Alphabet Inc. Class C (GOOGL.US) teams up with AMD (AMD.US) to create the next generation of AI chips, with Wedbush stating, "The industry may undergo significant changes."

Recent reports indicate that Alphabet Inc. Class C (GOOG.US, GOOGL.US) is collaborating with AMD (AMD.US) to develop its next-generation Tensor Processing Unit (TPU). Wedbush Securities believes that if this collaboration is confirmed, it could lead to significant changes in the semiconductor industry. Wedbush Securities analyst Matt Bryson wrote in a report, If Alphabet Inc. Class C chooses to partner with AMD instead of Broadcom Inc., Marvell Technology, Inc., or Intel Corporation (the latter of which has recently been vocal about such transformation projects), we believe this would be a noteworthy trend. We also want to point out that this rumor itself highlights the importance of two aspects: 1) ASIC design capabilities; 2) IP modules. This suggests that in an era where hardware resources are increasingly constrained, competition in the chip manufacturing sector is not a zero-sum game. Semiconductor research firm SemiAnalysis believes that Alphabet Inc. Class C will leverage AMD to design the tenth-generation TPU, capitalizing on AMD's CPU expertise and its technological advancements in packaging and networking. Alphabet Inc. Class C released two eighth-generation TPUs this AprilTPU v8t for AI model training and TPU v8i for inference operations. It is worth noting that although traditional large language model (LLM) training is still dominated by accelerators, the growing demand for inference tasks and reinforcement learning for agent-based AI is deepening the market's reliance on general computing resources. Therefore, some analysts speculate that Alphabet Inc. Class C may intend to create a tenth-generation TPU specifically optimized for reinforcement learning and agent workloads, with some building blocks provided by AMD. On Tuesday before the U.S. stock market opened, Alphabet Inc. Class C fell nearly 1%, while AMD dropped over 3%.