Why did Apple choose an external LLM? The 'AI winning strategy' changing with the Siri overhaul
On January 12, 2026, the 'AI alliance' between tech and healthcare moved in a big way.Appleadopted Google's Gemini for the Siri overhaul, while simultaneouslyNVIDIA and Eli Lillyinvested $1 billion over five years in a joint research lab for drug discovery. Furthermore, in retail,Walmartlaunched AI shopping within Gemini, expanding the scope of generative AI from 'answering' to 'making people buy/making things'.
1. Apple x Google: The impact of Siri choosing an 'external model'
1-1. What happened
The core of the report is simple.Apple and Google have entered into a multi-year AI partnership to integrate the Gemini model into the next generation of Siri. Apple is said to be proceeding with updates within a framework of on-device processing and private cloud, while maintaining its 'privacy standards'.
1-2. Why 'vertically integrated Apple' went outside
Apple has traditionally been a symbol of 'self-sufficiency.' Even so, they chose an external model because the requirements for generative AI have become an all-out war involving (1) intelligence (model performance), (2) distribution power (2B+ device scale), and (3) operational power (cloud/inference). A Bloomberg program also reported that 'Apple chose Gemini as the model to power Siri this year' (program summary).
2. NVIDIA x Eli Lilly: AI enters the 'front lines of drug discovery'
2-1. What is the goal of the '$1 billion lab'?
NVIDIA and Eli Lilly announced they will build a new joint research facility in the San Francisco Bay Area and invest $1 billion over five years.The goal is simply to 'speed up drug discovery.' Reuters reports that this lab will utilize NVIDIA's latest generation of chips (Vera Rubin) and other technologies to accelerate drug discovery and development.
2-2. The value of 'AI drug discovery' lies in the discovery itself
In the program, a statement from the Eli Lilly CFO was introduced to the effect that 'the ultimate vision is drug discovery,' and the potential for finding targets and treatments for intractable diseases was discussed. This is important; the effect of AI is shifting from 'clinical and internal efficiency' to the realm of finding candidates that humans alone could never reach.
3. Walmart x Google: Generative AI connects 'search to purchase'
Walmart is teaming up with Google to plan an experience that makes searching for and purchasing products from Walmart/Sam's Club more intuitive from within Gemini.According to the official announcement, they will combine the intelligence of Gemini with the new experience Walmart is building to provide a shopping experience on Gemini through the Universal Commerce Protocol. This is a design that not only has 'AI return an answer' but also leads the user to the next action (purchase).
4. Summary: The 'AI hegemony conditions' that run through these three news items
Distribution: The significance of Gemini entering Apple's device scale is huge. For AI, it is not just performance but the 'entry point' that determines victory or defeat.
Inference and implementation: Finding, making, and delivering drugs—incorporating AI into real-world processes requires the operation of GPUs, software, and data. This is where NVIDIA's strengths come into play.
Commerce: Walmart's move is a signal that generative AI is going after the purchasing experience, which is 'beyond advertising and search.' The winning strategy for retail is shifting from 'what to say with AI' to 'what to make people buy with AI'.
Ultimately, the main battlefield in 2026 is not just 'model intelligence'.Who controls the 'entry point' (OS/device/app), who enters the 'front lines' (factory/research/logistics), and who holds the 'flow of money' (payment/purchase). This trinity will determine the next champion.