For "regular" dilution, the way it works is that there's a pre-money and post-money valuation. If a company is worth 1M pre-money, and an investor puts in 1M, then the company is worth 2M post-money. Someone who owned 1% before (effectively 10k of the pre-money) now owns 0.5% (effectively 10k of the post-money). Ostensibly, their stake should be worth exactly the same before and after the deal, a smaller slice of a bigger pie. In practice, there's various incentives to inflate valuations, such that the early investor's slice is getting smaller faster than the pie is getting bigger.

Furthermore, investors tend to demand extra terms on top. The big one here is called liquidation preference, which is a clause that says, approximately, if/when this company is sold, this investor gets the first X amount of it, usually corresponding to some multiplier of their investment amount. Later rounds will ask to win out in preference, effectively creating a stack of liquidation preferences. In practice, liquidation preferences can often add up to so much that a moderately-successful sale goes entirely to preferred shareholders, and common shareholders see nothing. Perhaps the dealmakers put in a bit of a sweetener for the founders and/or current executives to grease the deal. Your average employee and angel or seed investors certainly see nothing in this deal.

Because there are preferred shares common shares. Large VCs or loans will demand preferred shares. They get cashed out first in any exit. What's left could be nothing for common share holders.

This has happened plenty of times where a startup in a dire situation had to sell itself or raise a down round with poor terms to survive and the earlier investors got wiped out.

Sold at a fire sale.