Siri overhaul plans may rely on Google’s fleet of Nvidia Blackwell chips when Apple launches its rebuilt assistant later this year, according to a new report from The Information. The report says Apple will use Google Cloud infrastructure, powered by Nvidia’s Blackwell B200 chips, to handle some of the more demanding AI requests behind the new Siri.
The arrangement would add another outside layer to Apple’s AI strategy. Apple has spent years promoting on-device intelligence and privacy-first cloud processing, but Siri’s long-delayed redesign appears to need more model capacity than Apple can deliver alone at launch. Google’s Gemini models would reportedly help power the assistant, while Nvidia’s chips would provide the compute inside Google’s data centers.
Apple has not announced the new Siri or confirmed the reported infrastructure plan. The company is expected to introduce iOS 27 and its redesigned assistant at WWDC26, with a September launch window tied to the next iPhone software cycle. If the report is accurate, Apple’s most visible AI product may arrive with a very un-Apple-like foundation: Gemini models running on Nvidia hardware through Google’s cloud.
Siri Overhaul Turns Into a Cloud Infrastructure Story
Siri overhaul reporting has already pointed to a major change in how Apple’s assistant works. The new version is expected to be more conversational, more capable across apps, and better at understanding personal context. It may also support typed and spoken interactions, file attachments, persistent conversations, screen awareness, and deeper system actions.
Those features require more than the older Siri architecture. A modern assistant needs language models, context handling, memory systems, app intents, privacy controls, and enough cloud capacity to answer complex requests quickly. Apple can do some of this on device, especially on newer iPhones, iPads, and Macs with Apple Intelligence support. More demanding requests still need servers.
That is where Google and Nvidia enter the story. Google’s Gemini models would give Apple access to stronger AI capability while its own models mature. Nvidia’s Blackwell B200 chips would give the system the performance needed for large-scale inference in Google Cloud.
The combination shows how expensive and infrastructure-heavy modern AI has become. Siri is no longer only a voice interface living inside iPhone. It is becoming a system that may depend on data centers, model partnerships, specialized AI accelerators, and global cloud capacity.
Why Nvidia Blackwell Matters
Nvidia’s Blackwell chips are built for the AI data center market, where companies need enormous compute for training and running large models. The B200 is part of Nvidia’s latest generation of AI hardware, designed for high-performance workloads that consumer devices cannot handle locally.
For Siri, that could help with more complex requests. A user may ask the assistant to interpret what is on screen, summarize a document, search personal information, compare files, rewrite a message, understand a long conversation, or take several actions across apps. Some of that can happen on device. Some may need larger models and cloud processing.
Using Nvidia hardware through Google Cloud would let Apple scale those requests without relying only on its own server fleet. Apple has its own Private Cloud Compute architecture, built around Apple silicon servers, but the reported Gemini-powered Siri plan suggests Apple may use outside infrastructure for parts of the assistant where capacity or model capability is needed.
That creates an unusual AI stack. Apple designs the user experience and privacy layer. Google supplies Gemini models and cloud access. Nvidia supplies the chips powering some of the work. The user may only see Siri, but the system behind it could involve three of the most powerful companies in AI infrastructure.
Apple’s Privacy Message Gets More Complicated
The report also makes Apple’s privacy story more delicate. Apple has positioned Apple Intelligence around on-device processing and Private Cloud Compute, arguing that personal requests should be handled privately, with stronger safeguards than conventional cloud AI services.
A Siri system using Google Cloud and Nvidia hardware would need to fit inside that promise. Apple may still control the interface, privacy policies, request routing, data handling, and user permissions. It could also use Gemini in a restricted way, with Apple acting as the privacy gatekeeper between user data and outside models.
Still, the optics are difficult. Apple has spent years presenting itself as different from companies that rely heavily on cloud AI and advertising-driven data models. If the most advanced Siri requests use Google infrastructure, Apple will need to explain exactly how personal information is protected.
Apple already allows users to access ChatGPT through Apple Intelligence for certain requests, but those interactions require user permission and are presented as an outside model option. A Gemini-powered Siri would be different if it becomes part of the assistant’s core experience.
