Jensen Huang spent four days in Seoul and left with partnerships at nearly every stop. Most coverage focused on the SK Hynix memory deal and the Doosan robotics announcement. The LG partnership got three sentences. It deserved more.
On June 8, 2026, after meeting LG Group Chairman Koo Kwang-mo in Seoul, Huang told reporters that NVIDIA is collaborating with LG on motor technology and mechanical systems for humanoid robotics, and separately on architecting what he called "the future data centers." Two workstreams. One meeting. Neither one has a timeline or a price tag attached to it yet.
What Jensen Huang said after meeting LG's chairman
Huang's exact words, as reported by Reuters: "We are working with them in motor technology as well as mechanical systems so that we can bring together humanoid robotics and the future of robotics." He added: "We're also working with LG in architecting the future data centers." For context on what NVIDIA's physical AI stack actually consists of, see NVIDIA's Physical AI Stack: Cosmos 3, Vera, and the Robot Operating Layer.
Those are two distinct commitments, not variations of one idea. Motor technology and mechanical systems on the robotics side. Data center architecture on the infrastructure side. For a single corporate meeting, that is a wide scope.
NVIDIA's official blog framed the collaboration more specifically: LG's production technology data and manufacturing know-how from its global factory sites will feed into AI-driven manufacturing and digital twin work. The goal, per NVIDIA, is an autonomous manufacturing ecosystem where the full production chain, from raw materials through logistics and delivery, is connected in real time through data and AI. NVIDIA describes this as a "new global smart factory standard" the two companies intend to establish together.
Two workstreams: humanoid robotics and data centers
The robotics workstream is more concrete in description, if not in deliverables. Huang named motor technology and mechanical systems specifically. These are the lower-level, hardware-intensive parts of humanoid robot development, not the AI brain side that NVIDIA is already known for, but the physical actuation and movement systems. NVIDIA's Isaac platform is typically what sits between the hardware and the AI layer in these setups; for a detailed look at how that works in practice, see NVIDIA Picked Unitree for Its Humanoid Robot Platform. LG's relevant capabilities here span consumer electronics, home robotics, and mobility components across the LG Group portfolio. Which specific LG subsidiary leads this work, whether LG Electronics, LG Innotek, or another entity, has not been clarified by either party.
The data center workstream sits on a different axis entirely. "Architecting the future data centers" is Huang's phrase. NVIDIA's blog expands this slightly, pointing to LG's capabilities in smart spaces and data center technologies. LG CNS, the group's IT services subsidiary, has been active in data center construction and management in South Korea. The combination of NVIDIA's full-stack AI factory platform with LG's existing data center operations is likely what Huang was describing. No build timeline, no capacity figures, and no location has been announced.
Separately, LG AI Research was already part of South Korea's MSIT Sovereign AI Foundation Models project, announced at the APEC Summit in late 2025. That initiative involves NVIDIA, LG, NAVER Cloud, SK Telecom, and several others developing sovereign AI models using NVIDIA's NeMo framework. The June 8 partnership builds on an existing relationship, not a cold start.
Where LG fits in NVIDIA's larger Seoul offensive
The LG deal is one of at least five significant partnerships NVIDIA announced or formalized during Huang's June 6-8 Seoul visit. The others: a multi-year memory partnership with SK Hynix covering next-generation memory for AI data centers; an AI cloud and factory roadmap with SK Telecom; AI data center and robotics collaboration with Doosan Group; and an AI factory agreement with Naver.
The pattern across all five is the same. NVIDIA is no longer treating South Korea primarily as a supplier of chips and components. It is treating Korean conglomerates as platform partners, embedding its full-stack AI infrastructure, from GPU clusters to simulation software to physical AI tools, inside companies that already have the industrial footprint to deploy it at scale.
LG fits that template particularly well. It has global manufacturing sites that generate production data at scale. It has a consumer electronics division with experience in home robotics and smart appliances. It has a data center operation through LG CNS. Huang said on arrival in Seoul that robotics is "the next major sector" for Korea. LG is one of the most natural partners to make that claim concrete.
NVIDIA also noted that Vera Rubin, its next-generation GPU system, is now in full production. Huang said the second half of 2026 will be "very busy." The Korean partnerships appear designed to ensure the demand infrastructure is ready to absorb that supply.
What's still unconfirmed
Quite a bit. No specific robot model or product line has been named. No data center has been announced, located, or sized. Neither company has disclosed deal value. The division of work between LG subsidiaries is unclear. And the timeline for any of this moving from collaboration to deployed product has not been stated.
