SK Hynix HBM4 Mass Production Drives 557% Profit Surge

SK Hynix delivered record-breaking financial results for the second quarter of 2024, driven by surging demand for high-bandwidth memory chips used in artificial intelligence infrastructure. The company reported an operating profit of 60.54 trillion KRW (around $42 billion), which represents a 557% increase compared to the same period last year. This massive profit growth highlights the critical role of memory components in the current AI hardware boom, directly impacting the supply chain for major server and GPU manufacturers.

SK Hynix ramps HBM4 output as AI memory demand accelerates

The financial surge coincides with the mass production of HBM4, the company's latest memory technology. SK Hynix officially began shipping HBM4 chips during the second quarter and plans to expand production capacity in the second half of the year. Additionally, the company delivered samples of the next-generation HBM4E to customers in the first half of 2024, ensuring a pipeline of advanced memory solutions for upcoming hardware generations.

SK Hynix secured long-term supply agreements with approximately ten customers to stabilize pricing and guarantee volume for these new memory types. The company's total revenue for the quarter reached 79.32 trillion KRW (around $54.3 billion). While these figures represent record highs, they fell slightly short of market expectations, which had anticipated revenue closer to 84 trillion KRW.

We looked at HBM4 closely in our earlier coverage on SK Hynix HBM4 Mass Production Starts to understand the technical implications of these 12-layer chips for Nvidia's upcoming Vera Rubin architecture. The company's ability to scale production and lock in customer contracts suggests a strong market position despite missing revenue targets. This production ramp-up is essential for meeting the growing computational demands of data centers worldwide.

SK Hynix now focuses on scaling HBM4 output to meet the sustained demand from AI infrastructure providers. The successful delivery of HBM4E samples indicates that the company is maintaining its technological lead in high-bandwidth memory. These developments confirm that HBM memory remains a bottleneck and a key differentiator in the AI hardware market.