Apple announced in June that it will provide the Apple Intelligence intelligent system on iOS 18, iPadOS 18, and macOS Sequoia systems. IOS 18 will be released in the fall of this year, and soon Apple Intelligence will officially provide AI functionality for some Apple phone users.
Although Apple has stated that it will provide Apple Intelligence for the above systems for free, it has not answered the question of how Apple Intelligence will bring profits to the company in the long run.
On August 8th, in an interview with First Financial News, Neil Shah, Partner and Vice President of Research at Counterpoint Research, a market research firm, stated that Apple's average software and services gross profit margin exceeds 70%. It is expected that at some point, Apple will have to make profits through Apple Intelligence or related services to maintain the high gross profit margin and operating profit that Apple currently has. Based on the value created by Apple Intelligence for users and Apple's strong marketing capabilities, the ideal price range for Apple Intelligence to charge may be starting at $9.99 per month or higher.
According to Apple's previous introduction, Apple Intelligence can bring many AI functions to Apple devices. For example, based on personal information and scenarios, Apple Intelligence can retrieve and analyze the most relevant data from various apps, such as learning relevant information from chat records and assisting users in checking traffic conditions and scheduling. Apple Intelligence can be driven by devices using A17 Pro chips and M-series chips. In situations where stronger computing power is required, Apple Intelligence also introduces a server based model to process requests, and the relevant data is sent to servers equipped with Apple chips for processing. The model runs on these servers, and Apple also uses Private Cloud Compute technology.
This means that Apple not only has the cost of training models, but also needs to provide additional support such as servers for Apple Intelligence after selling hardware devices.
Integrating Apple Intelligence into Apple devices is an expensive investment that requires a significant amount of research and development resources to develop and train models with billions of parameters, ensuring that data can be securely stored in Apple's' private cloud ', and ensuring future interaction with a large number of Apple applications centered around Apple Intelligence. "Neil Shah explained that this brings huge cost expenditures to Apple, and Apple cannot rely on increasing the hardware prices of its product portfolio to cover these costs.
We believe that Apple may have multiple business models. On the one hand, it integrates AI with Siri through 'App Intents', using Apple chips on devices or cloud AI computing power to charge developers. On the other hand, it can allow users to pay for the use of advanced AI features in Apple One subscription functions or individual advanced AI features. Neil Shah believes that considering that Apple has some of the highest quality and high-end users, some users will be willing to pay for Apple Intelligence.
Neil Shah estimates that based on Apple's performance and the popularity of Apple Intelligence, it may directly drive a 10% to 15% increase in Apple's service revenue in the coming years, which could already reach the $100 billion mark annually.
The reporter contacted the relevant person in charge of Apple Intelligence regarding whether it will be charged and how it will be priced. As of the time of writing, there has been no response.
From the progress, at the end of July, Apple has gradually pushed the Apple Intelligence beta version to American users. Apple CEO Tim Cook recently stated that the beta version will only release a portion of Apple Intelligence features, and more features and languages will be released within the year. Developers are already using Apple Intelligence for development work.
Apple is also promoting model development. Apple recently released a technical paper introducing the fundamental big model developed by Apple to support Apple Intelligence, including an AFM on device end-to-end model with approximately 3 billion parameters that can run on the device side, as well as a larger parameter model AFM server that can run on Apple Cloud servers.
Besides Apple, other mobile phone manufacturers are also developing and adding AI features. Will these hardware manufacturers also charge users in the future? Neil Shah told the reporter that similar to OTT manufacturers (providing application services to users through the Internet) and application developers who provide subscription models for their advanced software or functions, AI also needs to be commercialized. OEMs (Original Equipment Manufacturers) consider commercialization based on the value of their services and user base in order to optimize their hardware, whether using their own AI models or collaborating with third parties. But not all companies can directly commercialize like Apple, and other manufacturers may choose a revenue sharing model led by partners and may not directly charge users.