John Giannandrea leaves Apple, ushering in a new era for AI and Siri

  • John Giannandrea is leaving his senior vice presidency of AI and will become an advisor until his retirement in spring 2026.
  • Apple completely reorganizes its artificial intelligence division and places Craig Federighi at the center of the strategy.
  • Amar Subramanya, with extensive experience at Google and Microsoft, assumes the vice presidency of AI and will lead Apple Foundation Models and ML research.
  • The changeover comes after delays in Apple Intelligence and the new Siri, whose big debut is expected in 2026 with support from Google Gemini technology.

Change in Apple's AI strategy

Apple has confirmed that John giannandrea He is leaving the front line of command. After more than six years as the top executive responsible for its artificial intelligence and machine learning strategy, the executive is immediately stepping down from his position as senior vice president of Machine Learning and AI Strategy, but will remain with the company as an advisor until his retirement, expected in the spring of 2026.

This move is not a simple internal change: It marks the end of an era that began with great ambitions for Siri and Apple Intelligence and which has ended up starring delays, internal frictions and the need to rely on external technology to avoid falling behind in the AI ​​race.

From star signing at Google to an orderly departure

Giannandrea landed in Cupertino in 2018 from Google, where he had leading Search and AI, in what was considered one of Apple's biggest signings in the last decadeTheir mission was clear: to rebuild a fragmented artificial intelligence strategy, raise the bar for Siri, and position the company competitively against Alexa, Google Assistant, and later, the wave of generative models.

Within months he was appointed senior vice president and began reporting directly to Tim Cook. Under his guidance, They unified the Core ML, Siri, and Apple Foundation Models teamsInternal search and research in machine learning, which until then had relied on figures like Eddy Cue and Craig Federighi in a more scattered way.

Among its technical achievements is the massive integration of neural networks into Apple hardware, including the evolution of the Neural Engine present in the company's chips, a key component for running AI tasks on the device itself with a strong emphasis on privacy.

However, his term coincided with the explosion of large language models and generative AI. Apple reacted more cautiously than its rivals, prioritizing privacy and local execution versus the speed in releasing eye-catching features, an approach that not everyone internally shared.

Apple Intelligence and the new Siri: the promise that was delayed

Apple reorganization following Giannandrea's departure

Apple Intelligence was presented at WWDC 2024 as the great renewal of the user experience on iPhones, iPads, and Macs, with a deeply revised Siri and customized features that were supposed to understand the user's context and execute complex actions within apps.

The problems began when it became known that The public demonstration did not reflect the actual state of developmentInternal reports indicated that many of the capabilities showcased were not ready at the time of the presentation, and that several of them had not even been stably implemented in test systems.

The situation became even more tense in March 2025, when Apple acknowledged that The new personalized Siri wouldn't arrive in time and postponed its launch until spring 2026. From within, some officials admitted that the assistant worked well in limited demos, but was unable to sustain that performance on a large scale for hundreds of millions of devices.

These delays, coupled with the perception that Apple was lagging behind OpenAI or Google in generative AI, undermined Tim Cook's confidence in the project led by GiannandreaAccording to various sources, the executive himself acknowledged internally that the deadlines "had gotten ugly".

In parallel, The morale of the AI/ML team plummetedIt began brain drain towards companies like Meta and new startups, and some engineers even referred to the group disparagingly as "AIMLess", playing on the acronym AI/ML to underline the feeling of a lack of clear direction.

Clash of visions and realignment of power in Cupertino

After months of friction, the clash between Giannandrea's philosophy—more cautious and focused on privacy—and that of Craig Federighi—more focused on results visible to the user— it became unsustainable. The turning point came with an internal reorganization in March 2025.

At that moment, Apple decided remove him from the day-to-day development of Siri and to give operational control to Mike Rockwell, who had been in charge of Vision Pro. This move suggested that confidence in Giannandrea's leadership had weakened.

From then on, several of its strategic functions were redistributed. The supervision of the domestic robotics projectsFor example, it ended up in the hands of John Ternus, while other responsibilities gradually shifted to different executives.

The announcement of his formal departure as senior vice president and his move to the role of advisor implies, in practice, the culmination of a process of loss of influence that had been brewing for monthsApple frames the move as an orderly retirement after more than seven years with the company.

In parallel, the board of directors has been increasing Federighi's role in the company's AI roadmap, in a context where His name is increasingly being mentioned as a possible successor to Tim Cook.especially in the medium term.

The arrival of Amar Subramanya: Gemini DNA to power Apple Intelligence

Giannandrea's replacement comes with a high-profile signing: Amar Subramanya joins Apple as its new vice president of AIHe will report directly to Craig Federighi, further solidifying the head of software as the company's key coordinator of artificial intelligence strategy.

Subramanya brings a very particular resume. Before his brief stint at Microsoft as corporate vice president of AI, He spent 16 years at Google leading the engineering of the Gemini assistant. and being a key figure in the development of the models that bear the same name, currently one of the pillars of the company's generative AI.

