Apple plans to get into generative AI by offloading complex queries to the M2 Ultra chips that run in data centers before moving on to the more advanced M4 chips.
Bloomberg reports Apple plans to deploy the M2 Ultra to cloud servers to handle more complex AI queries, while simple tasks are processed on-device. The Wall Street Journal it was previously reported that Apple wants to develop custom chips to bring into data centers to provide security and privacy in what the publication calls Project ACDC, or Apple Chips in the Data Center. But the company now believes its existing processors have sufficient security and privacy components.
The chips will be placed in Apple’s data centers and eventually on servers managed by third parties. Apple operates and operates its own servers in the United States New center in Waukee, Iowafirst announced in 2017.
Although Apple isn’t moving as fast in generative artificial intelligence as its rivals, such as Google, Meta and Microsoft, the company is researching the technology. In December, Apple’s machine learning research team released MLX, a machine learning framework that can efficiently run artificial intelligence models on Apple silicon. The company is also releasing other research around AI models that hint at how artificial intelligence might appear in its devices and how existing products like Siri might improve.
When Apple announced its new M4 chip, it put a lot of emphasis on AI performance, saying its new neural engine is an “extremely powerful chip for AI.”