Mobile AI is a matter of two. At least that’s what it looks like when looking at the development of this market. Most manufacturers do not allocate their own resources to this section, and those who currently dominate this market (Apple and Google) seem to have freed up the path to dominate the hypothetical future filled with artificial intelligence functions.

Only Apple and Google are moving. Of the major mobile manufacturers, only Apple and Google appear interested in developing AI models for their devices. Apple Intelligence and Gemini, respectively, have begun offering generative AI functions on iPhones and Pixels; This makes them different from almost all competitors.

Samsung’s special case. The South Korean giant was the first to use artificial intelligence features on the Galaxy S24 at the beginning of the year, but the vast majority of integrated options were actually left to Google. However, the company continues to work on its own generative AI model and introduced Gauss2 a few days ago, although the company’s introduction to mobile phones is uncertain. However, they also have a “Compact” version of LLM, theoretically designed for smartphones.

Xiaomi. Lei Jun made clear at an event in August that his commitment to AI would depend on his recently announced alliance with Google. High-end devices in international markets will benefit from Gemini and Google Cloud. The company appears to be developing its own model, called MiLM, but for now it looks like it’s only intended to work in mainland China.

Google has a lot of votes in the Android market. What happens to Xiaomi will be the same as other manufacturers such as OPPO or Vivo also announce similar agreements with Google regarding AI functions. That’s definitely the logical path for all these manufacturers handing over the software experience to Google. They did this by first adopting Android, but we’ve also seen this in their adoption of Android’s services, apps, and components like Google Assistant. Only Samsung has tried to offer an alternative with Bixby.

Building an artificial intelligence model is too expensive. All these manufacturers have made a logical decision. Building an AI model from scratch and tailoring it to your specific needs is very expensive. We also see this in the huge investments made by companies such as Apple, Google and Microsoft in this market.

And it doesn’t seem to work for now. Not only that: the range of functions available on our smartphones is currently limited and, as with iOS notifications via Apple Intelligence, they can cause more negative than positive comments. The economic profitability of such investments is currently very low, and major tech companies have already warned that this is a long-term bet. For many companies, uncertainty about the size of these investments and their profitability means they choose not to make the move and go to partners who do.

What happens in search engines can also happen in artificial intelligence. First of all, the scenario is similar to what we already experience with search engines. Mobile phone manufacturers do not develop their own search engines, but instead make agreements specifically with Google to use the default search engines on their devices. The same could be true for AI, and after these deals we could see a future where AI functions are from Google by default.

But there’s so much competition. Unlike what is happening in the world of internet search engines, the competition in generative AI models is truly remarkable and we may see a more diverse scenario. Microsoft, OpenAI, Anthropic or Meta will definitely try to conquer our phones with their AI models because they cannot do otherwise. They will have a hard time, especially since the default options used by both Apple and Google will end up winning the game in most cases.

incentive issue. The situation does not seem likely to change in the short term, and the reason is simple: incentives. The current use of generative AI models on mobile phones is out of date, and everything indicates that the situation will not change in the short term. Maybe AI agents can suggest attractive uses to users, but until this or another trend proves it makes sense, investing billions of dollars to build and train a model looks like a very risky bet for the vast majority of mobile manufacturers.

in Xataka | Google released Gemini for iPhones. There’s quite a bit of overtaking Apple on the right