LG Group is reportedly making a massive KRW 700 billion (about $456 million) investment to acquire 10,000 of NVIDIA's latest Blackwell GPUs.
This significant move appears to be catalyzed by several converging factors, with the timing being particularly noteworthy. NVIDIA CEO Jensen Huang's recent visit to Taipei and planned trip to Korea, where he met with leaders from major tech firms including LG, likely accelerated the decision. Such high-level engagement often helps secure priority allocations for highly sought-after hardware, making it a crucial moment for Korean companies to finalize their AI infrastructure plans.
From a supply chain perspective, the decision is also well-timed. First, NVIDIA has indicated that production of its Blackwell systems is ramping up, increasing confidence that large orders can be fulfilled in the second half of 2026. However, a sense of urgency remains. Wait times for these advanced systems are still long, often stretching from six to nine months. This scarcity creates a powerful incentive for companies like LG to place their orders now to secure a spot in the production queue and avoid falling further behind.
Internally, this investment is a cornerstone of LG's clear and ambitious AI strategy. The primary driver is the need for immense computational power to train and refine its proprietary large-scale AI model, EXAONE. To support this, LG has already announced plans to build one of Korea's largest AI data centers. This shows that the company is not just buying hardware, but is holistically preparing the necessary power and cooling infrastructure to operate a supercomputing-class AI cluster.
Furthermore, the broader environment in South Korea is highly supportive of such an investment. The government has laid a strong foundation with the new AI Basic Act, which provides a clear legal and promotional framework for AI development. This is complemented by a significant increase in the national budget for AI, creating a powerful tailwind of public support that reduces the risks associated with large-scale private capital expenditure.
In essence, LG's purchase is not a singular event but the culmination of strategic internal planning, opportune market timing, and a favorable national policy environment. It represents a determined step to build sovereign AI capabilities and solidify its position as a key player in the global AI landscape.
- Glossary
- GPU (Graphics Processing Unit): A specialized electronic circuit designed to rapidly manipulate and alter memory to accelerate the creation of images in a frame buffer intended for output to a display device. They are essential for training large AI models.
- GB200 NVL72: A high-performance computing system by NVIDIA that integrates 72 Blackwell GPUs into a single, liquid-cooled rack. It is designed for the most demanding AI training and inference workloads.
- EXAONE: LG's proprietary multimodal AI foundation model, capable of understanding and generating text, images, and other forms of data.