We have developed a fully automated laser processing and data acquisition machine, called Meister Data Generator (MDG), which can be used via the Internet 24 hours a day, 365 days a year. All the data taken by MDG is acquired in a database. MDG can perform grid search as well as autonomous optimization with artificial intelligence such as Bayesian optimization. Huge amount of data can be obtained by the automated processing setup, and we applied the data to build a deep neural network that works as a laser processing simulator. The input to the simulator is laser parameters, and the output is a shape of the material after processing. We have tried to make two types of simulators: shallow drilling of silicone and deep drilling of glass. The simulator created by deep learning was found to be applicable to drilling under conditions that were not included in the training data. We were able to predict drilling conditions that achieved more energy savings than the conditions used in the training data. In other words, we now have a predictive AI simulator. We are working with several companies to expand the scope of MDG technology. This simulator could be applied to microfabrication for semiconductor back-end processing.