xAI is addressing the scarcity of high-performance NVIDIA GPUs, crucial for the generative AI boom, by investing billions into building its own supercomputing infrastructure. This vertical integration ensures access to raw AI compute power.

A 'gigafactory of compute' refers to xAI's strategy of building massively powerful and expensive hardware at manufacturing scale for data processing. This hardware-first approach aims to own the computational resources needed for AI development.

xAI's strategy mirrors AWS by building its own internal infrastructure for AI, then productizing excess capacity by potentially leasing it to other companies. This positions xAI as a new specialized cloud provider, or 'neocloud,' for large-scale AI workloads.

xAI is reportedly planning to build a supercomputer with potentially hundreds of thousands of interconnected NVIDIA H100 GPUs. This infrastructure aims to provide immense computational resources for training its own models and potentially for external clients.

A 'neocloud' describes xAI's potential role as a new kind of specialized cloud provider, tailored specifically for large-scale AI workloads. By leveraging its supercomputing infrastructure, xAI could lease compute power, much like legacy cloud providers, but as an AI-native company.