Say, there exists 2 humans and 98 bacteria. Consider all cells of a human one life.

50% of ALL life doesn't care which species the life is, and therefore there's a chance that 50 bacteria die. The probability of that happening is 98C50 / 100C50 = 98! 50! 50! / (100! 50! 48!) = (50)(49) / ((100)(99)) = 0.247

For my previous argument, I did not actually do the math. Now that I have done a little bit, the probability seems to converge at 25%

Obviously, this is based on the interpretation of "all life". For my interpretation, "all life" includes every life in a single set, and apply the 50% snap to that. For some others however, it may be interpreted as each species in their own set, and the 50% snap is applied on all those sets.

Moreover, that's not how probability works in independent events.

It'd be like saying "I flipped coin A 1 billion times and got half a billion heads, so now that I'm flipping coin B 100 times, I probably won't get any more heads."

It should be fairly obvious that you can say the exact same statement about tails, and get a completely contradictory statement.