When you short a stock, you are borrowing shares from someone with the promise to return those shares by a certain date, plus some fee.
Incorrect in one detail -- there is no date certain.
The way this actually works, you borrow from a broker for an indefinite period, and the broker borrows from their other clients. If there's a "net short" position at that broker, the broker will borrow from another broker, and that might have a date certain return time, I don't know about such things.
After borrowing the shares, you promptly sell them. You realize some price, $X. Sometime before the agreed return date, you have to go back into the market and buy shares back. You pay some price, $Y. As long as Y < X, you make money (though you still have to pay brokerage fees and the fee to the shareholder who lent you the stock). So, you don't need to have Y=0 to make money; you just need Y < X by enough to cover your transactions costs.
The critical point is that brokers require that you put up collateral when you borrow the shares. And the collateral has to be equal to the value of the shares (or to some fixed fraction of it or multiple of it, which is totally irrelevant for purposes of this explanation) -- and the required collateral
changes whenever the market price changes.
So a short-seller may have enough collateral to support borrowing 100 shares of Tesla at $30 a share. (Which he promptly sells.) But when Tesla goes up to $60 a share, suddenly the short-seller needs $3000 more in collateral. If the short-seller doesn't have enough collateral, eventually the broker can force the short-seller to sell everything else he owns in order to buy back the shares and return them, and the short-seller ends up bankrupt and in debt to the broker.
(This is why it is extremely unsafe to short-sell; you have unlimited downside risk. Professionals always purchase a call option to limit their downside risk, but they have to purchase a new one every three months, because options *do* have a date certain, so even with that strategy you can't stay "short" forever. So short strategies are rarely longer-term than a few years. I know Microsoft will go bankrupt within 20-30 years, but there is no easy way to bet on that; I don't have enough collateral.)
The problem comes when there are a lot of shorts in the market, and then the outlook for the company changes so that it seems unlikely that prices are heading down. Shorts start looking around for shares to buy to cover their position. But in the act of buying up shares, the shorts drive up the market price, prompting more shorts to try to close out their positions, and so forth. This is a "short squeeze", and it can lead to some very high prices to induce shareholders to part with their shares (especially as, by hypothesis, the company's prospects are looking good). Given that more than half of the TSLA "float" in the market is shorted, we could easily see a short squeeze on TSLA in the next few months.
Notice that the short squeeze is exacerbated by the collateral requirements ("margin calls"). The people who are in short positions may want to stay in short positions, but if they don't have great gobs of collateral, they won't be *able* to. The only entities with unlimited collateral are, effectively, central banks, and they don't do short-selling.
In experimental "mock markets" run by experimental economists, early on, people who knew what they were doing tried to sell short. Even though they were right in the long run, they all went bust because whatever they tried to short, would have bubbles large enough to blow through their collateral. "The market can stay crazy longer than you can stay solvent" is the resulting warning about short selling.
The very high short percentage of Tesla does indicate a coming short squeeze sooner or later, but I wouldn't be able to guess when the shorts will be squeezed; it depends on their level of collateral, partly. Deep-pocketed shorts take a long time before they get squeezed.