Nvidia plans to invest $100 billion into OpenAI. That's uh, a lot of money. (www.businessinsider.com)

Nvidia announced plans to invest up to $100 billion to build at least 10 gigawatts of AI data center capacity running Nvidia accelerators to supply OpenAI with large-scale compute. The deal ties a massive capital commitment to physical infrastructure — tens of thousands of GPUs, power, cooling and networking — and was enough to lift Nvidia’s stock on the news. Nvidia has framed the 10 GW figure as roughly comparable to the compute capacity it expects to ship in a year, underscoring the sheer scale of the deployment. For the AI/ML community this is notable because it effectively locks in a massive, vertically integrated compute supply for one of the largest model builders, enabling faster training and higher-volume inference for next‑generation models. Technically, 10 GW is a huge grid-level burden that requires data-center design, energy sourcing, and supply-chain coordination at an unprecedented scale; it also deepens concentration of GPU-based inference/training capacity around a single vendor partnership. The move accelerates the compute arms race (and potential vendor lock‑in), raises operational and regulatory questions about power and geographic distribution, and signals that finance for raw compute infrastructure is now a central lever in AI competitiveness.