Wall Street's next AI trade is smaller than a grain of rice: One Big Idea

Wall Street investors are shifting focus to multilayer ceramic capacitors (MLCCs) as the next potential bottleneck in AI server production, following a sell-off in chip and memory stocks. Goldman Sachs projects the AI-server MLCC market could grow more than fourfold by 2030, driving rallies in Asian component makers like Taiyo Yuden and Murata before recent pullbacks.

The AI trade is getting its first real stress test since Wall Street discovered the bottleneck playbook. Chips are sliding https://finance.yahoo.com/markets/article/the-2-trillion-chip-sell-off-hits-a-make-or-break-level-chart-of-the-day-155000822.html , memory is cracking https://finance.yahoo.com/markets/article/micron-samsung-sk-hynix-just-dragged-memory-stocks-into-a-bear-market-154549356.html , and investors are already hunting for the next scarce part of the server rack. Memory chips showed how fast that playbook can work, as poster child Micron MU https://finance.yahoo.com/quote/MU/ surged nearly 1800% off the April 2025 lows. But the market is already moving on to the next question: What else inside an AI server cannot scale fast enough? The latest answer is the part of the AI trade that is smaller than a grain of rice. Multilayer ceramic capacitors https://m.samsungsem.com/global/product/passive-component/mlcc.do , or MLCCs, are tiny components that help smooth voltage, filter electrical noise, and keep chips from glitching when power demand surges. AI servers do not draw power in a calm, steady stream. GPUs and custom chips ramp up, pause, and ramp up again as they process huge calculations. That makes MLCCs a natural target for Wall Street's bottleneck hunt. A late-May Goldman Sachs thesis https://x.com/zerohedge/status/2060558847735939447 put capacitors forward as the next potential AI squeeze after memory. Goldman's work put MLCCs behind only GPUs and memory among the biggest cost items in AI server builds, and estimated the AI-server MLCC market could grow more than fourfold from fiscal 2025 to fiscal 2030. The scale explains the chase. A single advanced AI server rack can require hundreds of thousands of MLCCs. If every next-generation AI rack needs more of them and suppliers cannot add capacity quickly, a boring component can suddenly become a pricing-power trade. That is what investors have been chasing. Selected Asian MLCC and passive-component stocks tied to the AI server supply chain have surged off the March 30 lows. Taiyo Yuden 6976.T https://finance.yahoo.com/quote/6976.T/ , Walsin 2492.TW https://finance.yahoo.com/quote/2492.TW/ , Samsung Electro-Mechanics 009150.KS https://finance.yahoo.com/quote/009150.KS/ , Yageo 2327.TW https://finance.yahoo.com/quote/2327.TW/ , Murata 6981.T https://finance.yahoo.com/quote/6981.T/ , and TDK 6762.T https://finance.yahoo.com/quote/6762.T/ all rallied hard, with several more than tripling before the latest pullback. This embedded content is not available in your region. Now comes the harder part. Several of those names are down more than 20% from 52-week highs, and a few are off more than 30%. That puts them in the same AI hardware unwind hitting chips, memory, and the broader semiconductor complex. However, capacitor stocks are not moving in one clean wave — a sign this is still more an emerging theme than a fully formed basket trade. The next test is whether the strongest MLCC suppliers can stop falling, rebuild support, and keep seeing pricing power. Goldman added fuel in early July, with fresh upgrades across the MLCC and capacitor supply chain https://x.com/zerohedge/status/2073951678492021116 , including Yageo, Samsung Electro-Mechanics, and LG Innotek. Goldman pointed to AI server demand, rising component content, and tighter supply as drivers of higher earnings estimates.