A manufacturing firm produces microchips where the defect rate follows a Poisson process with an average of $\lambda$ defects per wafer. If the number of defects $X$ per wafer is observed and we introduce a quality control mechanism that replaces a fraction $p$ of defective chips with non-defective ones, resulting in a new random variable $Y$. How does the variance of $Y$ compare to the variance of $X$?