inference

AI chip startup Groq is reportedly raising $650 million as it pivots from hardware to focus on AI inference services. This matters to investors because it shows the competitive pressure in AI infrastructure following Nvidia's recent major acquisitions and the shift toward specialized inference solutions.

Google has unveiled two new Tensor Processing Units (TPUs) designed for the 'agentic era,' with separate chips optimized for AI inference and training workloads. These specialized processors could accelerate AI model deployment and training efficiency compared to general-purpose chips.