Apple is reportedly in discussions with PrismML, a startup that has developed technology to significantly compress AI models for local execution on iPhones. PrismML claims its technology can reduce memory usage by up to 15 times, allowing AI computations to be performed directly on the device. This advancement could enhance the speed of Siri and improve user privacy by minimizing cloud data requests. If successful, this technology may influence market demand for memory and data center resources, although the AI sector is expected to continue its high consumption of chip resources.