Nvidia commits $500 billion to SK Group as the global AI buildout accelerates

Jensen Huang, the CEO of Nvidia, has revealed a partnership deal worth around $500 billion with SK Group of South Korea which may become the largest commitment in the field of AI infrastructure so far. Huang announced the amount of deal during the meeting with President of South Korea Lee Jae Myung in San Francisco, although the companies have not yet publicly disclosed details about the deal.

The comments reported first by the Yonhap news agency have underlined the speed with which the AI race is developing. Starting off as the competition to make faster chips, the race has moved on to establishing every aspect of AI supply chain that requires advanced memory, data centers and engineers to power them.

According to ChosunBiz, Huang said during the meeting, “SK Group and Nvidia will announce a business partnership between the two companies today… the size amounts to $500 billion.”

Why a memory partner sits at the center of the deal

The South Korean conglomerate SK Group owns SK Hynix, which is one of two companies that monopolize the region of high-bandwidth memory (HBM), which is stacked DRAM that allows AI accelerators to perform large data processing at a very high speed. That will put SK Group in a position closer to Nvidia and makes memory supply a key ingredient in their collaboration rather than only being about chip design.

This is significant because HBM is now one of the most significant limitation in this sector, and different suppliers are adopting various methods for the upcoming generation. TrendForce reported in the month of June that SK Hynix has decided to postpone the launch of its HBM4 products and focus on maximizing profit from its commodity DRAM. For Samsung, on the other hand, HBM4 sales have already exceeded $1 billion. For Nvidia, forming a direct partnership with SK Group guarantees that it has a greater certainty with regard to one of the major components for AI hardware.

The partnership also falls within the broader strategy of Nvidia. The company has moved beyond GPUs into rack-scale AI systems as well as custom AI CPUs, making it even more crucial to have reliable access to advanced memory as customers start deploying bigger AI clusters.

What Nvidia is planting inside South Korea

The intended SK Group collaboration is merely one phase of Nvidia’s larger project in South Korea. According to Chosun Ilbo, Huang stated that the company plans to invest in Naver, work together with Hyundai on a self-driving Genesis car, transfer several top AI researchers from California to South Korea, and cooperate with KAIST on the creation of a Korean large language model in his meeting with Lee.

Lee’s schedule mirrored South Korea’s broader ambition to be a worldwide AI center. On that same day, he also met with leaders of OpenAI, Anthropic, and Broadcom, as Seoul targeted an influx of capital from top AI companies.

“This is a golden age for South Korea,” Huang stated, citing Nvidia’s long-term cooperation agreements with Samsung Electronics and SK Hynix.

As per Yonhap, Lee noted Nvidia’s partnership with Hyundai, adding that “the expectations of young people in Korea and residents of the North Jeolla region are high.”

The larger conclusion to be drawn is not limited to a single cooperation. AI companies are beginning to position their research, talent, and computing infrastructure close to the associated supply chains. The development of Nvidia’s regional strategy, as well as its response to the increasing demand for advanced AI chips, gives an understanding of how Asia is becoming an important part of the company’s long-term growth plans.

The demand curve the spending is chasing

The announcement comes at a time analysts have started to uplift their forecasts about investments in AI technology. As per the SemiAnalysis report released on June 30, revenues associated with Nvidia’s compute systems business will exceed Wall Street estimates by approximately 20% in the second half of fiscal 2027, benefiting from better results of the company’s Rubin platform implementation after earlier problems related to supply of HBM4.

The forecasts for the industry overall are also aligned with this. According to Omdia, global investments in data centers are projected to be close to $1.6 trillion by 2030, while tech giants worldwide are likely to spend $600 billion on AI infrastructure in 2026 alone.

Given this context, the intended collaboration between Nvidia and SK Group seems less a standalone story and more of another step in the race for investments in artificial intelligence, which is now taking place on a scale of hundreds of billions of dollars rather than millions.

Don’t just read crypto news. Understand it. Subscribe to our newsletter. It's free.