Underfitting
Also known as: under-fit
Underfitting is the opposite failure: the model is too simple to learn the pattern at all. It does badly on the training data and on new data.
If overfitting is memorizing, underfitting is not learning anything worth keeping.
Usually the fix is more capacity (a bigger model) or more training — or a better representation of the problem.
Remember the seesaw: too little capacity → underfit. Too much capacity on too little data → overfit. You want just enough.