Apple’s AI servers idle as Apple Intelligence usage lags
▼ Summary
– Apple is reportedly in advanced talks with Google to host its new Siri AI models on Google’s servers instead of its own Private Cloud Compute infrastructure.
– Apple’s Private Cloud Compute is described as underpowered, underutilized, and fragmented, with its chips unable to run the latest AI models like Gemini.
– The company’s internal cloud infrastructure is inefficient, with isolated systems leading to idle capacity and duplicated costs that the finance team has criticized.
– Updating the Private Cloud Compute software is difficult, and the system is seen as inadequate for the expected high demand of the upcoming Siri chatbot features.
– This potential shift to Google, which already handles some iCloud services, represents a change in Apple’s approach, though longer-term in-house investment is still possible.
Recent reports suggest Apple’s ambitious Private Cloud Compute infrastructure is facing significant challenges, including low utilization and performance limitations. This has reportedly led the company to explore hosting its next-generation Siri AI services on Google’s cloud servers instead. While Apple designed its private cloud to uphold its strict privacy standards, the current system is not meeting the demands of advanced artificial intelligence workloads.
According to the latest information, Apple’s cloud technology landscape is surprisingly disjointed. Different internal teams operate their own independent systems rather than sharing a unified, centralized resource pool. This fragmented approach creates operational inefficiencies, where some servers sit completely idle while other projects lack the computing power they need. The company’s finance department has expressed concern over the costs of maintaining this duplicate infrastructure, yet there is reportedly reluctance to commit the billions required for a full-scale overhaul. Efforts to consolidate these systems have been attempted multiple times over the past ten years, but these initiatives have repeatedly stalled without resolution.
The situation is particularly acute for the Private Cloud Compute platform built to support Apple Intelligence. The system is described as both underpowered and underutilized, operating at roughly 10 percent of its total capacity on average. This low demand has resulted in newly manufactured Apple servers remaining unused in storage facilities. Beyond the idle hardware, the platform faces technical hurdles. Updating its software is said to be a slow and cumbersome process. More critically, the custom chips powering the servers, believed to be based on M2 Ultra processors, are reportedly not powerful enough to run the latest large language models, such as Google’s Gemini, which will form the foundation of the revamped Siri.
Compounding these issues is the slower-than-anticipated adoption of the initial Apple Intelligence features. With user engagement below expectations, the significant investment in the Private Cloud Compute network is being viewed internally as a questionable expenditure. Although Apple anticipates a major surge in demand when its new Siri chatbot capabilities finally launch, the existing private infrastructure is not seen as capable of supporting that scale.
Consequently, Apple is now in advanced negotiations with Google to host the new Siri within Google’s data centers. This partnership would leverage Google’s extensive experience in deploying massive server farms for its own Gemini AI models. The two companies already have an established relationship, as Apple utilizes Google’s cloud platform for certain iCloud services, including data storage. This potential shift highlights how the rapid evolution of AI technology is forcing Apple to reconsider its long-term infrastructure strategy. While a greater in-house investment may be part of the future roadmap, implementing such a fundamental change is a complex and lengthy process.
(Source: 9to5Mac)
