Apple Replaces AI Chief Amid Siri Development Struggles
John Giannandrea’s departure from Apple marks a pivotal shift in the company’s artificial intelligence strategy. After seven years as senior vice president of machine learning and AI, Giannandrea steps down following persistent delays in Siri upgrades and underwhelming reception to Apple Intelligence features. The executive, who joined from Google in 2018 to oversee Siri and broader AI efforts, will serve as an adviser through spring 2026. His exit comes as Apple integrates third-party models like Google’s Gemini to bolster its voice assistant capabilities.
Amar Subramanya assumes the role of vice president of AI, bringing expertise from 16 years at Google and recent leadership in Microsoft’s AI division. At Google, Subramanya headed engineering for the Gemini Assistant, contributing to advancements in multimodal AI processing. His responsibilities at Apple include development of core AI models, machine learning research, and AI safety protocols. This hire signals Apple’s intent to accelerate on-device inference and privacy-focused AI, areas where it trails competitors like OpenAI and Anthropic.
The leadership change follows a series of setbacks for Apple’s AI initiatives. Apple Intelligence, launched in October 2024, faced criticism for generating inaccurate notification summaries and lacking contextual depth in responses. A planned Siri overhaul, promised to handle complex multi-step tasks, was postponed indefinitely, prompting class-action lawsuits from iPhone 16 purchasers. Internal reports highlighted organizational silos between AI and marketing teams, alongside budget shortfalls that earned Giannandrea’s group the derisive nickname “AI/MLess.”
Giannandrea’s tenure saw incremental progress, including the integration of Core ML for efficient on-device machine learning. However, Siri lagged in natural language understanding compared to rivals, with benchmarks showing it resolving only 68 percent of complex queries versus ChatGPT’s 92 percent. Apple responded by reallocating Siri oversight to Mike Rockwell, head of the Vision Pro team, in March 2025, while stripping Giannandrea of robotics responsibilities. An exodus of AI talent to Meta and Google further strained resources, with over a dozen researchers departing in the past year.
Subramanya’s appointment aligns with Apple’s push to regain ground in generative AI. He inherits a team tasked with embedding AI across iOS 19, expected in fall 2026, featuring enhanced image generation via Stable Diffusion variants and predictive text powered by transformer models. Apple now relies on Gemini for Siri’s web-sourced responses, a concession in its longstanding rivalry with Google. This hybrid approach aims to deliver 40 percent faster query processing while maintaining end-to-end encryption.
Broader industry pressures amplify the urgency. An MIT study estimates 12 percent of the U.S. wage bill faces AI disruption, predominantly in white-collar sectors like software engineering. IBM’s CEO Arvind Krishna recently countered layoff narratives, attributing cuts to over-hiring rather than automation. Yet, Apple’s moves reflect a sector-wide scramble: Anthropic prepares for a 2026 IPO, while CrowdStrike reports surging demand for AI-driven cybersecurity tools.
The transition underscores Apple’s challenge in balancing innovation with its privacy ethos. Subramanya must navigate export controls on advanced chips and a memory shortage inflating HBM prices by 25 percent. With Big Tech raising nearly $100 billion in debt for AI infrastructure, Apple’s $3.25 billion acquisition of photonic interconnect firm Celestial AI by partner Marvell highlights ecosystem dependencies. Success hinges on delivering a Siri that rivals Gemini’s 15-day weather forecasting accuracy without compromising user data.
Giannandrea’s advisory role ensures continuity during the handover. Apple executives, including CEO Tim Cook, emphasize unified AI governance to avoid past missteps. Subramanya’s track record in scaling Gemini suggests potential for breakthroughs in federated learning, where models train across devices without central data aggregation. Investors await iOS 19 betas for signs of progress, as Apple stakes its ecosystem on seamless AI integration.
