Apple effectively brought their software and apps up to AI speed, closing the AI gap that has dampened investor confidence over the past year. Although the features won’t be available to late this year, I believe they will be enough to ignite a 2-3 year upgrade cycle starting in CY25.

For those who have closely followed the rapid progress of generative AI, Apple’s AI features may be considered playing “catch up.” That said, most Apple buyers are everyday people who have not experienced generative AI, including text composition tools, image generation and editing, along with organization tools that proactively manage our digital lives. A critical part to the Apple AI value proposition is the company’s unique position in having access to the user’s data (email, messages, and photos) to train their own models to personalize the user experience. Additionally, Apple is the only company that has control of the device ecosystem, which means Apple will make AI easy to use, a dynamic that will resonate with consumers. 

Siri will further Apple’s position as a player in AI. Its long-awaited update makes it conversational, both with voice and text. Users can now dictate to Siri within Apple apps (email, messages, photos, keynote, maps) to perform actions like finding or editing photos, creating unique navigation routes, or rewording an email to fit a desired tone. Additionally, more complex tasks can tap-in OpenAI’s ChatGPT to further explore the possibilities of generative AI. 

Here are the Apple Intelligence highlights:

The new AI features will start a slow wave of upgrades over the next 3 years. I estimate 85% of Apple’s active installed base do not have an iPhone 15 Pro or a Mac/iPad with an M1 chip or newer. This means people will eventually have to upgrade their devices to realize the full potential of Apple Intelligence.  In other words, Apple has about 1.9B active devices that will need to be updated to use these AI features. I envision an iPhone owner upgrading, falling in love with Apple Intelligence, and soon after upgrading their Mac and iPad so all of their devices can leverage the power of AI.

The features alone are not enough to spark an upgrade supercycle starting this year because they start rolling out in the Fall and Siri will be available next year. Starting in the December quarter of 2024 and into 2025, they’ll generate waves of upgrades as more and more people are exposed to their functionality. 

Apple Intelligence will be available on third-party apps too, expanding the user’s ability use AI on their devices. In other words, this will further support the case for upgrading.

Apple Intelligence will not be a new revenue leg like the iPad or Services was, but it will be an accelerator for iPhone and total revenue growth over the next several years, which will be beneficial for shares of AAPL. I expect iPhone sales growth to reaccelerate in FY24 to 5%, compared to being down on average 2% over the past year and a half, and growing 7% in FY25 and 10% in FY26. That means that overall growth goes from about 5% in FY24, to +10% in FY25 and 12-14% in FY26.

In terms of Apple’s valuation, investors are likely now expecting the company to earn around $9 in EPS in CY26 (Street is at $7.90). If we’re entering a 3-5 year AI bull market, it’s reasonable to expect multiple expansion. Investors are zeroing in on a 35x earnings multiple next year, Apple currently trades at 29x next year’s earnings.

In a year, the market will be pricing in expectations for 2026. Next year I expect a $9 (EPS) x 35 (multiple) = $315 (stock price), or a 49% upside to the current value (current price is $212). That implies about a $4.7T market cap, which is hard to fathom. Back in 2018 when Apple crossed the $1T mark, most thought $2T was impossible. As a point of reference, Apple added about $350B to their market cap in the 8 hours of trading post-WWDC.

Apple’s own models will be powering most of the showcased Apple Intelligence features. This might come as a surprise as it was well known Apple was late to the AI game and was partnering with OpenAI. However, with privacy at the forefront, Apple focused on building a small language model (SLM) that keeps as much compute on device as possible.

SLM: That SLM is a 3B parameter model that will likely power most of the AI usage including tasks we do many times an hour like composing texts, emails, playing music, taking and editing photos, looking for a file, and organizing anything digital.

LLM: For the times when users who want a “smarter AI” that is capable of more complex tasks, they’ll be able to easily access GPT-4 inside of Apple and third-party apps. As a point of reference, GPT-4 is a large language model (LLM) estimated to be 1T parameters, which means it’s about 300x bigger than Apple’s SLM.

The bottom line is anything run on-device (iPhone, iPad, Mac) or in the new Private Cloud is on an Apple model. OpenAI comes into play for more complex tasks, requiring an opt-in from the user. 

Another important tidbit is Apple did say in the future (my guess is sometime in 2025) they’ll allow users to integrate with the third-party LLM of their choice, like Grok, Gemini, or Claude.

During Apple’s March quarter earnings call, CFO Luca Maestri suggested there would be no changes to Apple’s capex plans as they build AI into their product line. His comment seemed out of touch with reality given the hundreds of billions of dollars big tech is spending each year to build out AI infrastructure. Following WWDC we got a better sense about how Apple can thread the needle between quickly getting up to speed in AI and at the same time maintaining profit margins.

The approach is twofold. First, make up lost ground by partnering with OpenAI. That relationship gives OpenAI distribution to Apple’s 2.2B active devices. In turn, Apple gets to use GPT-4 for free (as reported by Bloomberg’s Mark Gurman). Second, the hardware that runs Apple’s proprietary SLM is the device itself which of course is paid for by the consumer. Apple is building a private cloud, but I suspect the scale will be small in terms of capex needs.

This is yet another example (including the ~$15B Google Search deal) of how Apple has monetized its installed base.

The genius behind Apple Intelligence is removing the friction of using generative AI. While chatbots like ChatGPT may seem intuitive, there is a learning curve in prompting models to get the best and logical answers. My sense is there are a lot of people out there who are uninterested, don’t know how to begin, or are intimidated by AI. Apple is making it so easy for users that most of the time you won’t even realize you’re using AI. 

Taking a step back to Apple’s competitive advantage, Apple’s hardware, software, and services ecosystem is unmatched. Users’ accounts are integrated across all devices and tasks can start on one device and be finished on another. 

Lastly, and importantly, is Apple’s rooted identity in security. The Keynote emphasized their commitment to security, which was intentional with the addition of a third-party partner for the first time and added reliance on cloud computing for complex AI tasks. Privacy is another core competency Apple cannot afford to lose. To users, AI may seem daunting in terms of how their data is used and stored. Apple will utilize their own models via Apple Silicon to do as much compute on-device and in the Private Cloud to keep compute and data in-house. OpenAI’s ChatGPT is available for the more complex tasks, in which users opt-in and know when this is occurring, and only sending the relevant and necessary data to process the prompt. 

Bottom line is Apple’s ecosystem stack of hardware, software, and services will all utilize Apple Intelligence to be, “Powerful, Intuitive, Integrated, Personal, and Private.”

Disclaimer