That does not automatically mean user privacy is weakened. It means Apple will need a more detailed explanation. Users should know when requests run on device, when they use Apple’s cloud, when a partner model is involved, and what data is shared.
Siri Needs the Help
The reason Apple may accept that complexity is simple: Siri needs to catch up. The assistant has fallen behind modern AI systems in natural language understanding, contextual awareness, reasoning, and task completion. Apple’s delayed personalized Siri features have already damaged confidence in its AI rollout.
A stronger Siri could change the conversation quickly if it works well. The assistant could become useful for finding information across apps, acting on personal context, controlling devices, editing content, summarizing messages, managing schedules, and completing multi-step tasks. That would be a far better use of AI on iPhone than a separate chatbot alone.
Apple’s advantage is still system access. ChatGPT and Gemini can answer questions, but Siri can live inside iOS. It can connect to Messages, Mail, Calendar, Photos, Reminders, Notes, Safari, Shortcuts, Maps, Files, Music, Home, and third-party apps through Apple frameworks. If Apple combines that system access with stronger models, Siri can become more useful even if the model itself comes partly from Google.
This is why the infrastructure report is not only a technical footnote. It shows Apple may be willing to use outside AI capacity to make Siri competitive sooner. The company can keep building its own models while relying on partners for the first major overhaul.
September Becomes the Real Test
Apple is expected to present the new Siri at WWDC26, but the larger test will come in September, when iOS 27 is expected to reach users. Reports have suggested the new Siri may launch as a beta or preview, possibly with a waitlist. That would give Apple more control over access, server demand, and early quality.
A staged launch would make sense if the assistant depends on a more complex cloud system. Apple may need to test latency, reliability, privacy routing, model behavior, and user demand before opening the feature broadly. It may also need to avoid repeating the overpromising that surrounded delayed Apple Intelligence features.
The danger is that users and developers may hear another impressive demo but get limited access at launch. Apple needs the new Siri to feel real, not another promise pushed into the future. If the assistant launches with restrictions, Apple will need to make the rollout easy to understand.
The supported device list will also matter. Apple Intelligence already requires newer hardware. Advanced Siri features may be limited to iPhones, iPads, and Macs with enough memory and processing power, even if some requests are handled in the cloud. Apple will need to explain which features are local, which require cloud processing, and which devices can use them.
A Different Kind of Apple AI Strategy
The reported use of Google and Nvidia does not mean Apple is abandoning its own AI work. It suggests a more pragmatic strategy. Apple can use on-device models where privacy and speed matter, Apple silicon servers for Private Cloud Compute, and partner models where capability or scale is needed.
That mix may be the only realistic path in the near term. Building world-class models, data centers, chips, developer tools, privacy systems, and assistant behavior at once is a massive task. OpenAI, Google, Anthropic, Meta, Microsoft, and Nvidia have been moving quickly while Apple has taken longer to ship visible AI improvements.
Apple’s strength is not usually being first. It is turning technology into a product experience that feels integrated and safe enough for mass adoption. If the new Siri can use outside models without making the user experience feel fragmented, Apple may still control the part people notice most.
The risk is dependence. Google would become part of Siri’s AI foundation, Nvidia would sit behind some of the compute, and Apple would still be responsible for the final user experience. If the assistant fails, users will blame Apple. If it succeeds, Apple will need to decide how much of that success it wants to keep tied to outside infrastructure.
A Big Moment for Siri and Apple Intelligence
Siri overhaul plans running on Nvidia Blackwell chips through Google Cloud would mark one of the most unusual turns in Apple’s AI story. The company built Apple Intelligence around privacy, on-device processing, and its own cloud architecture, yet its most important assistant upgrade may need help from Google and Nvidia to reach the level users now expect.
That is not necessarily a weakness. It may be the practical cost of catching up in AI. Apple can still define the interface, security model, app integration, and user experience. Google and Nvidia may simply supply the model and compute muscle behind the scenes.
The question is whether Apple can make that arrangement feel like Apple. Siri cannot simply become Gemini with an Apple skin. It has to understand the iPhone, respect user privacy, work across apps, and complete tasks in a way that feels native to iOS.
If Apple gets that balance right, the Nvidia-Google foundation may fade into the background. Users will care less about the chip in the data center and more about whether Siri finally does what they ask.
