You have probably already read the news. If not, you can read this new article, Apple is reportedly exploring a partnership with Google for Gemini-powered feature on iPhones. The gist is that Apple is considering partnering with external partners to use Edge LLM on its iPhones. Is this a prudent move? As they say in French, “ça dépend”.

First, it shows that the Edge LLM race will begin soon. It has kind of already begun with the latest phones from Google and Samsung but as predicted in my couple of articles, it will be in full swing within next couple of years. That is nothing but good news for consumers and businesses alike. Consumers will have access to apps with exponential capabilities, and businesses will be able to expand the realm of innovation. Phones will become more powerful though. The RAMs on flagship phones will expand, along with some hardware upgrades to further expand the realm of Edge LLMs.

However, in this article, we are going to focus on the strategic aspect of Apple’s decision to leverage an external partner’s AI. Like every strategic decision, this one has pros and cons. As a positive person, I like starting with the pros, so let’s start there.

As mentioned previously, the LLM war has already begun. It is obvious from the news that Apple’s own AI development milestones are not where they would like them to be. And the heat is on from investors, as highlighted in Episode 39 of “Think About It”. If there are almost no new edge AI-enabled features in this year’s September release, Apple will be facing tough questions from investors. Considering this, it is prudent to collaborate with partners who have already explored the possibilities of Edge LLM.

Apple’s consumers are tech enthusiasts but they are also Apple fanboys. What this means is that while they will ignore one of two iterations of advances in phones like Samsung and stay with their iPhones, they will eventually get lured by AI-enabled innovation in competing phones if Apple does not move fast. So waiting is not an option. From that aspect, it is a prudent move.

While Google does offer its own phones, it is not a significant player in the market. Android phone market is crowded and fragmented, and except Samsung, does not compete with Apple. Any user that you find has switched sides, has almost always switched to a Samsung phone. So in Tim Cook’s mind, it is not like collaborating with a competitor, which may hurt in the long term. While they hone their own AI capabilities, they will have learnings from Gemini deployment on their phones.

The pros are over. Let us discuss the cons.

The cons come into play only when Apple delays its AI ambitions or, in the worst case, ditches them.

Long-term abandonment of Apple’s AI ambitions will mean that it will be Google, which will benefit more from edge deployment learning, leveraging the massive Apple iPhone user base. Google’s Edge LLM capabilities are already productional, and hence any transferred learning from deployment on iPhones will help fine tune Google’s own Edge LLM capabilities exponentially. This means that the delta between Apple’s own Edge LLM capabilities and Google’s capabilities will only increase. Because it is not about how powerful your models are but how effectively they are performing and adding value in the real world.

That will make Google the dominant player in the Edge LLM arena, which will also percolate into the area of non-Edge LLM capabilities. This may also not bode well for end-users, as Google already dominates Android arena.

Hence, the summary is that this collaboration may be a good move if Apple plans to leverage its AI capabilities in the near term and wants to use it as a stop-gap. If that is not the case, i.e., if Apple has given up on its Edge LLM ambitions or thinks it is years away, this may be a risky bet.

Collaborating with Open AI, another option that Apple is reportedly exploring does not make much strategic sense since Open AI has its own learning curve from an Edge deployment perspective. And since it is so widely used across the development landscape, chances of what I like to call “innovation creep” exist. This option does not make much sense unless there are capabilities that are not yet public.