Apple’s rumored M7 Ultra is being developed as a high-end Apple Silicon processor focused heavily on artificial intelligence, unified memory and professional computing. The chip has not been officially announced, so final specifications remain unconfirmed.

Current reporting suggests the M7 Ultra could support as much as 1.5TB of unified memory, approximately three times the 512GB maximum available with Apple’s M3 Ultra. The processor is reportedly being considered for a future Mac Studio or specialized Apple server platform, with a possible launch around 2028.

The enormous memory capacity could make the M7 Ultra especially valuable for running large language models, scientific simulations, professional video workloads and complex engineering applications locally. Apple’s unified-memory architecture allows the CPU, GPU and Neural Engine to access the same memory pool without repeatedly copying data between separate system and graphics memory.

That design can help large AI models fit inside one machine, although memory bandwidth, software optimization and actual compute throughput will remain just as important as raw capacity. Reports have compared Apple’s intended AI-performance target with Nvidia Blackwell-class systems, but no independent benchmark results, CPU-core count, GPU-core count or confirmed memory-bandwidth figures are available yet.

The M7 Ultra is expected to sit at the top of an accelerated M7 product family. Current roadmap reporting places the standard M7 in the first half of 2027, followed by M7 Pro and M7 Max chips later that year and the M7 Ultra in 2028.

Apple may also skip M6 Pro, M6 Max and M6 Ultra variants as it shifts more development resources toward AI-focused M7 hardware. For engineers and AI developers, the main appeal would not necessarily be replacing Nvidia’s largest data-center clusters, but creating an exceptionally memory-dense workstation capable of running increasingly large models without relying entirely on cloud infrastructure.