Morning Edition · Thursday, July 23, 2026Published at 1:46 AM EDT · New York
Nvidia Brings First DGX GB300 Online at the US Naval Postgraduate School
The Blackwell Ultra system was donated, not procured, as Wistron opens a Fort Worth plant to build Nvidia superchips on US soil.

Jensen Huang, Nvidia's chief executive, commissioned a DGX GB300 system at the Naval Postgraduate School in Monterey, the first command in the US military to receive the platform. The Blackwell Ultra machine, paired with Nvidia Mission Control software, gives roughly 1,500 in-resident students and 600 faculty on-premises training and inference for workloads spanning weather prediction, cybersecurity, and disaster response.
The system arrived as a donation rather than through a defense procurement contract, according to reporting on the launch, a distribution choice that establishes a captive base of military researchers on Nvidia's newest architecture. On the supply side, Wistron opened a 324,000-square-foot plant in Fort Worth to build and test Nvidia superchip systems domestically, part of the broader reshoring of AI hardware assembly.
The pairing matters more than either event alone. A donated flagship node inside a military graduate school builds institutional dependence on the DGX line at the same time that US-based manufacturing reduces the geopolitical exposure of that supply chain, the reverse of the export-control dispute over Chinese access to the same GB300 parts.
What this means
Nvidia is defending performance-per-megawatt as the deciding spec by placing GB300 nodes with anchor institutions and moving assembly onshore, which raises switching costs for buyers weighing AMD or custom silicon and hardens its position against export-driven demand shocks. The military, national labs, and regulated buyers gain owned frontier clusters rather than rented cloud capacity, sustaining demand for the newest platform through the capital channel rather than the consumption channel.
What to watch
- Whether other US service branches and national labs follow with GB300 procurements, which would signal owned-cluster buying is becoming standard for regulated and defense workloads.
- Wistron's ramp rate and yield in Fort Worth, since domestic assembly capacity is the binding constraint on how fast reshored supply can offset any tightening of overseas export routes.
Observations to monitor, not financial advice.
Synthesized from: NVIDIA Blog · NVIDIA Blog (Wistron) · Nextgov/FCW
Part of a tracked trend
Enterprises Build Owned Accelerator Clusters
Regulated industries increasingly buy dedicated frontier-scale accelerator infrastructure rather than renting cloud compute, sustaining demand for the newest Nvidia platforms and making performance-per-megawatt the deciding spec.
More from this edition
- White House Accuses Moonshot of Distilling Anthropic's Fable to Build Kimi K3, Treasury Threatens Sanctions
- OpenAI, Google, and Meta Pour Compute and Credits Into the Genesis Mission for National Science
- Judge Gives Final Approval to Anthropic's $1.5 Billion Settlement Over Pirated Training Books
- OpenAI Launches Presence, an Enterprise Agent Platform, as Codex Cuts NTT DATA Incident Analysis to 30 Minutes
- New Benchmark 'SysAdmin' Measures Whether Frontier Models Seek Power Beyond Their Task
- Two Papers Warn That Safe LLMs Do Not Compose Into Safe Multi-Agent Systems
- Anthropic Adds Screen-Recording to Teach Claude New Skills
- Tesla to Record Gigafactory Workers' Motions to Train the Optimus Robot
- Researchers Push to Standardize How Jailbreak Success Is Measured
- Study Finds Reasoning Fine-Tuning Collapses Behavioral Diversity in Game Play
- OpenAI Announces Project Camellia, a Georgia Data Center Built With Local Community Commitments