In recent years, more and more high-tech manufacturing plants have used virtual metrology technology to monitor the production quality of machines and processes. The principle of virtual metrology operation consists of two phases: the off-line modeling stage and the on-line conjecture stage. In the off-line modeling stage, various calculation methods are used (such as neural networks, regression techniques etc.) to build a virtual metrology model. In the on-line conjecture stage, the established virtual metrology model can be used to instantly estimate the manufacturing quality of the workpiece or the health of the machine. Therefore, the virtual metrology can solve the measurement delay problem without increasing the measurement cost, and achieve the full inspection realm that the quality of each production workpiece can be monitored online and immediately. Microsoft Azure Machine Learning Studio (AMLS) is a cloud machine learning service developed by Microsoft. It integrates the tools needed for machine learning on a cloud platform and uses drag and drop to analyze machine learning related data, model building, performance testing and service building which greatly reduce the threshold for learning. For providing factory-wide manufacturing service, this research used AMLS machine learning services to construct a virtual metrology cloud platform, so that all production machines have the virtual metrology capability with a highly integration solution. Finally, the actual production data of the factory was used to conduct the integration test and performance evaluation of the system to verify the availability and industrial utilization of the research.