Apple AI powered by Google Gemini models and Nvidia chips, but maintains distinct architecture

Apple partnering with Google and Nvidia for most advanced AI model

Apple making a 'solid step in the right direction' when it comes to AI, says Seaport's Jay Goldberg Apple on Monday revealed what it's been working on in artificial intelligence at its annual Worldwide Developers Conference in Cupertino, Calif. WWDC showed off demos of its redesigned Siri, which can speak back and forth with the user, a major improvement over previous versions of the assistant. In a demo, Siri was able to check concert dates, set a reminder to buy tickets, and even get directions to pick up a friend on the way to the concert venue. But the announcement also highlighted that Apple has taken a different strategy to many of its Silicon Valley rivals, choosing not to spend billions on infrastructure and the biggest, most advanced models, and instead focusing its message to potential customers on privacy advantages and convenience. Apple executives highlighted the difference in remarks on Monday. "Some appear to be racing forward, seemingly pursuing AI for the sake of AI, without clear regard for the people -- all of us -- that it's ultimately meant to serve," said Apple software SVP Craig Federighi in the launch announcement. But it turns out two of the traditional AI leaders, Google and Nvidia, are helping Apple out with its most advanced model, called Apple Foundation Model Cloud Pro, Apple executives told media in a talk at its headquarters on Monday. While Apple and Google announced their partnership for Apple Intelligence in January, this is the first time that the company has officially confirmed that some of its Apple Intelligence features will run on Nvidia chips.

Siri AI is powered by Gemini models, but is not Gemini - what does that mean?

We know that Siri AI and other Apple Intelligence features are powered by Google's Gemini models, but Apple has been at pains to point out that this is not the same as running Gemini on iPhone. While there are still some unknowns, a far clearer picture is emerging about exactly what all of this means ... Siri AI is not Gemini Assistant Google hasn't exactly helped matters by using the term Gemini to describe quite distinct things. Gemini is the name given to a series of Google AI models, while Gemini Assistant is the name of Android's equivalent to Siri. However, Google has often omitted the "Assistant" part, simply using Gemini to refer to the intelligent assistant. However, this much is clear: while both assistants use Gemini models, they are entirely separate. Siri AI is not simply a rebadged version of Gemini Assistant. Apple's models are powered by Gemini Apple talks about its own models. At the heart of this architecture is our third generation of Apple Foundation Models (AFM), a family of five foundation models However, that sentence continues: ... custom-built in collaboration with Google. But are customized for Apple Macworld's Jason Snell unpacks what Apple has and hasn't said to conclude that four of the five models are custom versions of Gemini running on Apple Silicon, while the fifth - and most sophisticated - is essentially Google's standard model running on Google servers but likely using a different set of training data. Siri AI doesn't pull info from Google's web search or knowledge graph; it uses its own. However, Federighi is not claiming that Apple's models themselves are not based on Gemini code. In fact, he explicitly says the four models made to run on Apple Silicon are "trained using proprietary data with reinforcement learning and refined using outputs from Gemini frontier models." It's likely that the biggest model is trained using both Google and Apple's proprietary data. So what does that mean? It seems like Apple started with Gemini's foundation models, optimized and rebuilt them for Apple Silicon and the model sizes it needs, and retrained them with its own data, weights, and guardrails. What does this mean for Apple user privacy? Two of the four models run on-device. This means that none of the data ever leaves your device, and so absolute privacy is assured. The next two run on Apple Silicon chips on the company's own Private Cloud Compute (PCC) servers. Apple has designed the architecture in such a way that absolutely no data is retained or exposed to either Apple or Google - and this fact is independently verifiable by security researchers. In other words, you don't have to trust what Apple says: any qualified expert can check for themselves. With the most powerful model, this is running on Google servers. However, these are servers dedicated to Apple use, and although NVidia GPUs are used in place of Apple Silicon, the company says that the PCC architecture still applies. An Apple security blog says that exactly the same protections apply. Our core PCC requirements remain exactly the same: stateless computation, enforceable guarantees, no privileged runtime access, non-targetability, and verifiable transparency. PCC on Google servers is not the same as PCC on Apple servers, but the company appears confident that it is just as safe. In my view, this is the one area where we effectively have to take on trust that the implementation of PCC principles works as well as it does on Apple's own servers. By this, I don't mean to imply that Apple would mislead us, but it is uncharted territory, so there may be vulnerabilities the company hasn't yet discovered.

