Apple is quietly ramping up its artificial intelligence (AI) capabilities, making a series of acquisitions, hiring employees, and hardware updates to bring AI to its next generation of iPhones.

According to the Financial Times , industry data and academic papers, as well as industry sources, show that the California-based company is focused on solving the technological problem of running AI on mobile devices.

The iPhone maker has been more aggressive in acquiring AI startups than its rivals. According to PitchBook research, it has acquired 21 startups since the beginning of 2017. The most recent was the purchase of WaveOne, a startup that provides AI-based video compression, in early 2023.

Wedbush Securities analyst Dan Ives predicts Apple will make a big AI deal this year because there is an AI race going on and the company cannot stay out.

k2lis0c7.png
Apple is actively acquiring AI startups and solving technological difficulties to bring AI to the iPhone. (Photo: FT)

According to a recent study from Morgan Stanley, nearly half of Apple's AI job postings include the term "deep learning," referring to the algorithms behind generative AI. The company hired Google's top AI executive, John Giannandrea, in 2018.

Apple typically keeps its plans under wraps even as major rivals like Microsoft, Google, and Amazon boast billions of dollars in advanced technology. But insiders say the company is also developing its own large language models—the technology that underpins AI chatbots like OpenAI’s ChatGPT.

In the summer of 2023, Apple CEO Tim Cook told analysts that they were "investing in a range of AI technologies" and investing and innovating "responsibly" when it comes to new technology.

Apple's goal appears to be to run AI on mobile devices, allowing chatbots and AI apps to run on the phone's hardware and software rather than being powered by cloud services in data centers.

That technological challenge requires reducing the size of large language models (LLMs), as well as requiring higher performance processors.

Other device makers have been moving faster than Apple, with both Samsung and Google releasing new devices running generative AI features on phones.

Apple's Worldwide Developers Conference (WWDC), typically held in June, is expected to be the venue for the introduction of iOS 18. Morgan Stanley analysts expect the new operating system to be geared toward generative AI and could include the LLM-based Siri voice assistant.

Igor Jablokov, CEO of enterprise AI firm Pryon and founder of speech recognition company Yap, said Apple tends to wait for a confluence of technologies. They may offer one of the best representations of that technology.

Apple also unveiled new chips with more power to run generative AI. The company said the M3 Max processor for MacBooks “unlocks workflows previously impossible on laptops,” allowing AI developers to work with billions of data parameters, for example.

The S9 chip for the new Apple Watch models, due out in September 2023, will allow Siri to access and log data without a network connection. The A17 Pro chip in the iPhone 15, which will be released at the same time, also has a neural engine that is twice as fast as previous generations.

From an architectural and design perspective, it's clear Apple is moving toward AI, according to Dylan Patel, an analyst at semiconductor consultancy SemiAnalysis.

Apple researchers published a paper in December 2023, announcing a breakthrough in running LLM on-device using Flash memory, meaning queries could be processed faster, even offline.

Two months earlier, they released “Ferret,” an open-source LLM in collaboration with Columbia University. Ferret is currently only available for research purposes and essentially acts as a second pair of eyes, telling users what they’re looking at, including specific objects in images.

Amanda Stent, director of the Davis Institute for AI at Colby College, said one of the problems with LLM is that it only experiences the world through text. So Ferret is exciting because it “can literally start to connect language to the real world.” But at this stage, the cost of running a “reasoning” query would be prohibitive, Stent said.

Such technology could be used as a virtual assistant that tells a user what brand of shirt someone is wearing on a video call and then orders it through an app.

Microsoft recently surpassed Apple to become the world's most valuable listed company as investors cheered its moves in AI.

However, Bank of America analysts upgraded Apple’s stock last week, expecting the iPhone upgrade cycle to be fueled by next-generation AI features expected to arrive this year and in 2025.

Laura Martin, a senior analyst at investment bank Needham, said Apple's AI strategy would be "for the benefit of the Apple ecosystem and to protect a strong user base."

(According to FT)