NVIDIA has formally declared the era of "Made in America AI supercomputers." The company announced plans to invest up to $500 billion (approximately 700 trillion won) over four years to produce its core AI chips through complete supercomputer final products entirely within the United States. On April 14, 2025, NVIDIA officially announced "for the first time in company history, we have begun manufacturing fully US-made AI supercomputers." Production partners: TSMC (Blackwell chip production, Phoenix Arizona), Foxconn and Wistron (supercomputer assembly, Houston and Dallas Texas), Amkor and SPIL (chip packaging and testing, Arizona) — over 1 million square feet of production/testing space secured.
Manufacturing sovereignty: amid explosive AI demand, chip supply bottlenecks and geopolitical risks have threatened the entire AI industry. NVIDIA is positioning itself as an "AI infrastructure manufacturer," emphasizing supply chain resilience. CEO Jensen Huang: "We are building the engine of the world's AI infrastructure in the US for the first time — US manufacturing is essential to strengthening supply chain resilience and proactively responding to AI demand." Advanced technology integration: NVIDIA Omniverse for digital twin factory design; Isaac GR00T robot platform for production automation; "self-optimizing factory systems" operated by AI — presenting "AI factories built by AI."
Economic and security impact: creating hundreds of thousands of US jobs; dozens of gigawatt-scale AI factories to be built nationwide. Partners including TSMC and Foxconn elevated from simple subcontractors to strategic co-partners. "The future of AI competition is no longer decided by 'what to build' but by 'who manufactures where, with whom, and how.'" This initiative sets a new benchmark for global AI industrial order, signaling that AI is no longer a "code-writing technology" but a factory-based intelligence production system.
