SAN FRANCISCO — In a potential shakeup for the tech world’s balance of power, Apple and Google — the industry’s archenemies turned reluctant partners — are engaged in talks that could bring Google’s cutting-edge generative artificial intelligence to the iPhone’s billions of users worldwide.
The two companies have discussed integrating Google’s flagship Gemini language model into future iOS software, three people with knowledge of the sensitive negotiations said, speaking on condition of anonymity. While the outlines remain fuzzy and no deal is certain, the mere possibility highlights how rapidly evolving AI is blurring boundaries and bonding erstwhile rivals.
On the surface, an Apple-Google pact around generative AI seems an unholy alliance — a bit like Samsung licensing smartphone chips from Intel, or Coke and Pepsi merging their soda fountains. The open-ended chit-chat of AI models like Google’s Gemini and OpenAI’s viral ChatGPT represents an existential threat to Google’s core search business, which profits by serving up simple query responses. It’s also a product category Apple has been slow to enter.
And yet, reality is forcing unusual shifts. Google’s AI efforts have struggled to match the public frenzy around ChatGPT, while Apple runs years behind in delivering its own generative AI chops. With huge competitive threats looming from Microsoft’s aggressive tech investments, an alliance of convenience could make sense — at least temporarily.
“No one could have imagined this even a year ago, but AI has changed all the rules,” said Samier Thinks, an analyst at Strategic Foresight Analytics. “We could be witnessing the start of ‘AI exceptionalism’ — a new landscape where the usual corporate rivalries dissolve because the payoffs are so massive for collaborating rather than competing.”
Others disagree that an Apple-Google deal would be quite so seismic. The two companies have maintained close partnerships around core services like Maps, Search and YouTube on iPhones for over 15 years, despite their heated competition in smartphones and other markets. Getting Google’s language models onto iPhones could just be an extension of that lucrative relationship.
“This wouldn’t necessarily be new ground, just the latest example of Apple and Google making nice when it benefits their shared interests in serving users and raking in subscription revenue,” said Rachel Clawson, an independent analyst studying Big Tech’s AI plays.
What is clear is the sheer implications of generative AI for the world’s most popular consumer technology. The ability to summarize books, compose essays, code software, analyze data and engage in open-ended dialogue could transform how we work, communicate, create and learn using our pocket supercomputers.
For Apple, adding an advanced language model to the iPhone’s capabilities would be a way to match up with rivals like Google, Samsung and Microsoft that have rushed to integrate generative AI into their newest devices and cloud services over the past year. The urgency has only grown since Microsoft’s multi-billion dollar investment in OpenAI turbo-charged the AI arms race.
“Apple can’t afford to be left behind on generative AI if it wants to maintain its premium product positioning,” said Wayne Lam, an independent analyst on mobile tech. “Sure, they could try building their own from scratch. But taking a proven shortcut via Google might be a more realistic move to get it out to consumers faster.”
At the same time, a deal could be a coup for Google in its efforts to make Gemini a leading AI platform that developers and companies build upon — much as Microsoft is doing by opening up OpenAI’s models through cloud services. Getting onto the iPhone’s vast global installed base would make Gemini exponentially more accessible.
Motives aside, challenges abound in bringing a large language model to smartphone software and apps. Among the biggest: governing how such a system operates to prevent safety risks and unintended harmful outputs.
In just the past few weeks, Google’s Gemini image generator was found to produce racist depictions of people and had to be temporarily disabled. Issues like data privacy, energy-hogging computing needs, and AI’s propensity for amplifying misinformation and biases also loom large.
“I can see why Apple and Google might want to work together on shipping powerful AI to end users, but they’ll have to be exceptionally careful about doing it responsibly,” said Aida Moallem, an AI governance expert at UC Berkeley.
“These models weren’t really designed for mobile use by mainstream consumers. So there are going to be a lot of places where they fall short or potentially cause problems when operating at that scale,” Moallem added. “Getting it right will require a total rethinking of how we govern and update these systems in the wild.”
Both Apple and Google have touted their principled stances on AI ethics and safety, though critics argue their rhetoric often exceeds their actions. Apple’s famed obsession with consumer privacy could face new tests as data-hungry language models are integrated into its locked-down ecosystem.
And while Google likes to tout its AI principles around accountability and mitigating harm, incidents like Gemini’s biased image issues show how hard it can be to align billions of language model parameters with human values.
Antitrust regulators are also likely to scrutinize any partnership closely. The U.S. Department of Justice’s ongoing lawsuit against Google partly hinges on the search giant’s payments to Apple — accusing it of anti-competitive behavior aimed at putting a “vice grip” on search distribution worldwide.
Adding generative AI would only deepen that vice grip, critics could argue. At the same time, the technology’s sweeping effects on software, computing and digital services may justify a degree of cooperation that would have once sparked bigger regulatory concerns.
“In the AI era, regulators may start affording tech giants like Apple and Google more leeway to coordinate on developing newer foundational technologies in ways that wouldn’t have flown for their existing monopolistic domains,” said Cecilia Nguyen, an AI policy analyst at Stanford.
Or as analyst Clawson puts it, with a dose of skepticism at potential rationalizations: “Of course Apple and Google will claim their AI friendship is about best serving users and making the world better. But let’s not kid ourselves. This is about power, profit and survival in an industry being remade before our eyes.”
In the months and years ahead, it will become clear whether generative AI lives up to its technological hype and economic promise, or goes down as another over-hyped fad destined for the tech debris pile. For now, Silicon Valley’s giants are making wild pivots simply to keep up with an unleashed new force that is changing the game daily.
For Apple, long accustomed to celebrating “magical” new product releases each year, an alliance with Google to bring Gemini-powered AI to iPhones could rank among its most bewitching moves yet. But like all modern sorcery, it comes with its share of responsibility — and spells only the tech giants themselves will be able to master.
