Nvidia's chip supplier prepares for an IPO worth $29.4 billion

SK Hynix, South Korean manufacturer of memory chips and key supplier to Nvidia, announced that it will seek to raise up to US$29.4 billion with an initial public offering in the United States. The operation aims to attract investor interest in companies linked to artificial intelligence.

According to Reuters, the company plans to list on the Nasdaq after becoming the most valuable company in South Korea, surpassing Samsung Electronics. The operation would also take place ahead of the anticipated IPOs of Anthropic and OpenAI.

Why SK Hynix wants to list on the Nasdaq

The company believes that listing on the Nasdaq will allow the market to evaluate it with parameters similar to those of Micron, one of its main competitors in the United States.

This change could favor a higher valuation and also impact the price of its shares traded in South Korea.

If the operation is completed for the expected amount, it will be the second largest stock sale in the world. It would only be behind the public offering of SpaceX, which reached US$85.7 billion this month, and would surpass those made by Saudi Aramco in 2019 and Alibaba in 2014.

How SK Hynix surpassed Samsung

The memory manufacturer became the company with the highest market value in South Korea, displacing Samsung Electronics, which had held that position since 2000.

The change reflects the growth in global demand for components for artificial intelligence and the role that SK Hynix plays within the supply chain of Nvidia, Google, and other AI system developers.

The company has a valuation close to US$1.2 trillion and its shares have quadrupled so far this year, outperforming Samsung and Micron.

What chips does SK Hynix manufacture

SK Hynix produces DRAM, NAND Flash, and high bandwidth memory (HBM).

DRAM memory is used as main memory in computers, servers, and mobile devices. NAND Flash chips are intended for permanent storage in SSDs, phones, and memory cards.

HBM memory has become one of the most important components for artificial intelligence because it offers superior bandwidth, lower latency, and more energy-efficient consumption than traditional memory.

Why HBM memory is key for artificial intelligence

The HBM architecture allows for processing large volumes of data in parallel, a necessary feature for training and inference of artificial intelligence models.

Its three-dimensional stacked design physically brings the memory closer to the processor, reducing latency and improving the performance of GPUs, specialized accelerators, and other processors used in data centers.

The company continues to invest in research and development of new generations of HBM and DRAM to meet the growing demand for infrastructure aimed at artificial intelligence and high-performance computing.