Amazon is executing a strategic infrastructure overhaul at its rural Indiana data center campus, consolidating multiple facilities into a unified AGI SuperCluster to accelerate its next-generation frontier AI development. According to internal planning documents, the initiative, dubbed the AGI Pivot, aims to deploy over 6,000 Trainium-powered AI servers across a redesigned networking, storage, and fiber-optic framework. The consolidation transforms independent buildings into shared annexes, allowing neighboring facilities to pool core infrastructure and operate as a single high-performance computing environment. The project directly supports Amazon's AGI division, which is racing to train its next major model before year-end, with a targeted launch window aligned with the early December re:Invent conference. Despite recent workforce reductions within the AGI organization and the phased retirement of the Nova model series, internal timelines indicate an accelerated deployment schedule designed to overcome industry-wide capacity constraints. The redesign also incorporates plans to replace older Trainium 2 systems with the newer Trainium 3 processors, while expanding storage capabilities to accommodate the intensive data requirements of multimodal AI training. Strategically, the AGI SuperCluster reflects CEO Andy Jassy's emphasis on maximizing long-term returns on AWS infrastructure investments. By reusing existing data center footprints and refreshing hardware on standardized cycles, Amazon intends to extend the productive lifespan of its facilities while meeting surging compute demand. The Indiana campus, which also hosts Project Rainier, a Trainium-based supercomputer dedicated to Anthropic, will continue supporting existing customer workloads without disruption. The infrastructure push coincides with Amazon's broader financial commitment to artificial intelligence. The company recently revised its 2026 capital expenditure forecast upward to $220 billion, citing persistent demand for AI computing and rising component costs. Parallel to this, Amazon's custom silicon division, encompassing Trainium and Graviton processors, is projected to generate over $25 billion in annual revenue. The AGI Pivot underscores the company's continued reliance on proprietary hardware to scale frontier model development independently of third-party chip manufacturers. This consolidation effort builds upon previous AWS infrastructure optimization programs, such as Houdini and Titus, and signals Amazon's intent to maintain a competitive edge in large-scale AI training through architectural efficiency and vertical integration.