Google launches two new AI chips to challenge Nvidia

Google announced two new specialized AI chips for its cloud computing services. The TPU 8tx, developed with AMD, focuses on confidential computing—keeping sensitive data and AI model weights protected at the hardware layer while training models. The TPU 8i Max essentially doubles the capabilities of Google's existing inference chip by combining two units into one package, offering more memory and processing power for handling larger AI queries.

These chips represent Google's latest move in the intensifying race among tech giants to reduce dependence on Nvidia's dominant GPUs. Microsoft, Amazon, OpenAI, and Anthropic are all developing their own AI-specific semiconductors. Google's new offerings target specific enterprise needs: secure confidential computing for proprietary data and enhanced capacity for large-scale AI inference workloads—signaling a more diversified and competitive AI hardware landscape.