How much Gemini is really inside Siri AI?

Despite using Gemini foundations, Siri AI offers a distinct experience from Google's implementation, with Apple maintaining full control over data security and processing. Apple this week announced a dramatically improved version of Siri, aptly named Siri AI. But instead of accolades, among the Apple enthusiasts in places such as X and Reddit, it's already been decided: Siri AI is just a slightly older version of Google's Gemini, with its own interface and voice. You'd be forgiven for believing this. After months of rumors that Apple was turning to Google's Gemini technology to bring Siri up to speed, and a purposefully vague joint statement to that effect this January, it certainly seemed like the new Siri would be exactly that. But then the big WWDC keynote came and went, and Gemini was barely mentioned at all. Following the keynote, Apple held a private "technical deep dive" for journalists after the event (which was not officially recorded and streamed), where Craig Federighi and three Apple VPs in charge of Siri and AI explained Siri's relationship with Google in greater detail. As it seems is always the case with AI, the truth is complicated, and every company involved is using very precise and opaque language that is more about what they don't say than what they do say. But there's a lot of information out there that can help us get a clearer picture of what Apple's new Siri AI actually is, how it works, and how Google's Gemini is involved. Apple's new Foundation Models Let's start at the bottom. Apple used the term "Foundation Model" a lot during WWDC. In a nutshell, it's a big AI model that is trained on a huge amount of data that is then used in whole or in part to deliver specific AI experiences in apps. They can be language models, vision models, image generation models, or audio processing models, though modern foundation models are what are called multi-modal, which means they understand and produce results among all these things together. Most companies scale their big foundation models to different sizes. The most advanced version of the model is so big it can only fit in and run well on huge AI servers with hundreds of gigabytes of RAM and massive, expensive, high-powered processors. So companies produce smaller versions with fewer "parameters" that can run on smaller servers, desktop computers, and laptops, and even little models that run directly on a smartphone. Apple has five new third-generation Foundation Models, as explained in a post on Apple's Machine Learning research site. The first two are the small models made to run directly on device: * AFM 3 Core: The next generation of our 3-billion-parameter dense model that delivers a step up in quality. * AFM 3 Core Advanced: Apple's most powerful on-device model. It's natively multimodal, enabling helpful features like expressive voices and higher-accuracy dictation. Built on cutting-edge Apple research, this 20-billion-parameter model uses a sparse architecture, activating just 1 to 4 billion parameters at a time depending on the request. This model only runs on the latest Apple devices. Those two are made to run directly on-device for all supported hardware. The AFM 3 Core Advanced model requires an iPhone 17 Pro or iPhone Air, Macs with an M3 and at least 12GB of RAM, or iPads with M4. You'll notice that Apple says it has a "sparse architecture," which means that it is broken up into chunks that specialize in different areas, and only the pieces that are needed are loaded up when you make a request. For example, a piece devoted to math wouldn't be loaded if you ask how tall the Burj Khalifa is, but would be when you follow up to ask how many Burj Khalifas fit between the Earth and the moon. The on-device models are joined by three new cloud-based models: * AFM 3 Cloud: Apple's server-side model, optimized for speed, efficiency, and performance. * ADM 3 Cloud (Image): Devoted to image generation and editing, which unlocks advanced photo-editing tools, the all-new Image Playground, and more. * AFM 3 Cloud Pro: Apple's most capable server-based model, which powers our most demanding use cases, including agentic tool use and complex reasoning. AFM 3 Cloud is the big server model that handles most things, but for the really complicated requests, there's an AFM 3 Cloud Pro. They are joined by a special image-centric model that is used for Image Playground (and all the apps that call on the Image Playground framework), genmoji, and all the new AI image editing tools: Clean Up, Extend, and Reframe. Apple is using its own servers (mostly) The first important point is that the first four models -- the on-device models and the first two cloud models -- run on Apple Silicon. The cloud models use Apple's