Apple reshapes its artificial intelligence leadership as Siri relaunch looms

Apple is quietly reshaping its artificial intelligence leadership and teams while preparing a 2026 relaunch of Apple Intelligence built around on-device foundation models and a more capable Siri.

Apple is methodically restructuring its Apple Intelligence organization ahead of a major 2026 relaunch, and the changes point to a refined strategy rather than a retreat from artificial intelligence. The article argues that John Giannandrea’s retirement should not be read as a collapse of Apple’s artificial intelligence push, since the underlying plan centered on on-device processing and foundation models remains intact. Despite reports of about a dozen artificial intelligence related departures over the past year, Apple has continued to execute on its original roadmap, with leaks about the spring Apple Intelligence launch reinforcing a focus on private, on-device systems supplemented by carefully controlled cloud components.

The leadership reshuffle clarifies responsibilities around Apple’s artificial intelligence stack. Giannandrea, hired from Google in 2018, originally oversaw Siri, artificial intelligence, and machine learning, then later absorbed Apple Car and robotics. In 2025, robotics was moved under the hardware chief and Siri under the Vision Pro team, leaving his group to concentrate on Apple Foundation Models. That structure persisted after his exit, signposting that Apple had planned the reorganization irrespective of his tenure, likely in response to internal frustration over Apple Intelligence delays. Former Google executive Amar Subramanya now serves as vice president of artificial intelligence reporting to Craig Federighi, leading the Apple Foundation Model effort, while parts of the prior organization have been redistributed and Eddy Cue is expected to handle search and knowledge work.

The article contends that headline-grabbing departures obscure the true scale of Apple’s artificial intelligence workforce. Public research papers show Apple permitting artificial intelligence researchers to publish in a way that surfaces at least part of the team. The author notes that, after examining only about 4 of 96 pages of research documents and listing each contributor that works at Apple, they had a list of nearly 200 names, turning a planned roster into an unmanageable project. This exercise is presented as evidence that Apple’s artificial intelligence team is “quite large” and still prolific, even after a year with about a dozen notable exits. The piece cautions against treating each resignation as evidence of strategic failure, arguing that many senior departures around the six year mark and frequent churn among lower level artificial intelligence staff are normal for the industry.

Looking ahead, Giannandrea’s on-device modeling work is expected to surface in consumer products rather than vanish with his retirement. Apple Intelligence is expected to get a serious revamp in early 2026 with iOS 26.4. The central feature will be the new LLM-backed Siri that can address on-device Apple Foundation Models to perform tasks via app intents. In parallel, Apple is expected to reveal partnerships that bring third party models to Private Cloud Compute, such as a custom Gemini model used for web search after receiving data from an on-device Apple model. The article concludes that, as broader culture grows weary of low quality artificial intelligence output and inflated promises from artificial intelligence focused executives, Apple is positioning itself to arrive with a more private, efficient artificial intelligence ecosystem rooted in the iPhone and integrated tightly into its existing software stack.

68

Impact Score

Andrej Karpathy outlines four strategies for Artificial Intelligence startups building on large models

Former Tesla Artificial Intelligence chief Andrej Karpathy argues that a new layer of “LLM apps” is emerging on top of general-purpose language models, with tools like Cursor showing how startups can specialize for specific industries. He outlines four core functions these applications should perform and explains how they can remain competitive with major labs such as OpenAI, Anthropic, and Google.

Cadence tapes out 64 Gbps UCIe chiplet interconnect on TSMC N3P

Cadence has taped out its third-generation Universal Chiplet Interconnect Express solution on TSMC’s N3P node, targeting high-bandwidth, energy-efficient chiplet designs for advanced Artificial Intelligence, high-performance computing, and data center workloads.

Best generative engine optimization tools for Artificial Intelligence search in 2025

As Artificial Intelligence search reshapes how users discover information, generative engine optimization tools are emerging to track citations, sentiment, and visibility across platforms like ChatGPT, Gemini, Perplexity, and Google AI Overviews. This guide reviews 10 leading platforms and explains how brands can integrate generative strategies into existing SEO workflows.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.