We do trials to determine if a new treatment is safe and effective. Let's say you got a "miracle drug" that cures whatever disease you're studying, but it is too toxic and kills patients over time. That drug may get hyped up in early development as a miracle cure, but you need to compare it to something else to be sure the toxicity seen is not driven by something unrelated. This is why it's not ethical to run late stage trials without a standard of care or placebo control arm, because in this case the standard of care would be the better treatment option.

This concept is called equipoise, as in the two treatments are equally poised to provide benefit to patients at the beginning of the trial. Otherwise if you had enough data to know for certain your new therapy is better, then the trial is unnecessary and it should just be a regular medicine/submitted to a health authority for approval instead of wasting >$100 million dollars on another trial.