Private Cloud Compute architecture that makes the code open for researchers to ensure that the only data sent to the cloud is necessary to complete the request. After the query, the data is deleted and never retained. The biggest cloud model, AFM 3 Cloud Pro, requires more muscle than the current Apple Silicon-based servers can provide. It is built to run on Google's cloud infrastructure with Nvidia GPUs, but this is not off-the-shelf server leasing. Apple is running its Private Cloud Compute infrastructure here, too. All the core PCC requirements are met: stateless computation, no privileged runtime access, non-targetability, and verifiable transparency. You can read more about how Apple is extending Private Cloud Compute to Google's servers with Nvidia hardware on Apple's Security Research site. How does Siri AI even work? When you make a request to Siri, it first gets interpreted, either by typing or through a voice recognition model. Then, a component called the System Orchestrator turns what you said into a sort of underlying invisible prompt and decides which model or models it should go to. If you're asking Siri to turn on a light at home, start a timer, or tell you the weather, the on-device model handles that. But if you want to generate a few paragraphs of text, the system orchestrator will send the prompt to the Private Cloud compute cluster for processing. It will also send the appropriate data necessary to fulfill that request. For example, if you're writing an email with a menu of items guests are bringing to a potluck, the system orchestrator might first pull relevant text messages from the search index. Perhaps it could include a screenshot of what's on your iPhone's screen if it includes relevant info. After the text is generated and sent back down to your device, the request and any associated data are deleted. All of this happens with as much encryption and pseudonymity as possible, so nobody at Apple or Google can access your requests, data, or results. This is one reason why some of the new AI image processing tools seemed slow in the iOS 27 demos, because images and data need to be uploaded and processed in the cloud. Turn on Airplane mode and disconnect from Wi-Fi, and you can't use the new AI image tools at all. Where does Gemini come in? In the post-keynote discussion at WWDC, Federighi explained why Siri AI is not Gemini: Of course, we don't have the Gemini app as our app. In fact, none of that client code is part of how we run on iOS. For these models, we use none of the models that Google deploys to their customers, nor do we use the infrastructure and means by which they deploy models to their customers. And then, when it comes to the knowledge base, we of course don't use Google Search or anything like that as the foundation of our system. So I hope that's clear. The amount of the Google Assistant we use is none. Read Craig's words carefully, and you'll notice he's specifically saying that the client experience (the app and assistant) is not Gemini, nor are the specific servers the same ones Google uses to serve Gemini to its customers. Furthermore, Siri AI doesn't pull info from Google's web search or knowledge graph; it uses its own. However, Federighi is not claiming that Apple's models themselves are not based on Gemini code. In fact, he explicitly says the four models made to run on Apple Silicon are "trained using proprietary data with reinforcement learning and refined using outputs from Gemini frontier models." It's likely that the biggest model is trained using both Google and Apple's proprietary data, or has some other distinguishing characteristic other than its size that made him leave it out of that statement. So what does that mean? It seems like Apple started with Gemini's foundation models, optimized and rebuilt them for Apple Silicon and the model sizes it needs, and retrained them with its own data, weights, and guardrails. As a user, you shouldn't expect the same performance, capabilities, and results from Siri AI on your iPhone as you would get from Google's Gemini on a Pixel phone. An analogy I like to use: Apple used Unix (technically, the Unix-derivative called Darwin) as the core for every operating system going back to Mac OS X. But that doesn't mean Apple's OSes share the same compatibility, features, or characteristics as Unix. Nor does it mean Apple lacks the world-class operating system engineers necessary to make a great OS. Unix is merely a foundation to start on, and a quicker way to get a leg up on development. In much the same way as it did back in 1999 and 2000 when building Mac OS X (and then later iPhone OS and so on), Apple used someone else's work to get started and then built its own thing that's indistinguishable from where it began.