On the broader Seoul visit: South Korean media reported that Doosan Electronic Materials is expected to exclusively supply CCL materials for NVIDIA's Rubin chips, with related sales projected to reach KRW 1 trillion by 2026. That figure has not been confirmed officially by NVIDIA or Doosan. It should be treated as a reported projection, not a stated commitment.
My Take
The LG partnership is actually the most interesting of the Seoul announcements, and it got the least attention. The SK Hynix deal is important but expected — NVIDIA needs memory, SK Hynix makes memory, the logic is obvious. The LG deal is structurally different.
Motor technology and mechanical systems are not NVIDIA's core business. NVIDIA does not manufacture physical actuators. That is LG's contribution. What NVIDIA brings is the simulation environment, the Isaac platform, and the AI compute that turns hardware components into a system that can learn to move. The fact that Huang named those specific technical areas, rather than vague phrases about AI collaboration, suggests there is an actual engineering workstream underway, not just a handshake.
The data center piece is less surprising but worth watching for a different reason. If LG CNS becomes a significant NVIDIA data center partner in South Korea, it creates a second major Korean hyperscaler alongside NAVER. That changes the domestic AI infrastructure picture considerably.
- NVIDIA and LG Group confirmed a partnership covering humanoid robot motor systems and mechanical systems, plus data center architecture — announced after Jensen Huang met LG Chairman Koo Kwang-mo in Seoul on June 8, 2026.
- NVIDIA's official blog adds that LG's global manufacturing data and production know-how will feed into AI-driven digital twin and smart factory work.
- The LG deal is one of five major Korean partnerships from Huang's June 6-8 Seoul visit, alongside deals with SK Hynix, SK Telecom, Naver, and Doosan Group.
- No robot model, data center timeline, or deal value has been disclosed by either company.
- LG AI Research was already part of NVIDIA's South Korea sovereign AI project from the APEC Summit 2025 — this is a deepening of an existing relationship.
FAQ
What exactly is NVIDIA and LG partnering on?
Two separate areas. First, motor technology and mechanical systems for humanoid robotics, where LG's hardware manufacturing capabilities combine with NVIDIA's AI and simulation platforms. Second, data center architecture, where NVIDIA's AI infrastructure stack is being applied to LG's data center operations. NVIDIA's blog also describes a broader autonomous manufacturing ecosystem goal, using LG's global production data alongside NVIDIA's digital twin tools.
Which LG company is involved — LG Electronics or LG Group?
The announcement names LG Group, the South Korean holding company, not a specific subsidiary. LG Group's relevant subsidiaries include LG Electronics (consumer electronics, home robotics), LG CNS (data centers and IT services), LG Innotek (components), and LG AI Research (AI models and research). Which entity leads which workstream has not been clarified by either company.
How does this LG deal differ from NVIDIA's other South Korea announcements?
The SK Hynix deal is about memory chip supply for AI data centers. The Doosan deal covers industrial robotics, energy solutions, and chip materials. The Naver deal is about AI cloud and data center builds. The LG deal is specifically about humanoid robot hardware and data center architecture, combining LG's manufacturing and electronics base with NVIDIA's physical AI platform. It is the only Seoul deal that names motor technology and mechanical systems explicitly.
Has NVIDIA worked with LG on robotics before?
Yes. At CES 2026 in January, NVIDIA listed LG Electronics as one of the partners already using its physical AI and robotics technologies. LG AI Research is also part of the MSIT Sovereign AI Foundation Models project in South Korea, which involves NVIDIA's NeMo framework. The June 2026 announcement deepens and formalizes an existing technical relationship.
Conclusion
The wire story was three sentences. What it described was two distinct engineering commitments between one of the world's largest chip companies and one of South Korea's most diversified industrial conglomerates. One of those commitments, motor technology and mechanical systems for humanoid robots, points to a part of the robotics stack that NVIDIA has not previously occupied directly. The other, data center architecture, extends NVIDIA's footprint into LG's infrastructure business. Neither has a timeline. Both are worth tracking.
The question is whether a conglomerate with LG's breadth, spanning appliances, chips, telecom components, data centers, and manufacturing, can focus these commitments into something that ships. Large industrial partnerships between large companies tend to move slowly. Humanoid robotics tends to move slower than the announcements suggest — and real-world deployments so far show exactly that gap.
0 Comments