Korea Investment's ACE Global AI Custom Semiconductor ETF surpasses 50 billion won in net assets

Korea Investment Trust Management announced Tuesday that its ACE Global AI Custom Semiconductor exchange-traded fund has surpassed 50 billion won ($36.2 million) in net assets.

According to Korea Exchange, the ETF's net assets stood at 50.4 billion won as of Monday, up 406.52 percent from the start of the year.

The fund posted a three-month return of 122.31 percent, ranking second among 23 overseas semiconductor ETFs listed in Korea, excluding leveraged products. Its six-month and one-year returns were 132.20 percent and 174.65 percent, respectively.

The ETF selects industries representative of custom semiconductors, or ASICs, and concentrates its holdings in a small number of core companies. The strong performance was driven largely by share price gains at Arm Holdings and Marvell Technology Group, both leaders in semiconductor intellectual property and custom chip design.

Arm, the ETF's largest holding at 29.88 percent as of Monday, saw its share price rise 208.06 percent over the past three months. The gain reflects growing demand for CPUs capable of reducing inference workload costs, as the spread of AI agents shifts computing demand from training to inference.

Arm holds a commanding position in power-efficient IP. The company has been expanding its supply of proprietary CPU chips to major global technology firms, broadening its business model from IP licensing to chip supply and prompting a revaluation of its corporate worth.

Marvell, the second-largest holding at 23.14 percent as of Monday, rose 249.92 percent over the past three months, contributing significantly to the fund's returns. The company holds roughly a 20 to 25 percent share of the global custom ASIC design services market. Its share price was boosted by accelerating sales of custom ASICs to Amazon Web Services and news that Nvidia had agreed to invest approximately $2 billion in Marvell's optical networking business.

Marvell drew particular market attention on June 2 when Nvidia CEO Jensen Huang directly referred to the company as a potential "next trillion-dollar company" at Computex 2026 in Taiwan. Analysts say Marvell's strategic value is gaining recognition as AI infrastructure competition expands beyond GPUs to encompass networking, optical communications, ASICs and data centers more broadly.

As of Monday, the ACE Global AI Custom Semiconductor ETF holds the highest combined weighting in Arm and Marvell of any ETF listed in Korea, according to ETF CHECK data.

Unlike many global semiconductor ETFs listed in Korea that carry heavy Nvidia weightings, this fund includes Broadcom, Cadence Design Systems and Synopsys, positioning the portfolio to capture growth potential among custom AI chipmakers.

Nam Yong-su, head of the ETF division at Korea Investment Trust Management, said that as the proliferation of AI agents makes power efficiency and cost optimization increasingly important, inference CPUs and ASICs are emerging as core investment areas within AI infrastructure. He added that Arm's architecture is extending its influence across the full AI computing stack — from PCs to data centers — as demonstrated by Nvidia's unveiling of RTX Spark, an AI PC platform powered by an Arm-based CPU, at Computex 2026.

"The ACE Global AI Custom Semiconductor ETF has delivered strong recent returns by maintaining high weightings in companies leading the CPU and ASIC industries," Nam said. "Investors who combine it with our ACE Global Semiconductor TOP4 Plus ETF, which tracks the broader general-purpose semiconductor market, will be well positioned to benefit across the full spectrum of AI industry growth."

Korea Investment Trust Management has also been active in launching new products. On Tuesday, the firm listed two new ETFs: the ACE K Semiconductor TOP2+ ETF, which concentrates investment in two of Korea's leading companies in the global AI semiconductor market, and the ACE Korea AI Power TOP10 ETF, which targets domestic companies expected to benefit from surging power demand and grid infrastructure buildout in the AI era.