Many investors began questioning earlier this year whether stock markets had become overly reliant on a small group of artificial intelligence winners. Franklin Templeton's global index portfolio management lead, Ding Na, and ETF investment strategy head, Marcus Weyerer, believe the benefits of the AI theme could extend far more broadly than the market commonly expects.

The wave of AI investment is gradually shifting from companies that develop AI technology toward infrastructure providers that support real-world AI applications. However, significantly boosting AI productivity still requires substantial capital spending across semiconductors, memory, power, data centers, and digital infrastructure. In other words, various bottlenecks persist until AI resources become abundant.

As capital expenditure rapidly flows into limited areas, the breadth of investment opportunities may far exceed what many investors anticipate. In the first half of 2026, the S&P 500 Index delivered roughly 10% returns, but the more noteworthy trend is the shift in leading stocks. Several international markets, including South Korea, Taiwan (China), Japan, and Brazil, have outperformed the S&P 500, though for different reasons.

Where the Next Opportunities Lie

In the second half of the year, market attention may no longer focus on whether AI remains a viable investment theme, but rather on who will become the next beneficiary. Investors' focus has moved from large cloud service providers building AI applications toward companies supplying related infrastructure and hardware. This trend reflects the maturation of the AI investment theme rather than its decline.

Investor positioning also reflects this change. By early July, ETFs focused on the South Korean and Taiwanese markets recorded substantial net capital inflows, with assets under management growing more than 240% and nearly 20% respectively from the start of the year. This indicates that market interest in AI infrastructure and hardware investments has surpassed that in AI model development.

Comparing the AI Boom to the Dot-Com Era

Of course, whether the current AI frenzy will repeat the late-1990s tech bubble remains a common market question, but the two situations are not identical. While some speculative activity exists in certain market areas, today's AI giants generally demonstrate more stable profitability, more mature business models, and valuations that are often more reasonable than those of leading companies during the tech bubble era.

In terms of timeline, this rally is significantly smaller in both duration and scale compared to the same stage of the internet bubble. Many companies that were highly sought after during that period lacked profitability, cash flow, and sustainable competitive advantages. In contrast, many leading AI companies operate across multiple aspects of the entire AI ecosystem, covering chip design, cloud infrastructure, software platforms, and proprietary data, which is expected to sustain their pricing power and profit margins over the long term.

Additionally, AI infrastructure has become a capital-intensive industry, giving it deeper economic moats and higher barriers to entry compared to previous technology cycles.