Roundhill Investments launched a new ETF this week, ticker CCML, focused on multilayer ceramic capacitors (MLCCs) and electronic components. These sit on the semiconductor circuit board and rarely get attention. This follows earlier Roundhill funds that target single parts of the AI trade: the Memory ETF, the Photonics and Optics ETF, and the Neocloud ETF. CCML goes deeper and stays pure-play on the circuit board.
What MLCCs Do
MLCCs are smaller than a grain of rice. A smartphone holds thousands of them. A single circuit board holds tens of thousands. The Vera Rubin chip will hold even more. Each one releases small amounts of electrical charge, and together they control power flow. A GPU gets taxed at different times, and these parts deliver power with precision so the system is not overloaded. When you send a query to a large language model and want a fast, strong answer, the MLCCs inside the board help make that happen.
The parts first appeared in 1960 as part of the NASA space program, where delivering power precisely without overload was the core need. As chip design grows more complex and more powerful, each system needs more of these components to work.
The Investment Case
MLCCs ship in the trillions of units each year. They are one of the components with the highest price increases on the bill of goods for a server rack or an Nvidia chip, especially the high-end parts needed for AI. As AI needs more power, the buildout needs more of these circuits - a point stated with no doubt.
Access is the hard part for US investors. Like memory or photonics, only a few pure-play companies exist, and most are based outside the US. The makers include Murata in Japan and Samsung Electro-Mechanics in Korea. The vast majority of production happens in Asia: Korea, Japan, China, and Taiwan. Murata alone controls 35% of the market. There are no US companies here that investors have known about, because the makers were nearly untouchable for most US buyers. The research draws on Goldman Sachs work and Japan Ministry of Finance data.
CCML packages these Asian names together to give precise exposure, the same approach used for the DRAM ETF and others. MLCCs are becoming another bottleneck in the AI trade.
Where the AI Buildout Stands
Where are we in the AI buildout? Early to mid innings, using a baseball comparison. Investors debate this daily, and performance of the precision ETFs (memory, photonics, neoclouds) swings - rewarded some days this week, not others.
Heading into the midterms, expect some political backlash from both parties over data center buildouts, tied to worries about harm to the environment and to communities. Some projects may be put on hold or stretched out. The US leads in AI and wants to keep that lead, so the buildout should continue. The early benefits - making consumers and workers more productive - are starting to pay off.
Other Roundhill ETFs mentioned: DRM, NCD, and LIGHT.
