UBS Reviews New Logic for HK Internet & AI Investments: Valuation Restructuring & Value Capture

On May 21, the Hang Seng Tech Index closed down 2.15%, bringing its year‑to‑date decline to nearly 15%. Amid the market gloom, Fang Jincong, head of China internet industry research at UBS, said at a media briefing that while the sector faces short‑term headwinds, the long‑term investment thesis for internet companies has fundamentally shifted.

1. Three Overlapping Pressures and Resilient Fundamentals

Fang Jincong attributed the index’s recent drop to three simultaneous pressures: a weak macro environment weighing on consumer‑facing businesses (e‑commerce, advertising, local services), rising AI‑related capital expenditures squeezing near‑term profits, and lingering uncertainty about the path to AI monetization.

Yet earnings reports reveal robust underlying strength. Internet giants continue to improve profit margins through cost‑cutting and efficiency gains, with core business losses narrowing sharply. They are also focusing on revenue‑generating channels such as API licensing, AI‑powered advertising, and gaming scenario integration. Fang believes negative news is already largely priced in, pointing to “low expectations, low valuations, and a bottoming‑out of fundamentals” — a sign that stock prices are hitting a cyclical trough.

2. Divergent Routes to Excess Returns: Large Models vs. Gaming

Looking ahead, UBS analysts outlined two distinct paths for investors seeking alpha:

Large‑model companies (growth logic): Seen as the most explosive segment, with potential for 10‑ to 30‑fold scale expansion. This is classic growth investing, targeting high returns by capturing cutting‑edge opportunities.

Gaming sector (certainty logic): Better suited for risk‑averse investors. Although expected growth is in the double‑digit range, top gaming firms trade at just 12–13 times earnings — offering a wide margin of safety and a high probability of excess returns.

3. Fundamental Differences in Valuation Approaches

Valuation frameworks diverge sharply between AI model providers and traditional internet companies:

AI startups (P/ARR method): Valued on price‑to‑annual‑recurring‑revenue multiples, with premiums driven by technological leadership, asset scarcity, and a liquidity premium around their initial listing. MiniMax and Zhipu’s Hong Kong stock performance, for instance, already reflects market recognition of their rapid growth.

Internet giants (profitability method): Still anchored to the profitability of legacy businesses. Because AI is capital‑intensive (heavy computing investment) and its contribution is hard to isolate, these companies cannot command the same valuation multiples as lightweight AI startups.

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

Fang Jincong noted that this AI wave is fundamentally different from the mobile internet era:

Mobile internet: The logic was traffic monetization, capped by users’ daily online time (roughly six to seven hours).

AI era: The logic is value creation — selling APIs and tokens that help enterprises or individuals improve efficiency and save time. “Sharing the value created” implies a monetization ceiling far higher than simple traffic‑based advertising. That, Fang argues, is the real long‑term opportunity for large internet firms in the AI age.

UBS believes that as AI commercial applications evolve from “pure information delivery” to “deep business penetration,” internet companies can re‑rate via scenario‑based use cases in the second half of the cycle. For investors, the current low‑level volatility may be a window to delve into fundamentals and position for long‑term AI beneficiaries.