Our website is made possible by displaying non-intrusive online advertisements to our visitors.
Please consider supporting us by disabling or pausing your ad blocker.
The application range for silicon capacitors is also expanding rapidly. They were initially used mainly in mobile devices, but demand has recently surged in AI data centers. Industry watchers expect their use to expand further in physical AI fields such as humanoid robots, self-driving cars, and industrial robots. As AI-based control becomes more common, ultra-fast computing and high reliability are becoming more important.
Kim Won-gi, head of the Silicon Capacitor Development Group at Samsung Electro-Mechanics, said, "The teams responsible for semiconductor substrates, MLCCs, and silicon capacitors operate as one team," adding, "We are proposing the optimal combination of components based on the voltage, capacitance, and substrate structure that customers want."
Samsung Electro-Mechanics' biggest strength is that it can provide an integrated solution rather than individual components. AI semiconductor packages use FC-BGA package substrates, MLCCs, and silicon capacitors together. Samsung Electro-Mechanics is one of the few companies that can supply all of them.
Their strengths are even more pronounced in AI semiconductors. Because they are made on silicon wafers, they can be made much thinner and mounted directly inside semiconductor packages. They also incur less power loss than MLCCs, allowing them to respond quickly to power fluctuations. They also effectively reduce signal interference generated during high-speed computing.
Silicon capacitors work in a similar way to MLCCs, but their structure, materials, manufacturing process, and applications are different. While MLCCs are strong at storing power by stacking ceramic materials in multiple layers, silicon capacitors are specialized in achieving an ultra-thin structure and excellent high-frequency characteristics by using semiconductor processes.
As the AI semiconductor era gathers pace, the importance of silicon capacitors is rising quickly. The latest AI servers are packed with Graphics Processing Units (GPUs) and High Bandwidth Memory (HBM), which leads to massive instantaneous power consumption and frequent minor voltage fluctuations. If power supply is unstable, it can cause calculation errors or lower performance. Stable power delivery technology is therefore essential for silicon capacitors.
At a recent media seminar on the 14th, Samsung Electro-Mechanics said it is building a full-scale mass production system for silicon capacitors and expanding its push into the global AI semiconductor market. A silicon capacitor is a component used in AI semiconductor packaging.
Last month, Samsung Electro-Mechanics announced that it would begin supplying silicon capacitors to a global company from January next year. The contract is worth 1.5 trillion won.