New ETF Targets Tiny Chips Behind AI Servers

Investors have spent the last two years chasing high-profile AI components like graphics processors and high-bandwidth memory. Now, attention is shifting to a much smaller part of the supply chain: multilayer ceramic capacitors, or MLCCs. These passive components are essential for stabilizing voltage in electronic devices, and they are becoming a focal point for traders looking for the next phase of the AI hardware boom.

On September 9, Roundhill launched a new ETF called the MLCC & Electronic Components ETF, trading under the ticker CCML. The fund offers concentrated exposure to global suppliers of these components. Its launch signals that MLCCs are evolving from a niche trade into a standalone investment theme, driven by the belief that AI servers require significantly more of these parts than traditional computers.

AI Servers Demand More Capacitors

As AI chips become more powerful, they draw more power and create sharper fluctuations in current. This requires more complex power-delivery systems that rely heavily on MLCCs to manage voltage stability. Estimates suggest that an AI server uses ten to fifteen times as many MLCCs as a standard server. For example, Nvidia’s recent hardware platforms reportedly require hundreds of thousands of these capacitors per rack.

Leading manufacturers are adjusting their forecasts to reflect this surge. Murata, a major Japanese supplier, has raised its estimates for the number of capacitors needed on a single AI server baseboard. The company now expects each board to contain between 15,000 and 25,000 units, up from previous lower estimates. This shift indicates that the opportunity lies not just in selling more servers, but in the higher value of components required inside each one.

Supply Tightness Drives Price Increases

The demand spike is colliding with limited production capacity. Producing smaller, higher-performance capacitors requires advanced materials and precise manufacturing techniques that are difficult to scale quickly. As a result, suppliers are seeing order books grow larger than their production output. By mid-June, major manufacturers like Murata and Samsung Electro-Mechanics reported book-to-bill ratios above 1.3, meaning they had 30% more orders than they could ship.

This imbalance is beginning to affect prices. Reports indicate that Samsung Electro-Mechanics has raised prices for specific high-grade MLCCs sold to computer manufacturers. The combination of rising volumes and tightening supply suggests the industry is entering an upcycle where both volume and pricing work in favor of producers.

Concentrated Portfolio Focuses On Leaders

The new CCML fund is heavily concentrated in a few key players. Its two largest holdings are Murata and Samsung Electro-Mechanics, which together make up nearly half of the portfolio. Other significant positions include Chaozhou Three-Circle, Yageo, and Taiyo Yuden. According to GN technics/hardware (en-US), this structure targets the companies best positioned to benefit from the premiumization of AI hardware components.

However, the trade-off is significant risk. Because the fund holds a small number of stocks, its performance is heavily dependent on the fortunes of these specific manufacturers. If demand for AI hardware slows down or if competitors expand capacity faster than expected, the fund could see sharp declines. Investors are betting on a sustained upcycle in component pricing, but this remains a concentrated wager on a specific niche of the tech supply chain.