New entrant Democratic presidential candidate Michael Bloomberg is going to skip the early caucus and primary states. These are the relatively low population states where free media makes much of the candidates, financing has less impact due to the low prices and limited market for paid advertising, and hence reporters plus a few thousand voters have great influence on who the Democratic Party nominates for president. They are Iowa, New Hampshire, Nevada, and South Carolina. For a self-funded entrant who could pay for the entire primary and general election campaign out of his own extraordinarily deep pocket but who does not have obvious magnetic appeal to voters in these four states, it makes a great deal of sense to focus entirely on the big states that vote soon after the early little ones.
The four earlies have about 4% of the delegates (193 as I decipher the opaque and labyrinthine system) out of the nearly 5,000 delegates to the Democratic nominating convention. By comparison, California (495), New York (319), Texas (262), Florida (248), Pennsylvania (210) account in the aggregate for about one-third of the delegates.
The Democrats do not have winner-take-all primaries. Delegates are awarded proportionally to the votes (whether primary or caucus), but a candidate needs 15% of votes to get any delegates. (Experts in this field should correct me. The system changes between every election and may even be changed now for all I know.) Furthermore, a state is awarded delegates roughly half due to its number of electors (perpetuating the inequitable allocation of electors in the Constitutional) and half due to the number of Democratic votes in the last election (which gives weight to the importance of the party in the state but also counts the surplus votes that make no difference in the election of the president). The net of this system is that it does not reflect how a Democrat will win the electoral college. If the party wanted to prioritize winning the election, all delegates to the convention would come from swing states. The forty states sure to be ignored in the general election would revolt at this prospect. So they demand a role in picking the nominee, after which they play scant meaningful role in the election of the president.
At any rate, the mayor understands well that his bottomless pockets give him a huge advantage in campaigning in the big states where paid media will play an enormous role in affecting voter preference. In his victorious campaign for a third term as mayor of New York in 2009, he spent $174 a vote.
In 2008 Barack Obama won a little less than 18 million votes in the caucuses and primaries (the actual count depends on how you treat the caucus results). He edged out Hillary Clinton by only .1 to .4% of the national popular vote, which of course was irrelevant. (If the national popular vote chose the president in the general election, then at least one party probably would use a national popular vote to choose the president, although it might still stagger the voting through different states over a few months.) If Bloomberg spent, say, $200 a vote to win, say, 20 million votes in the nominating process, the total cost would be $4 billion. That breathtaking sum is only 7% of his estimated fortune, and besides he has promised to give away half his wealth upon death, so why he should not spend the $4 billion now to make a mark on history escapes me. (Some also believe that people in their 70s do realize that they cannot take the money with them to whatever may or may not follow this life.) That amount is, moreover, about the same as the total amount of money both parties will spend on choosing the president between now and November 2020. So it is neither impossible to imagine Bloomberg dipping that deep into his fortune nor inconceivable that the system could absorb that much spending.
But, you might say, he wouldn't be able to get 20 million votes, even with $4 billion in spending. Perhaps. However, he could spend less, get fewer votes, and still either emerge as the nominee or play the queen/kingmaker role in choosing the nominee. If he won a third of the delegates in just the five big states I mentioned above, then he would have a block of 10% of the total number of delegates. Surely he could get some delegates elsewhere as well. The superdelegates account for about 15% of the total at the convention. Bloomberg's 10% plus and the superdelegates together very likely would be able to block the nomination of anyone who the polls showed could probably not win the electoral college vote against the Republican nominee. Then the delegates at the convention would enter into quite a free-for-all in picking the nominee. This is called a brokered convention. Bloomberg's willingness to spend what it takes to get what some bumper stickers call "any functioning adult" into the presidency could lead to this event.
All of this results from the electoral college. No one doubts that the president will lose the national popular vote. He has never obtained 50% or more approval from the American people. To be this unpopular and still have a reasonable path to reelection is unheard of, absurd, and totally a function of an archaic and ridiculous method of picking the national chief executive. If the national popular vote chose the president, then it would follow that the Democratic Party would nominate its candidate in a manner that matched up with winning the national vote -- probably incorporating ranked choice voting in the process so as to produce the nominee most likely to win the most votes in the general election. But instead we have the current wacky system and odd scenarios stem from it.