Apple is quietly but steadily ramping up its efforts in the realm of artificial intelligence, particularly focusing on integrating generative AI capabilities into the next generation of iPhone, according to a new report by Financial Times. This push is evident through a series of recent acquisitions, talent recruitment, and hardware updates.
Industry data, academic papers, and insights from experts suggest that Apple has prioritized tackling the technical hurdle of running AI directly on mobile devices. Compared to its tech rivals, the Cupertino company has been particularly active in acquiring AI startups, with 21 deals since 2017, according to PitchBook. Notably, their early-2023 purchase of WaveOne, a specialist in AI-powered video compression, further underscores this direction.
Analysts anticipate significant M&A activity from Apple in the AI space going forward. “They are gearing up for some substantial deals,” says Daniel Ives of Wedbush Securities. “I’d be surprised if they don’t make a sizable AI acquisition this year, given the ongoing AI arms race.”
Apple’s AI job postings hint at their focus on generative AI, with nearly half mentioning “Deep Learning,” the core technology behind these models capable of generating human-like text, audio, and code. Notably, they lured Google’s top AI executive, John Giannandrea, in 2018. Giannandrea is now Apple’s Senior Vice President of Machine Learning and AI Strategy.
While competitors like Microsoft, Google, and Amazon openly advertise their multi-billion dollar investments in AI, Apple has maintained its characteristic secrecy regarding its plans. However, industry insiders believe they are developing their own large language models (LLMs), the technology powering generative AI products like OpenAI’s ChatGPT.
Apple CEO Tim Cook acknowledged their broad research in AI last summer, emphasizing responsible investment and innovation in this new field. Their active goal is to run generative AI directly on iPhones, enabling AI chatbots and apps to function independently on the device’s hardware and software, rather than relying on cloud services.
This on-device approach presents a complex technical challenge, requiring both miniaturization of LLMs and more powerful processors. Other device manufacturers, including Samsung and Google, have already released products claiming on-device generative AI features, putting pressure on Apple to catch up.
Apple is expected to unveil the new iOS 18 operating system alongside other updates at WWDC 2024 as has been the case for the last several years. Morgan Stanley analysts predict that iOS 18 will be geared towards enabling generative AI and could potentially feature an LLM-powered Siri.
Apple’s recent chip releases further support their AI ambitions. The M3 Max processor for MacBooks, unveiled in October, promises “workflows previously not possible on a laptop,” particularly beneficial for AI developers working with massive datasets. Additionally, the S9 chip in new Apple Watches allows Siri to access and log data without an internet connection, while the A17 Pro chip in the iPhone 15 boasts a neural engine twice as fast as previous generations.
In December, Apple researchers published a paper announcing a breakthrough in running LLMs on-device using Flash memory, enabling faster query processing, even offline. Additionally, they collaborated with Columbia University to release “Ferret,” an open-source LLM currently limited to research purposes. This tool acts as a “second pair of eyes,” describing images the user sees, including specific objects.
However, limitations remain. “LLMs primarily experience the world through text,” notes Amanda Stent, director of the Davis Institute for AI at Colby College. “Ferret’s ability to connect language to the real world is exciting, but the cost of running such queries is currently immense.”
Potential applications for this technology include virtual assistants that identify clothing brands during video calls and facilitate online purchases through integrated apps.
Microsoft’s recent rise to the top spot among the world’s most valuable companies is partly attributed to investor enthusiasm for their AI initiatives. Despite this, Bank of America analysts recently upgraded their Apple stock rating, citing expectations that iPhone upgrade cycles will benefit from the introduction of new generative AI features in 2024 and 2025.
While Apple may not be the loudest voice in the AI race, their strategic acquisitions, talent recruitment, and hardware advancements suggest a determined push towards integrating generative AI directly into iPhones.
