Computing power is rewriting the 'underlying script' of China's internet: from an 'infinite game' to 'resource precision' — 3132 HK and 2812 HK dual-track capturing new era opportunities

May 13,$TENCENT (00700.HK)$ And, $Alibaba (BABA.US)$ (Hereinafter referred to as 'Alibaba') chose the same day to release their quarterly earnings report. At first glance, both are talking about AI: Alibaba’s Qwen, ModelScope, and Kunlun chips, Tencent’s Hy3 preview, WorkBuddy, and AI Agent. Both sides pointed out that AI revenue is growing rapidly and commercialization is being realized. However, upon closer reading of the numbers and management's wording, the growth logic of China's tech giants is being reconstructed.

Over the past two decades, the operational logic of Chinese internet companies has been extremely straightforward: the more users, the better; the larger the transaction volume, the better. Because serving one more user almost incurs zero marginal cost. But AI has broken this equation. Every question-and-answer session, every round of generation, every instance of Agent invocation burns real GPU (Graphics Processing Unit) computing power. Intelligence comes at a cost, and that cost is no longer zero. This time, the giants have encountered unavoidable computational constraints.

Alibaba: From selling goods to selling computing power

$BABA-W (09988.HK)$ The most crucial signal from this earnings report is not a specific revenue figure but a clear declaration by management: full AI technology investment has bid farewell to its long incubation period and officially entered the track of scaled commercial returns.

The core supporting this transformation is the cloud business. External commercial revenue from cloud services is growing at a rate of 40%, with AI-related products accounting for more than 30% of revenue for the first time. The annual recurring revenue (ARR) from AI models and application services, including the BaiLian MaaS (Model-as-a-Service) platform, is set to exceed 10 billion yuan and will surpass 30 billion yuan by the end of the year. Alibaba’s in-house developed GPU chip by Pingtouge has achieved scaled mass production, with over 60% of its computing power serving external commercial clients. From chips to models, from cloud platforms to application services, Alibaba is one of the few Chinese companies to have completed a full-chain closed loop. [1][2]

But the deeper change lies in the reversal of organizational logic. In the past, Alibaba's resource allocation was determined by the scale of the business — e-commerce being the largest, it got priority in budget allocation. Now, the decision-making power has shifted to the computing power pool. Whoever can enter the computing power scheduling process is considered a core business. Alibaba is no longer just about selling goods or advertisements; it now focuses on allocating computing power and delivering intelligence.

Tencent: Insufficient GPUs lead to centralized organizational power

Most cloud vendors are eager to monetize their computing power immediately, $TENCENT (00700.HK)$ but Tencent has taken the opposite approach, prioritizing internal service needs for its GPU capacity. The reason is simple — internal demand for computing power already far exceeds that of external customers. Large models like HunYuan, WeChat Agent, Channels, the advertising system, and WorkBuddy are all massive consumers of GPUs. AI isn't a standalone department within Tencent but rather a continuous effort to empower core business growth through AI.

In the past, Tencent's various business lines — WeChat, gaming, and advertising — operated with independent resources. However, in the AI era, due to insufficient GPUs, Tencent has centralized computing power as a company-wide resource, managing it uniformly. The center of organizational power has shifted toward infrastructure and the AI hub. Tencent is recalibrating which businesses deserve investment and can generate long-term returns. It no longer solely chases user numbers but instead focuses on whether AI can improve return rates and enterprise collaboration efficiency.

Complementary positioning of 3132 HK (Samsung Bloomberg Global Semiconductor ETF) and 2812 HK (Samsung CSI China Dragon Net ETF)

This structural shift creates two different levels of opportunity for investors, corresponding perfectly to two ETFs under Samsung.

$Samsung Bloomberg Global Semiconductor ETF (03132.HK)$— Seizing the foundational 'picks-and-shovels' providers of computing power
Whether Alibaba is busy deploying computing power to deliver intelligence or Tencent is internally feeding its GPUs, all these demands ultimately point back to the same source: the global semiconductor supply chain. Together, Tencent and Alibaba spent nearly 60 billion yuan on capital expenditures in Q1 2026, most of which went to GPUs and AI infrastructure. [2] As AI applications move from training competitions to large-scale inference deployment, demand for computing power is entering a structurally long-term expansion phase. $Samsung Bloomberg Global Semiconductor ETF (03132.HK)$ fully covers the global semiconductor leaders from design ( $NVIDIA (NVDA.US)$ 、 $Advanced Micro Devices (AMD.US)$ ), manufacturing ( $Taiwan Semiconductor (TSM.US)$ ), storage (Samsung, SK Hynix) to equipment ( $ASML Holding (ASML.US)$ ). Regardless of how Chinese internet giants diversify, these "pickaxe sellers" will reap every wave of computing power.

