Apple Inc. (AAPL) plans to invest approximately $1 billion annually to license and use Gemini, an artificial intelligence model developed by Alphabet Inc.’s Google (GOOGL) with a parameter scale of up to 1.2 trillion. This move aims to drive a comprehensive upgrade for its long-developed Siri voice assistant. It is reported that after in-depth evaluation, the two parties are close to reaching a final agreement. The core functions of information integration and task execution for the new version of Siri, expected to be released next spring, will be powered by the Gemini model.
Previously, Apple had also considered and tested other third-party models, including OpenAI’s ChatGPT and Anthropic’s Claude, but decided to proceed with Google earlier this year. However, according to Apple’s own statements, integrating Google’s technology is viewed only as a transitional arrangement. The ultimate goal remains to replace it with an in-house model once it possesses sufficient capabilities. Compared to the 150-billion-parameter model currently used in Apple’s intelligent cloud services, the custom Gemini system from Google represents a leap in parameter scale by an order of magnitude. This significantly enhances the system’s data processing power and its understanding of complex contexts, thereby making Apple’s current in-house models appear inferior.
Although this collaboration is strategic, it is not expected to be highly publicized. Google will function as a backend technology provider, a role different from the public partnership where it is the default search engine in the Safari browser. Furthermore, the current agreement also differs from previous negotiations about integrating Gemini directly into Siri as a chatbot – those discussions, though once close to fruition, did not materialize into an actual feature. Additionally, the collaboration does not include integrating Google’s AI search into Apple’s operating systems.
During recent earnings communications, Apple CEO Tim Cook mentioned that Siri might offer chatbot options other than ChatGPT in the future. Nonetheless, the company explicitly does not view Gemini as a long-term solution. Despite challenges such as the loss of AI talent, including the departure of the model team lead, Apple’s management remains determined to continue developing its own AI technology and hopes to eventually replace Gemini with an internal solution. Currently, the company’s model team is focused on developing a cloud model with trillions of parameters, striving to deploy it in consumer products by next year. Apple executives believe this model has the potential to match the quality of the custom Gemini version. However, analysts point out that Google’s Gemini 2.5 Pro currently leads in most large language model benchmark tests, suggesting that catching up will not be easy.
