# Apple AI loses steam as Siri chief departs, talent exodus to Meta, key features delayed to 2026

> Apple's AI plans hit delays as Siri chief Robby Walker exits and upgrades slip to spring 2026 after an architecture shift. Investor doubts grow amid talent losses and reorg churn.

- Type: AI news
- Page: https://completeaitraining.com/news/apple-ai-loses-steam-as-siri-chief-departs-talent-exodus-to/
- Published: 2025-09-14
- Updated: 2025-09-14
- Author: Jeroen Erne
- Jobs: Executives and strategy
- Reading time: 3 min

### Apple's AI push stalls: leadership exits, slipping timelines, and a hard pivot to 2026

Apple's AI ambitions have hit a wall. The former head of Siri, **Robby Walker**, is set to depart, and the company has pushed key Siri upgrades to spring 2026 due to ongoing quality issues and an architecture change.

Investor patience is thinning after a muted "Light AI" showing in September and a rating downgrade from D.A. Davidson. The message for operators: strategy without shipping cadence is a liability.

#### The leadership carousel

Walker, one of the few direct reports to AI chief **John Giannandrea**, led Siri from 2022 but was removed from the project earlier this year amid delays. Oversight shifted back to software lead **Craig Federighi**, with **Mike Rockwell** stepping in to run Siri engineering and reorganize the team.

Attrition is accelerating. **Ruoming Pang**, who led Apple's AI model team, moved to Meta along with several engineers and researchers. **Frank Chu**, a senior search executive, also left for Meta.

#### From Siri to "AKI": Apple's answer engine bet

In 2025, Walker transitioned to lead the "Answers, Information, and Knowledge" (**AKI**) team. The mandate: build a bottom-up answer engine that interprets questions, integrates information, and returns direct answers-an Apple-native alternative to relying on external models.

This was Apple's bid to own the end-to-end answer experience. Internally, Walker faced criticism for a conservative risk posture during his Siri tenure, according to prior reporting.

#### Strategy vs. shipping

Apple explored AI server chips for cloud-edge deployment, rehomed Siri under Federighi, and evaluated using models from Anthropic or OpenAI for the new Siri. The direction was clear; the execution wasn't.

Since unveiling [Apple Intelligence at WWDC 2024](https://www.apple.com/apple-intelligence/?ref=completeaitraining.com), there's been no substantive Siri upgrade. Apple now targets spring 2026 for personalized Siri features, citing persistent quality issues and a move to a more powerful underlying architecture.

#### Why this matters for executives

- Shipping beats signaling: Frequent reorganizations slow decision cycles and blur ownership. Define a single-threaded leader with authority over model, product, and infra.
- Talent risk is strategic risk: Mission clarity, visible progress, and equity upside reduce poaching. Attrition compounds delays.
- Build vs. partner: External models can accelerate feature readiness, but create dependency. Decide what's core (evals, orchestration, safety, retrieval) and what can be partnered.
- Architecture first: Late-stage architecture swaps are costly. Lock core interfaces early, invest in offline evals, and fund platform work before feature sprawl.
- Cloud-edge reality: If privacy and latency matter, plan for hybrid inference. Align chip bets, capacity, and model sizes with clear use-case tiers.
- Expectation management: Don't over-announce. Tie public milestones to GA-ready features and measurable quality thresholds.

#### Action checklist for your AI roadmap

- 90-day stabilization: Freeze reorgs, clarify RACI, and set weekly ship goals tied to user-facing improvements.
- Quality gates: Establish offline evals, on-policy A/Bs, red-teaming, and regression dashboards before scaling.
- Two-speed delivery: Separate platform stability work (observability, guardrails, retrieval) from feature experiments.
- Vendor strategy: Maintain a primary model and two backups. Pre-negotiate data, privacy, and rate limits.
- Talent plan: Retain critical staff with milestone-based grants. Backfill with targeted hires in inference, tooling, and product.
- Roadmap hygiene: Quarterly commit to a small set of high-impact user journeys. Kill or pause anything without adoption signals.

#### Timeline watch

- WWDC 2024: Apple Intelligence revealed; no major Siri upgrades shipped since.
- September event: "Light AI" underwhelmed; street skepticism increased.
- Spring 2026 target: Personalized Siri features after architecture shift; watch for public betas, on-device capabilities, and cloud-edge details.

#### Bottom line

Apple's reset highlights a hard truth: reliable assistants at scale demand clear ownership, the right architecture, and consistent delivery. Treat AI as a product system-not a press strategy-and your odds go up.

#### Further support

If you're formalizing your AI operating model and team skills, explore role-based tracks and certification paths:

- [Courses by job](https://completeaitraining.com/courses-by-job/)
- [Popular certifications](https://completeaitraining.com/popular-certifications/)

---

This is the Markdown form of https://completeaitraining.com/news/apple-ai-loses-steam-as-siri-chief-departs-talent-exodus-to/. The whole site is mapped at https://completeaitraining.com/llms.txt.
