Can we use 1d-CNN for time series once only the first step is available for prediction

Question

I have a dataset of 1000 rows (observations) and 100 columns (features). My features are all time-dependent features at different equal time steps, and my output is the last tim