Apple’s artificial intelligence strategy is taking shape around a practical but difficult goal: making generative AI work directly on the devices people already carry and wear. The company has been quieter than several Big Tech rivals, but the signals in acquisitions, job postings, chips and research point in the same direction.
The central question is not whether Apple is interested in AI. It is how Apple wants AI to run. Based on the source material, the company appears focused on bringing AI models closer to the iPhone, Apple Watch and MacBook, instead of relying entirely on data centers.
Apple’s AI buildout is happening in several layers
Apple has acquired 21 AI startups since the beginning of 2017, according to research from PitchBook cited in the source. That is presented as more active acquisition behavior than rival Big Tech companies in this area.
The most recent acquisition named in the source was Apple’s purchase in early 2023 of WaveOne, a California-based startup working on AI-powered video compression. That deal fits a broader pattern: Apple is adding pieces that could help it bring AI capabilities into products where speed, efficiency and device performance matter.
Hiring is another part of the picture. A recent research note from Morgan Stanley said almost half of Apple’s AI job postings now include the term “Deep Learning.” The source connects that term to the algorithms behind generative AI, including models that can produce humanlike text, audio and code in seconds.
Apple also hired John Giannandrea, Google’s top AI executive, in 2018. The source does not describe the full internal strategy behind that hire, but it places it alongside Apple’s wider effort to strengthen its AI capacity.
The iPhone is the hard target
The source frames Apple’s main technical challenge as running AI through mobile devices. That means making AI chatbots and apps work on the phone’s own hardware and software, rather than depending on cloud services in data centers.
That shift matters because large language models are demanding systems. The source says the challenge requires reducing the size of the large language models that power AI and pairing them with higher-performance processors. In plain terms, the model has to fit the device better, and the device has to become better at running the model.
Apple is not alone in pursuing this direction. The source says Samsung and Google have moved faster, with both releasing new devices that claim to run generative AI features through the phone. That creates pressure on Apple, which has traditionally been careful about when it introduces major technology shifts to users.
Igor Jablokov, chief executive of AI enterprise group Pryon and founder of Yap, described Apple’s timing this way: “They tend to hang back and wait until there is a confluence of technology, and they can offer one of the finest representations of that technology.”
Chips are becoming part of the AI story
Apple has also been updating hardware in ways that connect directly to AI. The M3 Max processor for the MacBook, revealed in October, was described by Apple as unlocking workflows that were previously not possible on a laptop, including AI developers working with billions of data parameters.
The Apple Watch is part of the same direction. The S9 chip for new versions of the Apple Watch, unveiled in September, lets Siri access and log data without connecting to the Internet. The A17 Pro chip in the iPhone 15, announced at the same time, includes a neural engine that Apple says is twice as fast as previous generations.
Dylan Patel, an analyst at semiconductor consulting firm SemiAnalysis, summed up the hardware trend in the source: “As far as the chips in their devices, they are definitely being more and more geared towards AI going forward from a design and architecture standpoint.”
For Apple, hardware is not a side issue. If generative AI is going to work on the device itself, processors, memory use and software integration all become part of the product experience. The source points to exactly that combination: chips, research, software expectations and acquisitions moving together.
Research points toward offline and visual AI
Apple researchers published a paper in December announcing a breakthrough in running LLMs on-device by using Flash memory. The source says this could allow queries to be processed faster, even offline.
Apple also released an open source LLM in October in partnership with Columbia University. The model, called “Ferret,” is limited to research purposes. Its role is described as acting like a second pair of eyes, identifying what a user is looking at, including specific objects within an image.
Amanda Stent, director of the Davis Institute for AI at Colby College, explained why that kind of work matters: “One of the problems of an LLM is that the only way of experiencing the world is through text.” She added: “That’s what makes Ferret so exciting: you can start to literally connect the language to the real world.”
The source also notes an important limitation. Stent said that at this stage, the cost of running a single “inference” query of this kind would be huge. That detail is important because it shows the gap between promising research and everyday product features.
One possible use described in the source is a virtual assistant that can identify the brand of shirt someone is wearing on a video call and then order it through an app. That example shows how language, images and device-level action could eventually be connected.
Apple’s AI strategy is tied to its ecosystem
Apple has not promoted AI spending in the same public way as Microsoft, Google and Amazon. The source says those rivals have touted multibillion-dollar investments, while Apple has remained typically secretive about its plans.
Chief executive Tim Cook told analysts last summer that Apple “has been doing research across a wide range of AI technologies” and is investing and innovating “responsibly” in the area. Industry insiders cited in the source say Apple is working on its own large language models, the same broad technology behind generative AI products such as OpenAI’s ChatGPT.
Apple’s Worldwide Developers Conference, usually held in June, is widely expected to be where the company reveals iOS 18. Morgan Stanley analysts expect that mobile software to be geared toward enabling generative AI and say it could include Siri being powered by an LLM.
The business stakes are also clear in the source. Microsoft recently overtook Apple as the world’s most valuable listed company, with investor interest tied to Microsoft’s AI moves. Bank of America analysts last week upgraded their rating on Apple stock, citing expectations that demand for new generative AI features could boost the iPhone upgrade cycle this year and in 2025.
Laura Martin, a senior analyst at Needham, said Apple’s AI strategy would be “for the benefit of their Apple ecosystem and to protect their installed base.” She also said: “Apple doesn’t want to be in the business of what Google and Amazon want to do, which is to be the backbone of all American businesses that build apps on large language models.”
That makes Apple’s likely path distinct. The source points to a company aiming less at becoming the default AI infrastructure provider for other businesses and more at making AI a reason to stay inside the Apple ecosystem. If that strategy works, the most important AI feature may not be a standalone chatbot. It may be a smarter device that handles more tasks locally, faster and more privately through Apple’s own hardware and software stack.