An Apple a Day Keeps the Competition at Bay
Apple have announced an interesting alternative AI solution to large language models (LLM) for their iPhone range. Apple Intelligence (API) is based less on LLM but more on learning relevance from the data it has accumulated from the two billion plus iPhone installed base. By pulling data from iMessage, email, Siri and photos, Siri's upgrade may be significant given Siri’s ability to better understand natural verbal conservations and to perform tasks by referencing relevant information from other applications.
Not only is it cheaper than LLM (API does not require Nvidia GPUs) but, arguably, it is more useful to consumers. API will be used to further differentiate functions between high- and low-end models and – one would have thought - motivate users to upgrade to the latest models thus boosting both volumes and average selling prices. We are led to believe Apple is making API available only on iPhone 15 pro and pro max models. One would expect that API is likely to build further ‘stickiness’ to the iOS ecosystem. Apple also announced a non-exclusive relationship with Open AI whereby requests that require ‘off-device’ knowledge can be sent to ChatGPT with the user's consent.
The Android camp will almost certainly try to catch up but faces several disadvantages. The Android camp continues to suffer weakness whereby the Original Equipment Manufacturers (OEM) do not design the silicon. As a result, the hardware roadmap will continue to be driven by the likes of Qualcomm and Mediatek. The smaller installed base of any single Android OEM also severely limits the AI data ‘fishing pond’ compared to Apple.
That said, China (a major market for Apple, accounting for over 20% of revenues) has a rather different ecosystem with most iPhone users using WeChat rather than iMessage or emails. Also, China will, undoubtably, have more regulatory monitoring of any on-device AI which may impact API’s performance.
If you would like to receive The Daily Update to your inbox, please email markets@epicip.com or click the link below.