Apple Confirms iPhone 18 to Feature On-Device AI Inference Engine
Apple has confirmed the A20 chip inside iPhone 18 will include a dedicated neural inference engine capable of running large language models locally without internet connectivity.
Cross-verified and synthesized across 2 independent reporting sources:
Close-up of Apple iPhone showing the screen and hardware design
Apple has officially confirmed that the iPhone 18, expected to launch in September 2026, will feature a dramatically upgraded neural engine within the A20 chip. The new architecture is capable of running LLM-class AI models entirely on the device.
Why This Matters
Running large language models on-device eliminates the need to send data to remote servers, addressing user privacy concerns that have dogged cloud-based AI features. It also enables AI functionality without an internet connection.
What Features It Will Power
Apple has hinted that on-device LLM capabilities will power a significantly upgraded Siri that can understand and respond to multi-turn conversations, summarize emails and documents, and generate text in context without sending data to Apple's servers.
Technical Specifications
The A20 chip is rumoured to feature a 12-core neural engine with 35 TOPS (trillion operations per second) of AI compute — approximately 3x the performance of the A17 Pro used in iPhone 15.
Competition Context
This announcement comes as Google's Pixel and Samsung Galaxy devices already feature similar on-device AI capabilities, but Apple's control of both hardware and software gives it a potential advantage in integration quality.
