From selling chips/systems to helping build AI factories, the ambitious deal that emerged during the AI boom at the end of July has finally been finalized: 1) NVIDIA announced a $1.5 billion investment in SB Energy (joining SoftBank and OpenAI as shareholders); 2) SB Energy will be responsible for building, owning, and operating data centers, signing a 20-year lease with OpenAI; 3) NVIDIA will provide credit support/residual value guarantees for LPS (Land, Power, and Shell) construction (initially 4.25 GW, with expansion options covering an additional ~3.75 GW, totaling ~8 GW); 4) Capacity will come online in phases, expected to start in 2028. Jensen Huang believes that LPS construction has become the next critical strategic resource for AI factories. In the AI economy, compute power equals revenue, and AI factories require a complete resource stack: advanced chips, packaging, memory, networking + LPS. Most customers still handle LPS on their own, but cutting-edge AI labs have massive demands for training and inference compute power, growing faster than their balance sheets and long-term credit capacity. NVIDIA is stepping in to help secure physical infrastructure. This also shows the evolution of NVIDIA's role: From chips → systems → networking → CUDA → full-stack AI factories → now further upstream, helping secure the "long-life foundation" of AI factories: LPS. This marks NVIDIA's official shift from "selling chips/systems" to "helping customers securely build the physical layer of entire AI factories." Last week, NVIDIA announced https://(x.com)/qinbafrank/status/2087182111832998287?s=46&t=k6rimWsEbo2D2tXolYcM-A the establishment of an "independent compute power financing platform" in collaboration with six major financial institutions, which is essentially an upgraded version of this partnership with SB Energy and OpenAI. #AI #NVIDIA #OpenAI #SBenergy #LPS #ComputePower