John Ternus Named CEO

Apple said on April 20 that John Ternus will become chief executive on September 1, 2026, succeeding Tim Cook after a board-approved transition. (apple.com) Cook will become executive chairman, and Ternus will move up from senior vice president of Hardware Engineering after about 25 years at Apple. Apple said the board approved the plan unanimously. (apple.com) Ternus has run hardware engineering since 2021 and has overseen work on iPhone, iPad, Mac, AirPods, and other devices. Apple’s leadership page says he joined the company in 2001 after working on the Power Mac and iBook teams. (apple.com) Apple is making the change as artificial intelligence shifts competition from apps and cloud services back toward chips, memory, and battery life. Reuters reported that analysts read Ternus’s promotion as a bet on Apple’s strength in devices and in weaving artificial intelligence into existing products. (usnews.com) That focus fits Apple’s current playbook. Apple says Apple Intelligence is built into iPhone, iPad, Mac, and Vision Pro, and says many requests are handled on the device instead of being sent to a remote data center. (apple.com) In plain terms, on-device inference means the model answers on your phone or laptop, using the chip already in your hand. Apple said in 2024 that when a task is too large for the device, Private Cloud Compute sends it to Apple-run servers built to process the data without storing it. (apple.com) Apple’s own research shows how central that hardware approach has become. The company said its Apple Intelligence system includes an on-device language model of about 3 billion parameters and a larger server model that runs on Apple silicon in Private Cloud Compute. (machinelearning.apple.com) Apple expanded that work in a 2025 technical report, saying its on-device model was optimized for Apple silicon with methods including KV-cache sharing and 2-bit quantization-aware training, both aimed at cutting memory use and compute cost. (machinelearning.apple.com) That matters for economics as much as for speed. Running more requests on iPhones, iPads, and Macs can reduce how often Apple has to pay for expensive server inference, while still using its own Apple-silicon servers for harder tasks. That is an inference from Apple’s published architecture and model design, not a stated financial target. (apple.com) (machinelearning.apple.com) Apple is also starting from scale. Tim Cook said in January that Apple’s installed base had passed 2.5 billion active devices, giving Ternus a large base of hardware that can receive new artificial intelligence features through software updates. (macrumors.com) The immediate test is whether a hardware engineer can push Apple’s artificial intelligence strategy faster without breaking the company’s habit of tight hardware-software integration. Ternus takes over on September 1 with that model now at the center of Apple’s next chapter. (apple.com) (apnews.com)