Apple's New AI Models Contain 'None' of Google's Gemini Assistant

Apple executives have detailed the architecture of the company's new Apple Foundation Models (AFM) and clarified exactly how Google's technology factored into their development. Craig Federighi, Apple's SVP of Software Engineering, held a post-keynote tech talk (via 9to5Mac) with press on Monday alongside AI VP Amar Subramanya, Siri lead Mike Rockwell, and software VP Sebastien Marineau-Mes to walk through how the third-generation AFM family was built and how it powers Apple Intelligence. "The amount of the Google Assistant we use is none," Federighi said, explaining that Apple uses none of the Gemini models deployed to Google's customers, none of Google's client-side code, and no Google Search infrastructure as the knowledge backbone. Of course, we don't have the Gemini app as our app. In fact, none of that client code is part of how we run on iOS. For these models, we use none of the models that Google deploys to their customers, nor do we use the infrastructure and means by which they deploy models to their customers. And then, when it comes to the knowledge base, we of course don't use Google Search or anything like that as the foundation of our system. Subramanya outlined the new AFM family, which spans two on-device models and three server-side models. The on-device tier consists of AFM Core, a next-generation dense architecture model, and AFM Core Advanced, which uses a sparse architecture and is natively multimodal. Subramanya said AFM Core Advanced is "unlike any on-device model we've run before," enabling new features including invitation and expressive voices without any cloud requests. On the server side, AFM Cloud handles latency-optimized Private Cloud Compute requests, while AFM Cloud Image powers image generation and editing features including spatial reframing. The key detail on the Google collaboration came in Subramanya's description of how these four models were trained. "All of these are custom built for Apple Silicon, trained using proprietary data with reinforcement learning and refined using outputs from Gemini frontier models," he said, making clear that Google's contribution was distillation-based, not a wholesale adoption of Gemini. The fifth and most capable model, AFM Cloud Pro, is designed for agentic tool use and complex reasoning tasks, with quality that Subramanya said is "similar to Gemini frontier models." This model marks a departure from Apple's standard Private Cloud Compute setup. To run it, Apple worked with both Google and Nvidia to extend its private cloud infrastructure to Nvidia GPUs hosted in Google's cloud. Marineau-Mes said Apple wanted to use Nvidia's latest chips but required them to be configured so they couldn't read the contents of Apple's servers. A recent Nvidia technology called "ambiguous confidential compute" provided the solution. We wanted to avail ourselves of the latest technology from Nvidia, and so we set out to extend private cloud compute to third-party cloud. Federighi described the broader system architecture as being organized around a System Orchestrator, a piece of software he called "key to the privacy architecture of our entire system." The orchestrator routes any given query to the appropriate model, on-device or cloud, based on the complexity of the request and the personal context required. It draws on an App Toolbox for in-app actions, a Spotlight Semantic Index for personal content, and on-screen context for real-time awareness. For queries involving current events, responses are found through Apple's own World Knowledge Service, which Federighi said the company has been building for several years. Apple also maintains that all Private Cloud Compute infrastructure, including the extended Nvidia GPU capacity in Google's cloud, can be independently verified by third-party researchers to confirm that user data is never stored or accessed.

