Inside information about the delay of the next-generation Gemini model has been revealed: computing power is tight, and internal teams have differences in development focus and resource allocation. Recently, news of the delay of the next-generation Gemini model, which has attracted much attention in the industry, has been reported

Inside information on the delay of the next generation Gemini has been exposed: computing power is tight, and the internal team has differences in development focus and resource allocation. Recently, news of the delay of the next generation Gemini model, which has attracted much attention in the industry, has been reported. According to reports, the release of this flagship model had to be delayed by two months due to internal team differences over development priorities and resource allocation, coupled with tight computing power supplies and cumbersome approval processes. In terms of computing power, despite having self-developed TPU chips, computing resources are still stretched thin in the face of multiple demands for model training, cloud services, and massive consumer and enterprise-level AI products. Senior management has recently begun to adjust the organizational structure to alleviate resource conflicts among various departments. At the same time, the top management and management power within the company are also undergoing subtle changes. Co-founder Sergey Brin has recently been frequently involved in core model training, pushing resources toward "recursive self-improvement." In addition, the management reorganization further tightened the laboratory's autonomy, and core leader Kore Kavukchuoglu began to fully dominate the development direction of Gemini. With the resignation of some technical backbones, more dispersed resources are being refocused and efforts are being made to catch up with the development pace of cutting-edge AI. via AI News (author: AI Base)