Samsung Electro-Mechanics is rapidly transforming from a smartphone-centric electronic components company to a key supplier of components for artificial intelligence (AI) data centers. In the past five months, the company has secured long-term supply contracts worth 4.3 trillion won, expanding its presence in the AI server component market.
This shift is driven by the competitive race among global tech giants to build AI data centers, which has intensified the battle for securing essential components beyond semiconductors. Samsung Electro-Mechanics is increasing its supply of high-value products such as multilayer ceramic capacitors (MLCCs) and semiconductor package substrates, accelerating its transition away from dependence on the smartphone market.
According to industry sources, Samsung Electro-Mechanics signed a long-term supply contract worth 360.5 billion won for memory semiconductor package substrates with a major global corporation on October 20. The contract will run from January to December next year. This brings the total number of large supply contracts signed since May to seven, with a cumulative contract value of approximately 4.3 trillion won.
The range of items being ordered is also diversifying. Starting with a 1.557 trillion won contract for silicon capacitors in May, the company has secured consecutive contracts for AI server MLCCs in June and July. Last month, it signed a contract worth 1.0721 trillion won for MLCCs, and this latest agreement expands its long-term contract portfolio to include memory semiconductor package substrates.
The surge in orders is attributed to the explosive growth of AI data centers. AI servers require more than ten times the number of MLCCs compared to standard servers to reliably supply power to high-performance graphics processing units (GPUs) and memory semiconductors. AI server MLCCs must operate for extended periods in high-temperature and high-voltage environments, making the technology complex and limiting the number of suppliers.
Similarly, the importance of semiconductor package substrates is increasing as the computational performance of AI semiconductors rises and memory capacity expands, necessitating substrates that can transmit electrical signals at high speeds. Notably, this recent contract includes not only high-performance AI accelerator substrates but also memory substrates, highlighting a trend identified by iM Securities that the substrate supply shortage is affecting even lower-margin memory products.
The transaction methods have also changed. Previously, MLCC orders often fluctuated based on demand from finished products like smartphones and PCs. However, as the supply of high-performance components for AI data centers tightens, customers are increasingly entering long-term contracts to secure future supplies. Samsung Electro-Mechanics has already signed long-term supply agreements for MLCCs with over ten major global tech and semiconductor companies.
The structural changes in the business are reflected in its financial performance. In the second quarter of this year, the revenue of Samsung Electro-Mechanics' components division reached 1.6494 trillion won, a 29% increase from the same period last year. The package solution division, which produces substrates for AI accelerators and servers, grew by 37%. In contrast, the optical solution division, which manufactures camera modules for smartphones, saw only a 10% increase in revenue. High-value components for AI infrastructure are emerging as a new growth driver.
Samsung Electro-Mechanics is also investing heavily to expand its production capacity. Last month, it announced plans to invest a total of 6.78 trillion won in its Sejong plant and its production facility in Vietnam. This decision aims to significantly boost production capacity for FC-BGA, a high-value semiconductor substrate for AI servers. The Sejong plant is expected to begin operations at new production facilities in 2028.
The series of long-term contracts is likely to change Samsung Electro-Mechanics' revenue structure as well. By securing supply volumes in advance for next year, the predictability of its revenue has increased. The growing proportion of high-value products and enhanced pricing power due to supply shortages are also expected to improve profitability. However, if investment in AI data centers slows down, the burden of large-scale expansions could increase.
An industry insider noted, "High-value products like MLCCs and FC-BGA for AI servers require customer certification, yield, and production technology, making it difficult to narrow the technology gap in the short term."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.