Google is set to launch a bold move to secure its position at the center of the artificial intelligence (AI) market. Having become a symbol of AI technology with AlphaGo in 2016, Google lost market dominance to OpenAI following the emergence of ChatGPT at the end of 2022. Accordingly, Google is expected to showcase its flagship AI model, 'Gemini 2.5,' along with an upgraded version, 'Gemini 2.5 Pro Preview,' and the multimodal AI agent 'Project ASTK' at this year's 'Google I/O (Developer Conference) 2025.'
According to foreign news reports on the 20th (local time), the I/O 2025, which opens at the Shoreline Amphitheater in Mountain View, California, is expected to focus on AI. Google has preemptively unveiled major updates related to Android 16 at a separate online show, and the main focus of this keynote presentation will be on AI. From the slogans adorning the venue to the organization of the announcement sections, everything is aligned with AI technologies, including Gemini.
The Gemini 2.5 Pro Preview, to be unveiled that day, has already had some features revealed in the form of a developer test last week. In particular, this model is specialized in large-scale code generation, interactive web app creation, and real-time bug detection, making it a developer-friendly AI tool. Currently, Google is testing an AI agent that automatically finds security vulnerabilities based on this model, and there are suggestions that live demonstrations may be possible.
Another key weapon for Google is 'Project ASTK.' First unveiled at last year's I/O, this technology analyzes real-time visual information recognized by the user's camera, responds to voice inputs, and performs tasks in a multimodal manner. This year, it is expected to be revealed in a more advanced form, and some features are currently available in the form of 'Gemini Live.' In the future, it is anticipated to be integrated with various hardware such as smart glasses and Android XR (extended reality) devices, expanding into a reality-based interface.
It is highly likely that there will be no direct announcements related to hardware. The mid-range 'Pixel 9a' has already been released, and the flagship model 'Pixel 10' is scheduled to launch in the second half of this year. However, there are forecasts that device demos, such as the MR headset 'Project Moohan' developed in collaboration with Samsung Electronics based on Android XR, may be showcased. This could be an opportunity to demonstrate how ASTK and Gemini integrate with such devices.
This technology exhibition is not merely an introduction of functions. The industry is watching whether Google can return as a technological leader in the AI era through this I/O, and how it can restructure its existing business model around AI is a key point of interest. Deepwater partner Jin Munster noted on his X account on the 19th (local time), "This I/O is an important occasion to prove that Google can overcome the 'innovator's dilemma.'"
There is fierce psychological warfare with OpenAI. Last year, OpenAI unexpectedly announced GPT-4o and Advanced Voice Mode the day before the Google I/O. In response, Google countered with the multimodal model Gemini 1.5 and ASTK, but the lead tilted towards OpenAI. This year, there were no separate announcements from OpenAI before I/O, but they continued to exert pressure by expanding the GPT-4.1 model to general ChatGPT users and proactively unveiling the Codex AI agent for developers. GPT-4.5 is currently in the preview stage, and the anticipated GPT-5, scheduled for release this summer, is also drawing industry attention.
OpenAI continues to maintain an advantage in user scale. As of March, the monthly active user count (MAU) for Google's Gemini was about 350 million, while OpenAI's ChatGPT surpassed 500 million. OpenAI expects to exceed 1 billion by the end of this year, and Meta has also stated that its AI's MAU is nearing 500 million.
The industry is paying attention to whether Google can go beyond a simple technology demonstration and convincingly explain why these functions are necessary. An IT industry insider stated, "AI is now more about what changes it can bring to users' lives rather than what it has created," adding that "storytelling strategy has become as important a keyword as technological capability."