Jensen Huang: Strengthening AI Ties in South Korea

Daily Nvidia Brief — October 19, 2025

“AI factories are the new data centers.” — Jensen Huang

  1. Jensen Huang headed to APEC in South Korea (Oct 28–31)
  • What’s new: Nvidia’s CEO will meet Korean leaders and memory suppliers (Samsung, SK hynix) alongside global policymakers. (reuters.com)
  • Bigger vision: Tightening ties with HBM suppliers fortifies Nvidia’s “AI factory” supply chain from GPUs to memory. (reuters.com)
  • AI impact: Stronger HBM pipelines mean more GPUs delivered, easing bottlenecks for model training worldwide. (reuters.com)
  1. First U.S.-made Blackwell wafer celebrated in Arizona
  • What’s new: Nvidia and TSMC showcased the first Blackwell wafer produced on U.S. soil at TSMC Phoenix. Packaging still occurs in Taiwan for now. (reuters.com)
  • Bigger vision: Onshoring front‑end manufacturing builds resilience for Nvidia’s full‑stack AI factory strategy. (reuters.com)
  • AI impact: More regional capacity should reduce shortages and speed next‑gen training/inference deployments globally. (reuters.com)
  1. DGX Spark “personal AI supercomputer” now on sale at $3,999
  • What’s new: Orders opened this week; partners like Dell, HP, Lenovo and others are offering their own Spark systems. (theverge.com)
  • Bigger vision: Extends Nvidia’s platform from hyperscale down to the desktop—seeding more developers into its ecosystem. (theverge.com)
  • AI impact: Wider access to serious compute should accelerate experimentation in agents, multimodal apps, and robotics. (theverge.com)
  1. CoreWeave to deliver 40k+ GB300 GPUs to Poolside’s Texas “AI factory”
  • What’s new: Shipments begin December 2025 for a campus targeting 2 GW over time; CoreWeave is operator and anchor tenant. (barrons.com)
  • Bigger vision: Confirms Nvidia’s partner‑led buildout of large, power‑secure AI campuses tied to its GB300/Vera Rubin roadmaps. (barrons.com)
  • AI impact: Adds significant U.S. training capacity, lowering costs and speeding progress in code‑generation and reasoning models. (barrons.com)
  1. Spectrum‑X Ethernet lands at Meta and Oracle
  • What’s new: Both are standardizing on Nvidia’s Spectrum‑X switches to link massive GPU clusters. (investor.nvidia.com)
  • Bigger vision: Owning the “nervous system” of the AI factory—GPUs plus networking—deepens Nvidia’s full‑stack moat. (investor.nvidia.com)
  • AI impact: Higher, steadier throughput shortens training times and reduces costs for trillion‑parameter‑class models. (investor.nvidia.com)
  1. China update: Nvidia’s advanced‑chip market share “now 0%”
  • What’s new: Huang reiterated that export curbs have effectively closed China to Nvidia’s top AI chips. (scmp.com)
  • Bigger vision: Nvidia is leaning into diversified regional builds and sovereign AI factories to offset the China gap. (scmp.com)
  • AI impact: Parallel tech stacks (China vs. RoW) will shape where models are trained and which platforms they target. (scmp.com)

— End of brief —