Some Apple AI Servers Sit Unused, Report Claims
A new report claims some of Apple's AI servers are sitting unused in warehouses due to lower-than-expected user adoption of its Apple Intelligence platform. The alleged overcapacity highlights the challenge of forecasting demand for AI features and the financial risks of over-investing in infrastructure ahead of validated user engagement.
The server capacity was partly built to support a full overhaul of Siri, but its most advanced contextual and cross-app features have been pushed to later in 2026. This delay, coupled with a phased rollout of other Apple Intelligence features that began with iOS 18.1 in October 2024, means the full demand on server infrastructure has not yet materialized. Apple's server strategy relies on "Private Cloud Compute" nodes running its own M-series silicon, starting with the M2 Ultra, which Foxconn reportedly began assembling in May 2024. This hardware is designed for privacy-sensitive tasks that are too complex for on-device processing, creating a hybrid model that differs from the cloud-first approach of competitors. This infrastructure build-out represents a fraction of the AI capital expenditures seen from rivals. In fiscal year 2025, Apple's AI-related capex was approximately $12.72 billion, a stark contrast to the combined hundreds of billions being spent by Google, Microsoft, and Meta on their own AI data centers. The lower-than-expected usage aligns with multiple user surveys. A December 2024 SellCell survey found 73% of iPhone users felt Apple Intelligence features added "little to no value." A separate CNET survey revealed only 18% of smartphone owners cited AI as a primary motivator for upgrading, ranking well behind battery life (61%) and storage (46%). Specific features have also faced headwinds. The "Notification Summaries" tool was disabled for news apps in early 2025 after it generated and distributed misinformation, including false reports about public figures, prompting complaints from major news organizations. Despite reports of low engagement, Apple CEO Tim Cook stated during a January 2026 earnings call that adoption was "strong" among users with compatible devices. He highlighted "Visual Intelligence" as a particularly popular feature, though the company has not released specific usage metrics to quantify this. This demand forecasting challenge is set against a backdrop of Apple's shifting AI strategy, which now includes a partnership with Google to use its Gemini models for some of the more advanced Siri features expected in 2026. The move signals a pragmatic approach to catching up in the large language model arms race while it develops its own next-generation server chips, codenamed "Baltra," for 2026 and beyond.