Apple makes no secret of the fact that it particularly values ​​this experience. In its official statement, it emphasizes that The new vice president of AI combines a solid research background with a strong product focus., and highlights its ability to bring scientific advances to concrete functions integrated into mass consumer services.

Subramanya's mandate will be clear: accelerate the development and implementation of Apple Intelligence, strengthen the base of its own models (the Apple Foundation Models) and ensure that the new Siri arrives on time and in competitive conditions in 2026, even if it means relying on third-party technology.

In practice, Apple has hired the engineer who helped build the “brain” of Gemini At the same time that it's preparing to use a customized version of that Google model to enhance certain Siri features. A move that, far from being hidden, is presented as a pragmatic shortcut to avoid being left out of the current battle.

This is how the new AI map at Apple looks.

Giannandrea's departure implies more than just a name change on the organizational chart. Apple has decided dismantle its large centralized AI division to distribute its assets among several of its most influential executives.

On one hand, Amar Subramanya takes the lead three areas considered criticalApple Foundation Models, machine learning research, and the AI ​​Safety and Evaluation teams will focus on the models and technical infrastructure upon which future intelligent features are built.

The remaining teams that reported to Giannandrea are being reassigned. The units closest to the operation and the supply chain will now report to Sabih Khan, Chief Operating Officer (COO)while the areas related to services and business will be under the supervision of Eddy Cue, Senior Vice President of Services.

In this way, Apple seeks integrate AI much more closely into the product and business linesInstead of keeping it as a large, isolated "tower," artificial intelligence is becoming a cross-cutting fabric that touches everything from iOS and macOS to subscription services, hardware, and the entire value chain.

At the top of this new structure is Craig Federighi, who expands his scope of action beyond traditional software for coordinate the next generation of AI-powered experiencesFor the company, this represents a commitment to a more distributed leadership model but with a clear center of gravity.

A change of leadership with an eye on 2026: the new Siri and Europe's place within it

In the immediate future, all eyes are on the new Siri that Apple plans to launch in spring 2026This is expected to coincide with a major iOS update (internally estimated to be around iOS 26.4). It will be the major public test of Federighi and Subramanya's collaborative work.

The company promises an assistant capable of to better understand the user's personal context, and to access relevant device and app data. and execute complex chains of actions, all while maintaining the emphasis on privacy that Apple has historically championed.

At the same time, there is the expectation that Some of these capabilities will depend on a customized version of Google's Gemini model.This is especially true for the more advanced functions of understanding and generating natural language. This balance between proprietary models and external technology will be particularly delicate in markets like the European Union.

In Europe, the entry into force of the Digital Markets Act (DMA) and emerging regulations on AI are forcing Apple to Provide a more transparent explanation of how the data is processed, what models are used, and where the most sensitive operations are performed.Any deep integration with third-party services, such as Gemini, will need to fit within this regulatory framework.

In addition, the company has recently had to manage Regional delays and limitations in the rollout of Apple IntelligenceWith features that haven't arrived simultaneously in the United States and the EU due to regulatory uncertainties, the new leadership will have to ensure that the next wave of features not only arrives on time but also in a consistent manner across key markets like Spain and the rest of Europe.

Internal attrition, talent drain, and competitive pressure

Giannandrea's departure comes after a period that many within Apple describe as turbulent for AI teamsIn addition to organizational tensions, there has been a steady trickle of departures to other large technology companies and to startups specializing in foundational models.

Much of that talent has set its sights on companies like Meta or new firms linked to former Apple executivesincluding the young company of Jony Ive, the brand's former head of design. This dynamic has fueled the feeling that, for some years, the company lost its appeal to top-tier professionals in artificial intelligence.

Meanwhile, the competitive environment has offered no respite. OpenAI, Google, and other rivals have set the pace for generative AI. with continuous releases, while Apple dealt with announced but unavailable features, and with Siri continuing to be criticized compared to other assistants.

In this context, the change at the top of IA and the choice of Subramanya are interpreted as an attempt to reset the narrative and regain credibility, both in terms of internal talent and in relation to the market and end users.

Tim Cook used the announcement to thank Giannandrea for his role in building Apple's AI technology foundation and, at the same time, to emphasize that Artificial intelligence has been “central” to the company’s strategy for yearsThe message seeks to convey continuity despite the changes, but also makes it clear that a different phase is beginning.

With Giannandrea preparing his final departure for 2026, Federighi gaining influence, and Subramanya in charge of models and research, Apple faces a crucial couple of years to prove it can catch up in AI without abandoning its core identity.The upcoming versions of Apple Intelligence and, above all, the new Siri will be the litmus test to see if this change of course comes in time and if the company's ecosystem is prepared for an era in which artificial intelligence is increasingly more important than the hardware itself.

Related article:
John Giannandrea talks about the importance of machine learning at Apple

Hey siri
It may interest you:
Over 100 fun questions to ask Siri
Follow us on Google News