Apple Outsources Siri Cloud Workloads to Nvidia Blackwell GPUs
Apple is preparing for a landmark September 2026 launch of its revamped Siri, which will rely heavily on external cloud infrastructure. In a significant departure from its historical vertical integration model, Apple is outsourcing complex, cloud-bound Siri queries to Google Cloud, where they will run on Nvidia's flagship Blackwell B200 GPUs.
Overhauling Siri via Google Cloud & Nvidia Blackwell
Definitive reporting from The Information on June 4, 2026, confirmed that Apple is on pace to launch its rebuilt Siri in September 2026.
- The Hybrid Architecture: While simple, on-device Siri requests will continue to be handled by Apple's in-house Apple Silicon, complex reasoning and retrieval tasks will be redirected to Google Cloud's AI infrastructure.
- The Silicon Core: These complex cloud queries will run on Google's massive fleet of NVIDIA Blackwell B200 GPUs. The system will leverage a licensed version of Google's Gemini models to power the advanced conversational and reasoning capabilities of the new Siri.
- Confidential Computing Integration: To maintain Apple's strict user-privacy guarantees while executing workloads on a third-party cloud, the deployment will utilize NVIDIA Confidential Computing. This technology provides hardware-level encryption and secure attestation, ensuring that sensitive user data is fully isolated and protected from the host cloud environment.
Strategic Implications
This alliance is a massive win for both Google and Nvidia:
- For Nvidia, securing Apple's Siri workloads—even indirectly through Google Cloud—firmly embeds its Blackwell architecture into the iOS ecosystem. It also highlights the critical role of Nvidia's Confidential Computing in unlocking enterprise and consumer workloads that require absolute data privacy.
- For Google, it positions Google Cloud as the premier infrastructure partner for high-performance AI, demonstrating that its massive capital expenditure on Nvidia chips is successfully attracting top-tier enterprise workloads.
- For Apple, it demonstrates pragmatism in the "agentic" AI race. By utilizing Google's pre-built infrastructure and Nvidia's state-of-the-art silicon, Apple avoids the massive, multi-year delay of constructing its own hyperscale data centers1, allowing it to compete immediately in the consumer AI agent space.2
-
An instance of Multi-billion-dollar AI hardware builds have outgrown dilutive equity. — It shows how physical compute limitations force tech giants to rely on outsourced, pre-built hyperscaler infrastructure. ↩︎
-
An instance of Rapid hardware depreciation turns proprietary data centers into a toxic liability for consumer platforms. — Apple circumvents the capital drag and construction limits of buildouts by outsourcing its complex cloud-based Siri queries to Google Cloud's AI infrastructure. ↩︎