Apple's Private AI Will Run on Google's Servers

Apple today said it is expanding Private Cloud Compute (PCC) beyond its data centers, partnering with Google and NVIDIA to run Apple Intelligence workloads on Google Cloud. Private Cloud Compute is Apple's cloud intelligence system for private AI processing, used to keep Apple Intelligence requests secure while handling processing in the cloud. PCC has been limited to Apple silicon servers in Apple data centers, but Apple is now relying on Google servers to handle some Apple Intelligence processing. Apple partnered with Google to use the technologies behind Google's Gemini AI models for its own Apple Foundation Models. While some processing is done on-device, agentic tool use and complex reasoning require cloud processing. Apple says it worked with Google and NVIDIA to extend its PCC infrastructure to Google Cloud systems that run NVIDIA GPUs without compromising privacy and security protections. Our core PCC requirements remain exactly the same: stateless computation, enforceable guarantees, no privileged runtime access, non-targetability, and verifiable transparency. What's new with PCC on Google Cloud is the implementation: NVIDIA Confidential Computing with NVIDIA GPUs, Intel CPUs with TDX, and Google's Titan chip. All server components and software are part of a trusted computing base subject to verifiable transparency and no-privileged-access guarantees, plus Apple has a cryptographically verifiable ledger of all Google Cloud hardware that is part of the PCC fleet to mitigate the risk of supply chain attacks. PCC on Google Cloud also uses many of the same architectural security patterns as PCC on Apple silicon. Apple says the efforts it has made to bring PCC to Google Cloud will mean user data continues to be protected by PCC's security and privacy properties even outside of Apple hardware and data centers. Apple maintains control over PCC software and Apple devices will only trust PCC software cryptographically approved by Apple. PCC on Google Cloud is not fully implemented, and Apple plans to gradually add the full set of protections throughout the beta testing process. PCC on Google Cloud binaries will be available for public inspection. Apple plans to provide public research tooling and access to live PCC nodes in research mode through its Apple Security Bounty Program.

Here's How Much Gemini Is Actually in Apple Intelligence

Apple spent a lot of time talking about the upgraded Apple Intelligence platform and the new Siri AI app at WWDC 2026, and in the days since, a few more details have emerged about how the AI model partnership between Apple and Google will affect the new software -- but answering the question of how much of Siri AI is Apple, and how much is Google, is still complicated. Back in January, we got official news that Apple would be tapping into Google's Gemini AI models to help power Apple Intelligence, that the deal would last multiple years, and that Apple's "industry-leading privacy standards" would be maintained. Neither Apple nor Google explained much at the time about how this partnership would actually play out, but it was clear that this was more significant than Apple's earlier ChatGPT deal, where Siri simply shunted off prompts it couldn't reply to. I expect there was plenty of debate within Apple about whether a technological deal with an arch-rival was worth it, even if it meant catching up more quickly with its AI. Ultimately, CEO Tim Cook and his fellow executives decided that it was -- and after WWDC 2026, we have more information on the details. Siri AI is not the Gemini app... Over to the WWDC 2026 keynote, where Apple's senior vice president (SVP) of software engineering Craig Federighi told us the vastly improved Apple Foundational Models (AFM) had been developed through a "deep collaboration" with Google. Apple had been "leveraging" the technology behind the Gemini models, in Federighi's words, to create the AI that now powers Siri AI and the other new Apple Intelligence features. And you can certainly see the Gemini influence: Apple's AI is now truly multi-modal, capable of processing audio, voice, and text, and much better at producing text and images of its own. Image editing is much improved -- very similar to Nano Banana 2, you might say -- and Apple's AI now has much better world knowledge too, which is another area where the Gemini models excel. However, these are still ultimately Apple's own models. For local models, we have the on-device AFM 3 Core and AFM 3 Core Advanced models, which sit on iPhones, iPads, and Macs -- though the latest hardware is needed for for the Advanced version (the one that allows you to tweak Siri's pace and expressiveness). Per Apple, AFM Core Advanced needs an iPhone Air, iPhone 17 Pro or Pro Max, an M4 iPad or later with at least 12GB of RAM, or an M3 Mac or later with 12GB of RAM. In follow-up comments (via 9to5Mac), Federighi said, "we don't have the Gemini app as our app." That is, Apple Intelligence doesn't use the Gemini AI models, or the client code Google Gemini app users get, or a knowledge base built from Google Search. All that work has been done by Apple. There's no doubt Apple needed the Gemini AI models to get its own models up to par this quickly, but Apple executives are understandably keen to not make too much of the partnership, for the same reasons that they won't talk about the billions of dollars Google pays each year to remain the default search option in Safari. Federighi still had time for some barbed comments though: "Some appear to be racing forward, seemingly pursuing AI for the sake of AI, without clear regard for the people -- all of us -- that it's ultimately meant to serve," he said, in a dig at the competitors that have left his company in the dust on AI (while Apple has paid out $250 million in settlements for promising AI features that never showed up). ...but Apple is using some Google servers That covers the local Apple Foundation Models, which keep all of your queries and data private and protected on whatever device you're using. It's with the cloud-based models that the waters get a bit muddier -- these are the models that are called in to deal with larger, more complex tasks that can't be handled solely on-device. As described by the Apple team (via Ars Technica), the AFM 3 Cloud is for general-purpose use. Then there's ADM 3 Cloud for image generation, and AFM 3 Cloud Pro for "more sophisticated" queries (and, it sounds like, the beginnings of agentic work). Like the on-device models, they use some Gemini smarts at the most basic levels, but with Apple's own contributions and tweaks on top. Those first two models run on Apple servers, but queries sent to AFM 3 Cloud Pro are going somewhere else: They'll be sent to Google data centers, to be processed by Nvidia GPUs. However, according to Apple, the exact same Private Cloud Compute (PCC) protections will be in place for those data centers as for those run by Apple. That means no data is stored (it'll be wiped after the query is processed), no one else can see it (not even Apple or Google), and your identity is masked. Apple lets third-party security auditors check its PCC code, and that's going to be the case with the AFM 3 Cloud Pro models and Google's servers, too. There is one small wrinkle: Apple says "PCC on Google Cloud will be gradually ramping towards the complete set of protections throughout the summer preview period." So if you're running one of the developer betas, some of your most complex AI queries might not yet be as fully protected as you would like them to be. Apple is promising more technical detail on all of this as we get further towards the full launch of the new software updates and Siri AI, but based on the information we have now, it seems to have managed a balance between boosting its AI with Gemini while retaining all of the Apple-ness that its users are going to expect.