2812 HK – Capturing the value revaluation of platform economies
When the market starts pricing internet platform companies based on their 'AI monetization capability' rather than 'traffic scale,' Tencent, Alibaba, Meituan, PDD Holdings, and Baidu are facing a renewal in valuation logic. Improvements in advertising ROI, a jump in cloud AI revenue share, and the unlocking of self-developed chip value are all validating this new framework. $Samsung CSI China Dragon Internet ETF (02812.HK)$ conveniently holds a basket of these first-tier Chinese internet AI commercialization players. As the rules of the game shift from 'endless expansion' to 'efficiency is king,' platforms with massive user scenarios and AI invocation capabilities are expected to receive structural valuation repair.

For investors looking to participate in this structural change, through $Samsung Bloomberg Global Semiconductor ETF (03132.HK)$and$Samsung CSI China Dragon Internet ETF (02812.HK)$ 's dual-product allocation, it forms an AI-themed portfolio across two dimensions: 'computing power supply' and 'platform application.'

Data source:

[1] Wall Street News (May 13, 2026)

[2] Bloomberg, data as of May 14, 2026

Disclaimer and Important Notices

• Investment involves risks. Past performance is not indicative of future results. The price of funds can go up as well as down, and investors may suffer all or substantial losses of their investments. Investors should not make investment decisions solely based on this material.

• The Samsung Bloomberg Global Semiconductor ETF may be exposed to key risk factors such as: investment risk, new index risk, stock market risk, concentration risk, semiconductor industry risk, emerging market risk, risks associated with depositary receipts, currency risk, securities lending transaction risk, other currency distribution risks, risks of distributions being paid from capital or effectively from capital, passive investment risk, trading risk, risks arising from different trading hours, reliance on market makers, tracking error risk, and termination risk.

The Samsung CSI China Dragon Internet ETF is a sub-fund of Samsung ETF Trust II. The investment objective of the sub-fund is to provide investment performance that closely tracks the CSI Global China Internet Index before deducting fees and expenses.

The main risk factors that the Samsung CSI China Dragon Internet ETF may face include: general investment risks, currency risks, stock market risks, concentration risks in the internet and technology sectors, China market risks, multi-counter risks, risks of distributions from capital or effectively from capital, risks associated with the Stock Connect mechanism, RMB currency and conversion risks, securities lending transaction risks, passive investment risks, tracking error risks, trading risks, risks of trading differences, termination risks, and risks related to reliance on market makers and liquidity.

• The above fund has been authorized by the Securities and Futures Commission of Hong Kong (SFC). Authorization does not imply official endorsement of the product. This material is for reference only and does not constitute an offer or solicitation to any person to buy, sell, or adopt any investment strategy.

• The Manager may, at its discretion, make cash distributions to unit holders from capital or total income (while charging all or part of the Fund's fees and expenses to the capital or paying them from the capital), thereby increasing distributable income for payment of distributions, which actually amounts to a distribution from capital.

• Distributions paid or effectively paid out of capital amount to a return of part of an investor's original investment or withdrawal of part of the original investment or capital gains attributable to that original investment. Any practice involving distributions being paid out of the capital of the product or effectively paid out of the capital of the product may result in an immediate reduction in the per-unit net asset value.

• This document has been prepared by Samsung Asset Management (Hong Kong) Limited (SAMHK) and has not been reviewed by the Securities and Futures Commission or any other regulatory authority. Investors should determine whether any investment product or strategy is suitable based on their individual financial situation, investment experience, and objectives. If you have any questions regarding the relevant information, you should seek advice from a professional advisor as needed.

• Certain information contained herein is compiled from third-party sources. SAMHK has made efforts to ensure the accuracy, completeness, and timeliness of such information and taken measures to verify and reproduce it accurately. However, SAMHK assumes no responsibility or liability for the accuracy of the data, any use thereof, or reliance on it. This content may contain forward-looking statements based on SAMHK’s opinions, expectations, and projections. SAMHK has no obligation to update or revise any forward-looking statements, and actual results may differ materially from those anticipated in forward-looking statements. All copyrights for the contents of this material (including all information, images, computer codes, text, logos, and designs) are owned by SAMHK. Without SAMHK’s consent, the information provided herein must not be reproduced or redistributed.

134K Views