South Korea: The New AI Factory Hub!
Daily Nvidia Brief — October 31, 2025
- South Korea becomes an “AI factory” hotspot
- What’s new: Nvidia announced a slate of Korea deals at APEC — SK Group will build an AI factory with 50,000+ GPUs; separate programs with Samsung and Hyundai target smart manufacturing and mobility; and a broader Korea initiative adds national infrastructure and ecosystem support. Reuters pegs total Korean deployments at 260,000+ chips across government and industry. (nvidianews.nvidia.com)
- Bigger vision: This is Nvidia’s AI‑factory playbook exported: GPUs + networking + CUDA‑X software, co‑designed with local champions to seed “sovereign AI” capacity. (nvidianews.nvidia.com)
- AI impact: More regional compute means faster progress in robotics, digital twins, and auto AI — and a more diversified global supply of training and inference capacity. (nvidianews.nvidia.com)
- Amazon’s capex drumbeat lifts sentiment
- What’s new: Nvidia shares ticked up after reports that Amazon plans to keep piling money into AI infrastructure in 2025, reinforcing a multi‑year buying cycle for GPUs. (barrons.com)
- Bigger vision: Hyperscaler spending underwrites Nvidia’s full‑stack roadmap — keeping fabs, systems, networking, and software on an aggressive cadence. (barrons.com)
- AI impact: More cloud capacity brings cheaper, broader access to state‑of‑the‑art training and inference for startups and enterprises. (barrons.com)
- Quantum + GPU: NVQLink debuts
- What’s new: Nvidia introduced NVQLink, an open interconnect to tightly couple quantum processors with GPU supercomputers; 17 quantum builders and nine U.S. labs are on board. (nvidianews.nvidia.com)
- Bigger vision: Extends the “AI factory” into hybrid quantum‑classical systems — keeping Nvidia’s GPUs at the center as orchestration and control engines. (nvidianews.nvidia.com)
- AI impact: Better links between qubits and GPUs could speed breakthroughs in materials, chemistry, and error correction — useful for future AI model training and simulation. (nvidianews.nvidia.com)
- Robotaxis: Uber aligns on Hyperion 10
- What’s new: Nvidia and Uber plan to support a global robotaxi rollout targeting about 100,000 vehicles starting in 2027, built on the DRIVE AGX Hyperion 10 reference stack. (blogs.nvidia.com)
- Bigger vision: Brings “physical AI” into the streets — Nvidia supplies the compute, sensors, and software platform, expanding its ecosystem beyond data centers. (nvidianews.nvidia.com)
- AI impact: At scale, autonomous fleets could pressure test long‑horizon reasoning and real‑time perception models, pushing advances that flow back into general AI. (nvidianews.nvidia.com)
- Palantir tie‑up targets “operational AI”
- What’s new: Palantir will integrate Nvidia CUDA‑X, cuOpt, and open Nemotron models into its Ontology/AIP stack; Lowe’s is an early customer for AI‑optimized supply chains. (nvidianews.nvidia.com)
- Bigger vision: Deepens Nvidia’s software moat — not just selling GPUs, but standardizing how enterprises run decision‑making agents on its platform. (nvidianews.nvidia.com)
- AI impact: If packaged agents and workflows spread, expect faster adoption of AI in logistics, healthcare, finance, and the public sector. (nvidianews.nvidia.com)
- Market and policy watch
- What’s new: Nvidia touched a $5T market value this week even as U.S.–China chip rules remain a flashpoint; leaders met but did not resolve export limits. (reuters.com)
- Bigger vision: Nvidia’s growth now leans on diversified demand (U.S., Korea, others) and onshoring — insulating its AI‑factory roadmap from policy swings. (reuters.com)
- AI impact: Export rules will shape who trains the largest models — and where — influencing the global pace and distribution of AI capability. (reuters.com)
— End of brief —