Alphabet (GOOGL.US, GOOG.US) develops new generation AI server chips, stock price rises more than 3%

According to media reports citing informed sources, Alphabet Inc. Class C parent company Alphabet (GOOGL.US, GOOG.US) is developing a new generation AI server chip codenamed "Frozen v2", aiming to further improve the efficiency of artificial intelligence model operations. Boosted by this news, Alphabet's stock price rose over 3.5%. The report stated that "Frozen v2" will be directly integrated into Alphabet Inc. Class C's Gemini large model architecture, thereby improving the overall efficiency of the model operation. Employees involved in the project expect that the chip will be able to process a number of Tokens at unit power consumption that is expected to be 6 to 10 times that of Alphabet Inc. Class C's latest generation tensor processing unit (TPU). The efficiency evaluation is mainly based on the Token throughput provided by unit power consumption. With the rapid development of generative AI, global tech giants are generally facing the issue of AI computing power shortage. Alphabet Inc. Class C is also continuously increasing its self-developed AI chips to reduce reliance on external computing resources and improve model performance. It is reported that this new chip is expected to be deployed as early as 2028. In April of this year, at a cloud computing conference in Las Vegas, Alphabet Inc. Class C released its eighth generation TPU products, including the TPU 8t for AI model training and the TPU 8i for inference tasks, further improving its AI infrastructure layout.