for his presidential campaign—$30 million for TV ads just this week—raises again the question of what he’s up to. When Bloomberg signaled he was entering the race for the Democratic nomination, commentators overwhelmingly scoffed that he had no chance of winning. That’s almost certainly true, if winning means winning the nomination.

But there’s another possibility. Bloomberg and several other candidates—here I include Bernie Sanders—may be running to accumulate delegates in the hope of influencing the final choice of the Democratic convention, even if they’re not the nominee.

Both in their late seventies, Bloomberg and Sanders are well past the age of any previous major-party nominee and on that basis alone might be counted out. For most of their careers, they weren’t Democrats and consequently don’t have the trust of many people in the party. And just as the convention will likely have an anti-Bloomberg majority, so it will likely have an anti-Sanders majority. But none of this means that one of them can’t be a decisive influence on the party’s nominee, depending on how the nomination battle develops.

Unlike the Republicans, the Democrats don’t have winner-take-all primaries; delegates are allocated proportionally under a complex formula that awards them to candidates who garner at least 15 percent of the primary or caucus vote in a district or statewide. The new rules adopted in August 2018 for the 2020 nomination also deny superdelegates a vote on the first ballot.

Given the continued fragmentation of support among Democrats, these rules make it possible, perhaps even likely, that no candidate will emerge from the primaries with a first-ballot majority. That could trigger a battle for delegates between the last primary and the convention, which could stretch into the convention itself. Candidates controlling large blocs of delegates might then have considerable leverage. If the convention goes to a second ballot, the superdelegates would be back in play.

Earlier in the race, I thought that Sanders and Elizabeth Warren were undermining each other’s chances because they would divide progressive support, possibly holding them both below the 15 percent threshold needed to win delegates in particular districts and states. Although that’s still possible, I’m not convinced that the net effect will be negative. Sanders has more of a working-class base of support, while Warren is drawing more from college-educated professionals. The two together may get more votes, and possibly more delegates, than if only one of them was in the race.

The same analysis may apply to Bloomberg, Joe Biden, and other candidates such as Pete Buttigieg who are competing for centrist support. While they may undercut one another, they may also attract votes from different constituencies and thus draw more support together than any one of them could draw alone.

Thanks to his association with Barack Obama, much of Biden’s support in polls has come from African Americans, who are unlikely to back Buttigieg or Bloomberg. For his part, Buttigieg has attracted some of the white, college-educated voters who aren’t drawn to Biden or Bloomberg; Buttigieg’s rise in Iowa seems to have come in part at Warren’s expense.

Bloomberg’s support, such as it may be, is also likely to be upscale. But in primaries that are open to voters regardless of party registration, Bloomberg may attract independents and moderate Republicans who would otherwise not vote in the Democratic primary at all. Indeed, this could be a key source of support for Bloomberg, much as independents on the left who are not registered Democrats have been for Sanders.

Bloomberg is skipping the early primaries and using his enormous wealth to spend massively in the states holding primaries on Super Tuesday in March and later. Skipping the early states defies precedent and doesn’t seem to make any sense. But Biden has already run into fundraising problems, and along with other candidates he may not have the resources to compete in all or even most of the Super Tuesday states. His candidacy may also fizzle if he does poorly in the early primaries. Bloomberg might then accumulate delegates to shore up the centrist side of the Democratic Party.

Granted, there’s a lot of speculation here. If this year’s Democratic race follows the usual pattern, one candidate will build up momentum from early primary victories and gain a majority of delegates. Or a candidate will win so large a plurality of votes and delegates as to have a presumptive claim to the nomination, perhaps on the second ballot with the help of the superdelegates.

But if at the end of the primaries no candidate is near a majority of delegates, we could be in for a multi-sided struggle for the nomination.

One part of that struggle would likely take place within the progressive and moderate wings, as each consolidates around a single candidate. Let’s say Warren and Sanders together have won about 40 percent of the delegates, and the question is which one could attract the additional delegates to obtain a majority. It would almost certainly have to be Warren because of the stronger opposition to Sanders in the rest of the party. Sanders could presumably exact some promises in return for his support.

The moderate wing would face an analogous question, but the resolution seems to me less clear. This may reflect my own prejudices since I find it hard to imagine Biden, Bloomberg, or Buttigieg as the party’s nominee. If supporters of all three recognize that none of them can achieve a delegate majority, perhaps the moderates will consolidate behind a dark horse—Amy Klobuchar, Deval Patrick, or one of the other centrist also-rans. The entry of Bloomberg and Patrick into the race has not solved the moderates’ problem that no candidate of theirs has inspired much enthusiasm.

Which wing of the party would prevail in a post-primary battle? It’s impossible to say. This could be a free-for-all. The individual delegates chosen in the primaries to represent candidates are not legally bound to vote for them at the convention, much less to follow their directions about supporting someone else.

But if the battle goes beyond the first ballot, the approximately 764 superdelegates at the convention could well decide the outcome. The superdelegates can also vote on rules questions that come before the convention on its first night; how they vote could be a signal of things to come.

Soon after the Democratic National Committee voted to strip superdelegates of their votes on the first ballot but to allow them to vote later, I suggested that those twin decisions were a recipe for a legitimacy crisis for the eventual nominee if the superdelegates proved decisive. But at least the DNC left the superdelegates a backup role. It wouldn’t be the worst thing for the Democrats if the final decision rests with a group that is likely to be interested in one thing above all: who can lead the party to victory over Donald Trump.