Goldman Sachs predicts capacitor stocks will surge as AI data centers devour components

The humble capacitor, a component most people haven’t thought about since high school physics, is having its moment. Goldman Sachs Research is telling investors to pay attention to the tiny electronic components that regulate voltage and store energy, because AI’s insatiable appetite for data center infrastructure is creating what the bank’s analysts describe as potentially the longest and largest upcycle in capacitor sector history.

The MLCC bet

At the center of Goldman’s call is the multi-layer ceramic capacitor, or MLCC. Goldman has revised its 2026 MLCC pricing forecast upward, moving from a flat outlook to a predicted increase of 0-5%. Goldman now projects the AI-driven MLCC upcycle will extend to roughly 2030, a notable stretch from prior expectations that pegged the cycle ending around 2028. The revision came after Goldman analysts met with Murata Manufacturing, one of the world’s dominant MLCC producers.

The companies in the crosshairs

Goldman has maintained a Buy rating on Murata Manufacturing with a target price of 5,400 yen. The bank has also rated Nantong Jianghai Capacitor Co., a Chinese manufacturer with direct exposure to the growing data center power infrastructure buildout, as Buy.

Data center electrical component stocks have already delivered an extraordinary run. The cohort has achieved an 85 percentage point absolute outperformance versus hyperscaler stocks since early 2025.

The power demand picture

Goldman’s latest projection anticipates 220% global power demand growth by 2030, a sharp revision from a previous estimate of 175%. The driver is accelerated capital expenditure by AI hyperscalers. Companies like Microsoft, Google, Amazon, and Meta have collectively committed hundreds of billions of dollars to building out AI infrastructure, and every new data center campus requires massive electrical systems filled with capacitors.

What this means for investors

The 85 percentage point outperformance versus hyperscalers since early 2025 raises the obvious question: has the easy money already been made? Goldman clearly thinks the answer is no, given that the upcycle extension to 2030 means several more years of elevated demand ahead.

The risk factors are worth considering. MLCC pricing increases of 0-5% are modest, and if the AI buildout slows or hyperscalers pull back on spending, the upcycle thesis evaporates quickly. Geopolitical tensions between the US and China also add uncertainty, particularly for Chinese component manufacturers like Nantong Jianghai, where tariff escalations or export controls could disrupt the supply chain dynamics that Goldman’s model depends on.