Make room for AI for mobile in the future

This transition will not be quick or easy, however. Many phones today have some rudimentary AI hardware, but for compute- and memory-intensive AI powered by machine learning (ML), back-end services will need to run in the cloud. Also, the algorithms that will be incorporated into the devices to run on AI subprocessors still need work, particularly if they are to enable the extension of existing programs and processes, such as ERP systems or personal productivity apps. Many hope to move ML to the device itself, but I expect that it will be impractical to acquire the necessary resources to accomplish that other than for very simple processes. It will be easier for companies that use SaaS applications to deploy smart apps than organizations that exclusively use on-premises apps that need to identify and deploy the proper APIs to the new ML services.