Gemini's Rise to Dominance in AI
The AI landscape is a fierce battleground, where only the most comprehensive and innovative strategies can lead to victory. To truly win in AI, one must excel in multiple critical areas. Firstly, a model must be among the best in the market, setting the standard for others. Secondly, it requires an immense amount of resources to continuously enhance and deploy the model on a massive scale. Thirdly, at least one AI-based product should be widely used, and ideally, multiple products should be in the market. Lastly, access to extensive user data, including personal information, online behavior, and computer files, is crucial for improvement and customization.
These requirements are challenging to meet, as evidenced by OpenAI CEO Sam Altman's plea for trillions of dollars in compute power (https://www.theverge.com/ai-artificial-intelligence/759965/sam-altman-openai-ai-bubble-interview). However, Google stands out as the company that seems to have all the necessary pieces in place. Over the past year and even in recent days, Google has made strategic moves, indicating its readiness to become the most significant and influential force in AI.
Google's infrastructure work began last year with the release of Gemini 3 in November (https://www.theverge.com/news/822977/gemini-3-google-ai-model-launch). This model is widely recognized as the best overall large language model on the market, excelling in most benchmark tests (https://www.theverge.com/report/827555/google-gemini-3-is-winning-the-ai-race-for-now). Experts agree that Gemini 3 is among the top choices for various tasks (https://www.theverge.com/report/825332/google-gemini-3-pro-ai-hands-on-test), although its reign may not be eternal, as new models emerge every six weeks. Nonetheless, Google has consistently demonstrated its ability to produce industry-leading models.
One key factor in Gemini 3's success is its training method. Google utilized its own TPUs, highly specialized chips designed for this specific purpose (https://www.theverge.com/2024/4/9/24125074/google-axion-arm-cpu-ai-chips-cloud-server-data-center). While Google faces manufacturing challenges and RAM price hikes like other companies, it is not dependent on Nvidia's supply chain. This full-stack control over its AI destiny allows Google to optimize its system for better performance, speed, and cost-effectiveness. No other company possesses this level of control.
With the technology in place, Google is now focusing on practical applications. On Monday, Google and Apple announced a partnership where Gemini will power the next-generation Siri, set to launch later this year (https://www.theverge.com/news/860521/apple-siri-google-gemini-ai-personalization). This collaboration is a significant win for Apple, as they invest $1 billion annually to transform Siri into a more useful AI assistant (https://www.theverge.com/news/814654/apple-intelligence-google-gemini-ai-siri).
Siri, with its 1.5 billion daily requests (https://www.wsj.com/video/series/joanna-stern-personal-technology/apple-software-chief-explains-ai-delays-future-of-siri-and-more/7F3883B2-0404-4618-97EE-5ABA1AC1602C), will become one of the most popular ways people interact with Gemini. While this partnership doesn't mean Gemini will fully replace Siri, it still matters. The more user activity and data these companies can collect, the better their models and products can be. This concept is similar to the recent search trial, where user data played a significant role (https://www.theverge.com/2023/10/2/23900233/microsoft-ceo-satya-nadella-us-google-antitrust-trial-testimony).
Google's other announcement this week is even more impressive. They introduced 'Personal Intelligence,' an opt-in feature that connects Gemini to Google's vast user data, including recent searches, YouTube videos, emails, photos, and files (https://www.theverge.com/news/861576/google-gemini-ai-personal-intelligence-gmail-search-youtube-photos). This feature allows Gemini to provide more accurate and personalized responses without requiring users to provide extensive context or detailed prompts. Google's extensive knowledge about its users, now shared with Gemini, is a significant advantage.
Initially, Personal Intelligence is in beta for a subset of paying AI customers, but Google aims to make it accessible to everyone. The ultimate goal is to integrate it into Google's most important product, its search engine. AI Mode in Search is a new tab alongside general search results, but Google sees it as the future of search, aiming to transform Gemini into a portal to all of Google's user data, internet information, and global knowledge.
In 2022, when ChatGPT launched, Google seemed unprepared. However, Google has demonstrated its ability to focus and marshal resources effectively. With chatbots being the future of AI, Google is now the company best positioned to compete. It has the models, resources, distribution, and data required to make its bots unique and useful. While ChatGPT has brand power and daily active users, Google has almost everything else, including the iPhone. This strategic advantage positions Google to dominate the AI landscape.
