On May 21, the Hang Seng Tech Index closed down 2.15%, with an annual cumulative decline of nearly 15%. Under the dark clouds of the market, Fang Jincong, head of China internet industry research at UBS, stated at a media briefing that although the sector faces short-term pressure, the long-term investment logic of the internet industry has undergone fundamental changes.

1. Three Pressures Behind the Decline and the Resilience of Fundamentals

Fang Jincong attributed the current index downturn to three overlapping pressures: weak macroeconomic conditions have put pressure on To C businesses (such as e-commerce, advertising, and local life services); increased capital expenditures on AI have eroded short-term profits; and the uncertainty in the AI transformation path has led to hesitation.

However, the earnings season showed strong resilience. Internet giants are continuously optimizing profit margins through "cost reduction and efficiency improvement," with significant narrowing of core business losses and a focus on revenue-generating channels such as API licensing, AI advertising, and game scenario empowerment. Fang Jincong believes that the market has already largely priced in negative factors, indicating that the stock price is hitting the bottom of the cycle, as reflected by "low expectations, low valuations, and a bottoming-out fundamentals."

2. The Diverging Investment Paths for Excess Returns: Large Models vs. Games

Regarding potential excess return sectors in the future, UBS analysts provided different path choices:

Large model companies (growth logic): Considered the most explosive sector, it may achieve a scale growth of 10-30 times. This is a classic "growth investment" aimed at capturing high returns by seizing cutting-edge sectors.

Game sector (certainty logic): Suitable for investors who prefer low risk. Although the expected growth rate is around double digits, the current P/E ratios of top game companies, only 12-13 times, offer a very high margin of safety and probability of excess returns.

3. Essential Differences in Valuation Logic

There is a significant gap in valuation methods between AI model providers and traditional internet companies:

AI start-ups (P/ARR method): Valued using Price/ARR (Annual Recurring Revenue), with premiums derived from technological leadership, scarcity of the asset, and liquidity premium during the initial listing period. For example, MiniMax and Zhipu's Hong Kong stock performance has already reflected the market's recognition of rapid growth.

Internet giants (profitability method): Still primarily focused on the profitability of traditional businesses. Due to the "heavy asset" nature of AI (high computing power investment) and the difficulty in isolating its contribution, AI business is difficult to compare with lightweight model start-ups in terms of valuation multiples.

4. The Core Philosophy of the AI Era: From "Selling Time" to "Selling Value"

Fang Jincong pointed out that this round of AI waves differs fundamentally from the mobile internet era:

Mobile Internet: The logic is monetizing traffic, with the ceiling limited by users' online time (about six or seven hours a day).

AI Era: The logic is value creation, selling APIs and Tokens to help enterprises or users improve efficiency and save time. "Sharing the value created" means that the monetization ceiling is much higher than simple traffic advertising. This is the real long-term potential for large companies in the AI era.

UBS believes that as AI commercial applications shift from "pure information provision" to "deep commercial penetration," internet companies are expected to revalue through scenario applications in the second half. For investors, the current low-level fluctuation may be a window period to deeply explore fundamentals and position for long-term AI benefits.