An algorithm is O(2^n) if it explores all possible subsets or combinations. Recursive Fibonacci without memoization is exponential.
Each increment of n doubles runtime. This explodes fast. n = 30 might take seconds. n = 50 could take years.
Exponential algorithms are impractical for large inputs unless you can prune the search space.