In a study using the Horvitz-Thompson estimator to account for unequal selection probabilities $\pi_i$ in a survey of size $n$, if the inclusion probability $\pi_i$ is extremely small for some observations, what is the primary consequence for the variance of the estimator $\hat{Y}_{HT} = \sum_{i=1}^n \frac{y_i}{\pi_i}$?