Apple Removes The Fog Around Its New Cloud-Based, And 20-Billion-Parameter On-Device AI Models, Brushes Aside Google's Contributions While Hyping NVIDIA's

Apple has established a sprawling and intricate compute architecture, one that ropes in Google and NVIDIA to paper over its embarrassing AI-related shortcomings. Even so, Apple's WWDC 2026 keynote answered as many questions as raised new ones. Thankfully, the Cupertino-based tech giant is now issuing clarifications at the speed of lightning, resolving lingering uncertainties on a war footing of sorts. Apple craftily obfuscates Google's contributions to its new Apple Intelligence architecture, taking pains to point out its own technologies at the core of this new paradigm We already know that Apple Intelligence consists of a combo of on-device and cloud-based models. Even so, this distinction was not very granular. Thankfully, Apple has just provided a critical update, noting that the gigantic cloud-based Apple Foundation Model (AFM) is its own creation, albeit distilled from an equivalent Google Gemini model. Of course, we already know that Apple licensed a 1.2-trillion-parameter Gemini model from Google a few months back. It seems the iPhone maker had only licensed Google's technology for model distillation purposes. Apple also takes pains to note that it conducted its own pre-training and post-training operations on the AFM Cloud. Apple has also detailed the architecture of its Private Cloud Compute (PCC) framework, going on to note: Apple has further clarified that the AFM Cloud itself is divided into 2 categories: a Pro model that runs on NVIDIA GPUs within Google Cloud, and a vanilla model as well as an image generation one that runs on Apple's own servers. As far as on-device Apple Foundation Models are concerned, the AFM Core Advanced has 20 billion parameters, but only needs the quantum of parameters strictly needed to process a given inference request. Critically, this model was entirely designed by Apple, and requires the A19 Pro chip to run on an iPhone. Of course, Apple has also prepared a less powerful on-device model for generalized inferencing on older iPhones. When a user submits a request, for instance, via the Siri AI, a localized orchestrator calls the required tools, collects data, and then generates the prompt for the AFM Cloud. Critically, raw data is not sent to the cloud, just the structured prompt. Of course, this comes as Apple spent the better part of the technical presentation downplaying Google's role within the new Apple Intelligence and Private Cloud Compute framework. Follow Wccftech on Google to get more of our news coverage in your feeds.