Jensen Huang Meets Seoul’s Builders as NVIDIA Pushes to Align the AI Supply Chain for the Second Half

NVIDIA founder and CEO Jensen Huang is in Seoul this week, meeting partners and builders behind South Korea’s fast-growing AI and robotics push. In an update posted by NVIDIA, Huang positioned the trip as a time to “align the AI supply chain” ahead of what he described as a demanding second half - after a strong start to the year marked by ongoing infrastructure buildouts.

Speaking to media during the visit, Huang said NVIDIA is already seeing momentum from its current-generation systems and that new production capacity will keep the company and its ecosystem busy. He pointed to Grace Blackwell, describing it as “doing very well,” and to Vera Rubin, which he said is now “in full production.”

Grace Blackwell and Vera Rubin refer to NVIDIA’s datacenter system platforms that combine GPUs (graphics processing units), CPUs, and high-speed interconnects to run large-scale AI training and inference. Vera Rubin, in particular, is designed as an “AI factory” engine for agentic workloads - AI systems that can carry out multi-step tasks with reasoning, retrieval, and tool use - according to NVIDIA’s investor materials. NVIDIA said Vera Rubin is ramping into full production and that it delivers “10x agent throughput” at scale versus the Grace Blackwell platform.

Huang also emphasized robotics and physical AI - efforts to apply AI to the physical world, such as manufacturing automation and other real-world deployments. In comments carried in the NVIDIA blog update, he called robotics “the next major sector here in Korea,” arguing that it is a strong opportunity for Korea to invest in AI and build capabilities beyond software and into deployed systems.

That framing aligns with a broader theme in NVIDIA’s recent coverage of South Korea: building AI infrastructure that can support not only research and cloud services, but also industrial transformation. In a separate NVIDIA newsroom release, the company said South Korea and NVIDIA are working with government and industrial partners to expand sovereign AI infrastructure and AI factories, including a deployment plan described as “over a quarter-million” NVIDIA GPUs across sovereign clouds and AI factories, with additional details on initiatives involving national and enterprise providers.

In the same release, NVIDIA described specific types of infrastructure investment across Korea’s industrial base: the government’s plan to deploy tens of thousands of GPUs through national compute initiatives and cloud partners; and large-scale “AI factory” builds tied to sectors such as semiconductors, manufacturing, telecommunications, and automotive. NVIDIA also described work with industry and research institutions on next-generation communications infrastructure that supports network-edge intelligence - an important ingredient for deploying AI in physical environments where latency, power use, and connectivity matter.

Even with the high-level emphasis, NVIDIA did not provide a detailed list of which Seoul partners Huang met on this trip, nor did it disclose any new, named procurement contracts or exact shipment volumes tied to the “second half” timeline referenced in the blog. The update also does not break out which specific components of the AI supply chain - such as server builds, networking, storage, or software layers - will be the focus of the alignment effort in Korea.

What to watch next is how broadly Vera Rubin’s production ramp translates into deployed capacity across Korean sovereign clouds and industrial AI programs. If NVIDIA’s stated ramp pace continues, the second half could bring more visible scaling of AI factory-style deployments aimed at agentic workloads, alongside more robotics and physical AI experimentation as Korea’s ecosystem tries to convert sovereign infrastructure into real-world productivity gains. In the near term, expect follow-on announcements from Korean cloud providers and system integrators as they translate NVIDIA’s hardware readiness into operational deployments.