Having seen through the profound meaning behind this Google management restructuring, Jeff resigned last night, and Google subsequently announced a major adjustment to the management structure of Google DeepMind. The specific personnel and responsibilities changes are as follows: 1) Demis Hassabis (former CEO of Google DeepMind and Nobel laureate) stepped down from his daily operational responsibilities and became the Chairman of GDM and Chief Scientist of Alphabet, continuing to lead Isomorphic Labs (an AI drug discovery subsidiary). 2) Koray Kavukcuoglu (formerly CTO of GDM and Chief AI Architect of Google) has been promoted to Senior Vice President of Google DeepMind, reporting directly to Sundar Pichai (no longer holding the independent title of GDM CEO). Fully responsible for Gemini model development, cutting-edge AI research, Gemini applications, and developer team. It can be seen that after the adjustment, DeepMind's daily operations are more closely integrated into the Google system (SVP reports directly to the CEO), and strategic/scientific and product execution responsibilities are further separated. What is the intention behind discussing Google or chopping firewood in depth? 1. On the surface, it appears to be an adjustment caused by the delay of the Gemini model. Flagship versions such as Gemini 3.5 Pro were originally scheduled to be released around June 2026 (Pichai had previously stated "next month" on I/O), but were delayed due to insufficient capabilities (especially coding, reasoning, and agentic performance) that did not meet internal goals and required additional iterations and testing, and have not yet been fully launched. The company is simultaneously promoting lightweight versions such as Flash and Gemini 4 training. Delay triggers internal setbacks and external doubts about competitiveness. During this time window, the daily product/model delivery (Gemini development) will be more clearly handed over to Koray, who focuses on operations/technology implementation, while Hassabis will withdraw from strategy and science (reducing daily management burden), and handle the departure of core figures such as Jeff Dean. This objectively achieves a clearer division of responsibilities 2. The ultimate goal should be to prepare the organization for the next stage of AI, transitioning from current generative models to stronger agentic systems and AGI 1) Separation of Strategy and Operations Hassabis, as the 'Chief Scientist+Chairman', focuses on the big picture, AGI safety/impact, global policy dialogue, and scientific applications (especially health/medicine), avoiding being dispersed by daily operations; Koray focuses on product implementation and cutting-edge model delivery, reporting directly to the CEO to enhance execution agility. 2) Strengthening the advantages of full stack and scientific implementation Utilize Google's complete capabilities from infrastructure, cloud, models to applications, while accelerating scientific breakthroughs such as Isomorphic, to demonstrate the practical value of AI for humanity. 3) Organizational integration DeepMind is more deeply embedded in the Google system, reducing the color of "independent laboratories" and strengthening unified command. Previously, DeepMind had stronger independence, and after adjustments, it was able to directly control and integrate with Google's business more effectively. Faced with fierce competition, large companies naturally tend to adjust their organizational structure first. A good organizational structure that prioritizes things over people can unleash powerful energy. This also marks a rebalancing of Google's AI organization from "research driven+integrated" to "strategic AGI oriented+strong product execution". This article is sponsored by @ bitget_zh, titled 'Bitget Buying US Stocks: Instant Entry, Smooth Trading'