Apple, Nvidia and Google Unite to Power Smarter Siri

This September could be the moment Siri finally feels like an intelligent companion rather than a polite assistant. Reporting from The Information suggests Apple has struck an unusual trio partnership with Google and Nvidia to power a next-generation, AI-augmented Siri — leaning on Google’s Gemini model and Nvidia’s new Blackwell chips.

For Apple, this marks a tactical shift. The company typically prefers in-house solutions or multiple suppliers. Now, some Siri queries will be routed off-device to Google Cloud, running on a licensed variant of Gemini. Why hand off to a competitor? Because large language models still need scale to deliver nuanced answers, and Google’s model is among the contenders that can provide that reach.

A privacy-first handshake between rivals

There’s a catch that will matter to users: Google’s data centers would use Nvidia Blackwell B200 processors equipped with confidential compute. That means the chip encrypts data while it’s being processed, keeping the raw content shielded even as it’s analyzed in the cloud. Apple plans to combine this encrypted cloud inference with on-device handling when possible, a hybrid approach aimed at delivering smarter responses without abandoning privacy promises.

Think of it as a staged performance: small, routine tasks stay local for speed and privacy; heavier, context-rich queries get escalated to the cloud where Gemini and Blackwell do the heavy lifting under encryption. Latency, battery life, and model fidelity are all part of the trade-offs Apple faces. The goal is clear — make Siri more capable without handing over user data in the clear.

This collaboration also signals how big tech is reshaping alliances. Competition remains, but for some features the fastest path to progress is cooperation. Expect iterative updates and cautious rollouts rather than a sudden leap.

Look for the first public rollout in September.