Can you use single timestep data for LSTM autoencoder?

Question

I have a dataset with 1000 samples, 1 timestep, and 1 feature (1000,1,1). I see a lot of discussions where they create subsequences (split one timestep column into multiple