Where Should You Invest in Physical AI? "Five Promising Themes Including Energy and Optical Communication"

"Companies That Excel at Using AI Will Be the Winners"
Traditional Manufacturers Also in the Spotlight

NH-Amundi Asset Management, which introduced Korea's first active ETF focused on Physical AI, shared its insights into the industry's current landscape after a year of tracking developments and presented five promising themes: energy, optical communication, semiconductors, AI models, and applications.

Choi Donggeun, Head of the ETF Operations Team at NH-Amundi Asset Management and manager of the HANARO Global Physical AI Active ETF, explained at the "Physical AI Investment Strategy Forum" held at the FKI Tower in Yeouido, Seoul, on the 22nd, that Physical AI is already expanding rapidly across advanced manufacturing processes, intelligent logistics systems, unmanned ports, and precision surgical robots. As a result, the operational methods of all industries, from agricultural infrastructure to healthcare, are undergoing a fundamental transformation. Based on this analysis, NH-Amundi Asset Management introduced five promising themes within the Physical AI value chain: energy, optical communication infrastructure, semiconductors, AI models, and applications.

Choi Donggeun, Head of the ETF Management Team at NH-Amundi Asset Management, is giving a presentation on Physical AI at the FKI Tower in Yeouido, Seoul on the 22nd. Photo by Kim Youngwon

First, the company analyzed that securing energy infrastructure itself has become a key AI competitiveness factor due to the enormous power demands of AI computing. Choi stated, "The most expensive part of implementing AI is electricity," adding, "Sam Altman, CEO of OpenAI, once mentioned that a simple 'thank you' from a ChatGPT user can cost hundreds of billions of won annually." He further explained that although memory prices are rising, their semi-permanent nature and amortization mean their share of actual costs decreases over time. GE Vernova was cited as a core company in this area.

Regarding optical communication infrastructure, Choi commented, "No matter how fast a sports car is, it's useless if the road is rough or narrow," emphasizing that optical communication infrastructure, which can solve various issues of copper-based networks, will become a driving force of the next supercycle. Among optical communication companies, Lumentum Holdings and Tower Semiconductor were highlighted.

In the semiconductor sector, Choi pointed out that supply shortages are occurring across all industry segments, including not only memory but also substrates and power semiconductors. In particular, NVIDIA's next-generation AI platform, 'Vera Rubin,' has adopted a high-voltage direct connection method for efficient power delivery, making power semiconductors an area to watch. Choi stated, "We are especially focused on Intel," noting, "With CEO Lip-Bu Tan recently taking the helm, the company is making continued efforts to regain the technological leadership it once lost."

In the field of AI models, Choi sees companies in Greater China as promising. "Chinese companies have a strong presence in the open-source community," he said, naming Zhihu AI and MiniMax Group. After listing on the Hong Kong Stock Exchange earlier this year, Zhihu AI and MiniMax Group saw their share prices rise by over 600% and 400%, respectively.

Choi cited applications as the most noteworthy theme. The advancement of AI is significantly boosting productivity across traditional industries. He explained that established manufacturing companies with cutting-edge manufacturing technologies in hydraulic and fuel control systems, bearings, and precision control systems are being reborn as essential hardware component suppliers in the era of Physical AI. He particularly expects strong growth from Parker Hannifin and RBC Bearings.

Lastly, Choi emphasized the importance of identifying companies that are not just "good at making AI" but are "good at using it," so-called "AI Hyper-Adopter" companies. He stated, "We should focus on the 'rebirth of traditional companies' that transform AI into a core business tool, thereby significantly improving profitability."

On the same day, Kwon Yongmin, Head of the ETF Product Research Team at NH-Amundi Asset Management, presented the one-year performance of the HANARO Global Physical AI Active ETF, which was listed in Korea for the first time last year as an ETF investing in the Physical AI theme. As of the 20th, approximately one year after its listing, the ETF recorded a return of over 127%. This figure exceeds the Nasdaq 100 index by 78 percentage points over the same period.

As the background for this performance, Kwon cited the company's ability to quickly capture narrative shifts through an active strategy. He explained, "We pursued balanced investments across the five promising themes, but last summer, our allocations to energy and semiconductors increased. More recently, the weight of applications—namely, traditional manufacturing companies—has also grown, allowing us to respond flexibly."

This content was produced with the assistance of AI translation services.