How does PCA behave when there are redundant duplicate features?

Duplicate features inflate variance but do not add new information, which PCA compresses into fewer components.

When identical or highly correlated features exist, PCA detects redundancy through covariance structure. These duplicates contribute to the same variance direction, causing PCA to assign higher weight to that direction. However, no additional informational dimension is added, so PCA effectively compresses them into a single principal component.