“Apple has a new CEO; he’s a hardware guy.”
That quick distillation of Apple’s impending leadership change spread fast across Silicon Valley and the broader tech world. The company’s choice, John Ternus, rose through the ranks on the hardware side, taking over iPhone engineering in 2020 and all hardware engineering a year later.
Analysts say Ternus’s elevation to succeed Tim Cook signals that Apple will enter the AI era with a family posture: using AI strategically to make its devices work better, but not stretching to incorporate AI into all of its services and businesses.
While its peers are pouring tens of billions of dollars per year into AI research and data centers, Apple’s spending on those areas has remained relatively flat. Its AI research group has not become the company’s center of gravity.
But appointing a “hardware guy” as CEO doesn’t mean Apple’s AI efforts will be suppressed or confined to inconsequential features like erasing an unwanted object from a photo. That’s because Apple’s big opportunity lies in running powerful AI models on its own hardware, not in the data centers of some unaccountable corporation.
The company’s personal AI models would live in a secure enclave within an Apple chip, much like Apple Pay, which keeps financial information invisible to Apple or anyone else. Running on-device, these models could process personal and sensitive data with speed and efficiency without sending that data to the cloud while also maintaining privacy.
Do these things really matter? You bet they do. As distrust of big AI labs grows and regulators lag, guarantees of security and privacy will become potent selling points. Apple has spent years building credibility on data privacy. AI is its chance to cash in.
Right now, running giant AI models on laptops and phones is still a work in progress. But under Ternus, Apple may have the right leadership mix to get there. He helped drive the transition to Apple Silicon, which is foundational to the company’s AI strategy. Johny Srouji, who built and ran Apple’s silicon engineering effort, is moving into Ternus’s former role leading hardware.
Ternus also has a long and productive working relationship with Apple software chief Craig Federighi, who is taking control of most of Apple’s AI research group and will play a key role in integrating AI models into Apple’s operating systems and apps. To make large models run on small chips, Apple’s hardware, silicon, and software teams will need to work in tight coordination.
Apple’s record on AI includes plenty of misses. Siri remains a broken promise. In 2024, Apple said it would transform the command-based assistant into a systemwide AI agent powered by large language models and deliver highly personalized features to iPhones. It has yet to follow through.
But as I wrote at the beginning of this year, the company still has a chance to lead from behind. Apple is unlikely to catch up with OpenAI, Anthropic, and others in building massive general-purpose models. What it can do is use those models to power Siri, as it has said it will, while focusing its own research on smaller models tuned for the unique information tasks of individual users.
My only concern about Ternus is his reputation as a perfectionist. Hardware design rests on mathematical certainties. AI does not. At its core it is probabilistic, not deterministic. It can’t be perfected; it has to be iterated, often in the wild, to improve through real-world use. That may be a difficult shift for a company built on polished, “it just works” products.
Ternus will feel pressure to play it safe. He will soon be running a $4 trillion company and will be accountable to shareholders. It would be easy to prioritize avoiding mistakes over taking risks. But AI just hits too close to Apple’s core identity to play things safe.
Apple’s superpower is providing an artful hardware-software experience that mediates between a human user and digital technology. Eventually, somebody will wield AI to humanize, personalize, and bring more intelligence to that experience. Why not Apple?