Apple is reportedly preparing a significant upgrade to its Silicon chip portfolio that could redefine the high-performance computing space. According to a Bloomberg report, the upcoming M7 Ultra chip is expected to support a remarkable 1.5 terabytes of unified memory, doubling the capacity of the anticipated M5 Ultra model and challenging competitors like NVIDIA in AI processing power.
In a strategic departure from its usual product cycle, Apple plans to skip the Pro, Max, and Ultra variants of the M6 generation, moving directly into the M7 family. The initial base M7 chip could arrive as soon as the first half of next year, followed by Pro and Max models toward the year's end, culminating in the M7 Ultra release in 2028. This accelerated schedule reflects Apple's growing emphasis on artificial intelligence capabilities within its chip designs.
One driving factor behind this ambitious roadmap is Apple’s pivot away from its discontinued autonomous vehicle project, which pushed the company to develop advanced machine learning hardware. This led to the creation of the Neural Engine, a specialized component within Apple’s chips dedicated to AI tasks on devices. Bloomberg highlights this innovation as a key reason why Macs have become highly capable AI platforms, especially for running large AI models locally.
The M7 Ultra’s enormous memory pool is designed to handle immense AI workloads. Running large language models (LLMs) and other AI applications demands substantial VRAM just to load model weights. The rumored 1.5TB unified memory could allow Macs to operate frontier-level AI models that previously required discrete, specialized hardware. The expansion in memory may also prompt changes in device design to accommodate the new chip’s requirements.
Apple's ambitions extend beyond consumer devices; engineers are reportedly working on an AI server chip based on the M7 Ultra for a 2029 launch, following a planned M5 Ultra server chip codenamed J246. Further ahead, Apple is developing 1.4-nanometer M8 chips under the codename Soko, alongside other high-end Mac processors named Cardinal.
While Apple Silicon has gained popularity among AI researchers for budget-friendly solutions like the Mac Mini—valued for its ease of setup and energy efficiency rather than intrinsic hardware advantages—the new M7 Ultra chip signals a push toward professional and enterprise-grade AI processing power. This positions Apple as a potential rival to established AI accelerator producers like NVIDIA with its Blackwell architecture.