Apple has taken the industry's next process leap. On August 25, 2026, the company introduced the M6, its first system-on-chip built on a 2-nanometer process — TSMC's N2 node — and the first Apple silicon shipping with a Dual 16-core Neural Engine explicitly tuned for on-device generative AI.
What The 2nm Move Buys
The M6 packs a 12-core CPU (two super cores, four performance cores, six efficiency cores), a 12-core GPU with a Neural Accelerator inside every core, and up to 32GB of unified memory with 170GB/s of bandwidth. Apple claims the world's fastest single-threaded performance, roughly 1.2x faster multithreaded performance than the M5, and nearly 30% more peak GPU compute for AI compared with M5 — 8x more than the original M1.
Why The Neural Engine Doubled
The Dual 16-core Neural Engine is the headline for on-device AI. System frameworks can now split model inference across both engines automatically, letting apps run bigger language and vision models locally without waiting on Apple's server-side stack. That matters for Apple Intelligence and the AFM-3 foundation model family, which increasingly relies on 20B-parameter sparsely activated models Apple wants to keep off the network.
TSMC N2 Ships In A Mac First
The Mac mini gets the debut, launching Tuesday, September 22 with pre-orders open now. That is a notable reversal from Apple's usual iPhone-first rollout of new process nodes, and reflects the reality that the 2nm ramp at TSMC is being spread carefully across Mac and iPhone families as yield ramps toward volume production for the A20 generation next year.
The Ecosystem Signal
M6 lands one week after Apple's M5 Ultra debut and confirms that Cupertino is willing to leapfrog its own product cadence when the process economics line up. For competitors targeting on-device AI PCs — from Qualcomm's Snapdragon X Elite family to Intel's Panther Lake — the M6 raises the bar on inference-per-watt and forces a rethink of how much of the AI stack should live locally versus in a hyperscaler cloud.
Reporting based on coverage from Apple Newsroom, MacRumors, TrendForce and DigiTimes.