BotBeat
...
← Back

> ▌

AppleApple
PRODUCT LAUNCHApple2026-06-09

Apple Confirms Siri AI Powered by Google's Gemini on Nvidia Servers While Maintaining Privacy Claims

Key Takeaways

  • ▸Apple's Siri AI uses Google's Gemini models and runs on Nvidia hardware in Google data centers, marking a major shift from Apple's historical on-device and on-server processing approach
  • ▸The advanced AFM 3 Cloud Pro model for complex reasoning runs on third-party Google Cloud infrastructure, while simpler models remain on-device or Apple-controlled servers
  • ▸Apple introduced an expanded Private Cloud Compute system for third-party hardware using multiple security technologies to encrypt data and verify trusted execution
Source:
Hacker Newshttps://arstechnica.com/apple/2026/06/apple-says-its-ai-is-still-private-even-when-its-running-on-googles-servers/↗

Summary

Apple officially confirmed at its Worldwide Developers Conference that its new Siri AI upgrade, powered by Google's Gemini language models, runs on Nvidia hardware installed in Google's data centers. The announcement reveals a significant infrastructure shift as Apple acknowledges the computational limits of its own hardware for supporting advanced language models. Rather than undertaking a massive data center buildout, Apple partnered with Google to leverage third-party servers while introducing an updated Private Cloud Compute system designed to preserve privacy guarantees even on non-Apple hardware. The new system uses Nvidia's Confidential Computing, Intel's Trust Domain Extensions, and Google's Titan security chip, alongside encrypted communications and cryptographically verifiable hardware ledgers to maintain Apple's traditional privacy promises.

  • The partnership addresses Apple's recognition that advanced AI models require computational capacity that would necessitate a prohibitively expensive data center expansion

Editorial Opinion

Apple's decision to run Siri AI on Google's servers while claiming to maintain the same privacy standards highlights a critical tension in modern AI infrastructure. Although the technical safeguards—encryption, signed software verification, and hardware attestation—are credible, outsourcing inference computation to Google-controlled hardware fundamentally changes the trust model compared to Apple-controlled infrastructure. Users evaluating these privacy guarantees should recognize that this partnership represents a pragmatic compromise rather than preservation of Apple's traditional privacy advantage, and the long-term security implications of third-party AI compute remain unproven at scale.

Large Language Models (LLMs)Generative AIMultimodal AIPartnershipsPrivacy & Data

More from Apple

AppleApple
UPDATE

Apple Enables Distributed LLM Inference and Training on Local Macs with MLX and JACCL

2026-07-23
AppleApple
UPDATE

Apple Fixes Hide My Email Privacy Vulnerability After Year-Long Delay

2026-07-21
AppleApple
POLICY & REGULATION

EU Exempts Apple Watch and AirPods from Battery Removal Requirements

2026-07-20

Comments

Suggested

CloudflareCloudflare
UPDATE

Cloudflare Expands AI Bot Controls With Nuanced Classification System

2026-07-25
AnthropicAnthropic
PRODUCT LAUNCH

Anthropic Releases Claude Opus 5: Mid-Tier Model Balances Performance and Affordability

2026-07-25
ApertusApertus
PRODUCT LAUNCH

Apertus 1.5 Brings Image Understanding and 4x Context Window to Open-Source LLM

2026-07-25
← Back to news
© 2026 BotBeat
AboutPrivacy PolicyTerms of ServiceContact Us