Goldman Sachs Maintains Buy Rating on Tencent, Target Price HK$700
AI summary card
Goldman Sachs Maintains Buy Rating on Tencent, Target Price HK$700
Tencent's AI strategy has made substantial progress; the Hy3 model demonstrates excellent performance, and agentic AI shows strong user retention. Improved domestic chip supply supports computing power expansion, paving the way for valuation repair.
- The Hy3 model leads its peer group, with a larger-scale iterative version expected within the next 6-9 months.
- Agentic AI is emerging as a new growth driver following chatbots, with successful validation in code generation scenarios.
- WeChat AI agents feature longer memory and multi-context processing capabilities, driving high user willingness to pay.
- Domestic ASIC chip supply is becoming increasingly abundant; Tencent has prepared for over a year to support inference across diverse chips.
- AI accelerates game content production, enhancing monetization efficiency without significantly increasing costs.
- The advertising business benefits from AI technology, with inventory resilience outperforming other formats such as e-commerce.
- Maintaining a Buy rating with a 12-month target price of HK$700, implying approximately 53% upside potential.
Report interpretation
Overview
This research report consists of meeting minutes compiled by Goldman Sachs based on the remarks made by James Mitchell, Tencent's Chief Strategy Officer, at the Asia Communacopia + Technology conference on May 18, 2026. The core view is that Tencent has achieved significant progress in its strategic transformation in the AI field, demonstrating long-term competitiveness ranging from foundation model reconstruction to the implementation of agentic AI applications, and further to the localization adaptation of the computing supply chain. Although valuation multiples have compressed since the beginning of the year, the firm believes that as the AI narrative deepens and cloud business returns to the ranks of China's top three hyperscale cloud providers, the path to valuation repair is clear. Consequently, the 'Buy' rating and a target price of HK$700 are maintained.
Core views
AI Foundation Models and Agentic Applications: Tencent has comprehensively restructured its underlying large model strategy, including rebuilding the LLM team, redesigning the pre-training and reinforcement learning stacks, cleaning and expanding datasets, and shifting evaluation criteria from public benchmarks to general intelligence. The resulting Hy3 model performs excellently within its scale category, with iterations toward larger and smarter models expected within the next 6-9 months. At the application layer, Agentic AI is becoming the mainstream use case following chatbots. CodeBuddy and WorkBuddy serve as core components, with code generation acting as a clear early validation point, gradually expanding into general computer usage and white-collar productivity scenarios. WeChat, WeCom, and QQ serve as user interaction interfaces; among them, WeChat AI agents possess longer memory than traditional chatbots, can handle multi-party contexts, exhibit high retention rates among active paying users, and tend to consume more tokens and upgrade to higher-priced tiers. Computing Supply Chain and Cloud Business: Facing changes in China's computing supply landscape, Tencent has adopted a more aggressive AI strategy. After more than a year of preparation, Tencent's technology stack can efficiently absorb diverse domestic ASIC chips and conduct large-scale inference, enabling it to direct new computing capacity toward external Tencent Cloud demand. ROI on AI investments is divided into short cycles (such as ad tech and Tencent Cloud; the former has higher profit margins, while the latter has larger revenue scale but lower margins compared to gaming/ads) and long cycles (such as Hy models, which create long-term value through differentiated data and WeChat scenarios). Empowerment of Gaming and Advertising Businesses: In the gaming sector, AI has primarily accelerated the production of monetizable content (such as virtual clothing), increasing content update speed without a proportional increase in costs; early IP development is also trending towards modular 3D asset creation. Total revenue from evergreen games has reached record highs, with new games providing incremental contributions. In marketing services, user consumption in Moments and Video Accounts stems from autonomous intent, making its ad inventory more resilient against traffic substitution threats driven by agentic AI compared to formats like e-commerce. AI-led ad technology and inventory expansion have maintained healthy growth.
Analysis framework
Goldman Sachs, through the format of meeting minutes, deconstructs Tencent's AI strategy across four dimensions: the technical foundation (model reconstruction), the application layer (agentic AI), infrastructure (domestic chip adaptation), and commercial monetization (gaming, advertising, cloud). The analytical logic focuses on verifying the actual output (ROI) of AI investments, distinguishing between short-cycle and long-cycle returns, and paying attention to the supporting role of supply chain security for computing power expansion. Regarding valuation, the firm views the current low valuation (compressed from 18x at the start of the year to 13x) as the market not fully pricing in its AI progress, believing that subsequent valuation repair depends on the advancement of the AI narrative in the second half of 2026 (such as the launch of WeChat AI agents, the introduction of GSe, etc.).
Methodology notes
SOTP Valuation Method
Goldman Sachs employs the Sum-of-the-Parts (SOTP) method to value Tencent, which involves separately evaluating the value of different business segments such as gaming, advertising, fintech, and cloud, and then summing them up to derive a 12-month target price of HK$700. This method is suitable for large technology companies with diversified businesses and can more accurately reflect the true value of each segment.
Computing Supply Chain Transmission
The report analyzes how improvements in the upstream supply of domestic ASIC chips transmit to Tencent's midstream computing infrastructure, thereby supporting the growth of downstream cloud service demand. This top-down industry chain perspective helps understand the practical support of technological breakthroughs for business models.
Ecosystem Barriers and Data Advantages
Emphasizes that the differentiated data scenarios and user stickiness brought by the WeChat ecosystem constitute Tencent's moat in the AI era. Specifically, the multi-context processing capability of WeChat AI agents is a key competitive advantage distinguishing it from other pure technology vendors.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- Tencent Holdings (0700.HK)Direct beneficiary, core carrier of AI strategic transformation
- Strengths
- High barriers in the WeChat ecosystem, excellent performance of the Hy3 model, strong adaptability to domestic chips, robust cash flow from gaming and advertising businesses
- Weaknesses
- Relatively lower profit margins in cloud business, significant AI R&D investment brings short-term reinvestment risks
- Comparison
- Compared to other major internet companies, Tencent has greater user stickiness and data advantages in AI application implementation within social scenarios
- Risks
- Intensified industry competition, delays in game license approvals, AI progress falling short of expectations
Key data
- Current Share PriceHK$456.40Closing price as of May 15, 2026
- Target PriceHK$700.0012-month SOTP valuation target price
- Implied Upside53.4%Calculated based on current price and target price
- Valuation Multiple Change18X -> 13XP/E ratio compressed from 18x at the start of the year to the current 13x
- Hy3 Model Iteration Expectation6-9 monthsExpected launch of a larger-scale, smarter next-generation model
Impact & implications
The research report believes that Tencent is transforming from a follower in foundation models to a competitive participant. Substantial progress in its AI strategy (such as the Hy3 model and agentic AI retention data) will provide momentum for valuation repair. As domestic chip supply issues ease, Tencent Cloud is expected to re-establish its position as one of China's top three hyperscale cloud providers. For investors, the focus should shift from mere traffic anxiety to the productivity improvements brought by AI and the monetization capabilities of new business models (such as the token economy).
Risks
- Intensified industry competition in the performance advertising sector
- Unexpected delays in game launches or license approvals
- Slowing growth in fintech and cloud businesses
- AI development progress slower than expected
- Reinvestment risks arising from large-scale capital expenditures
What to watch
- Potential launch status of WeChat AI agents in the second half of 2026
- Progress in the launch of GSe (possibly referring to related AI products or services) in the second half of 2026
- Evolution of the competitive landscape for OS-level AI agents in the consumer agent era
- How WeChat AI agents will achieve differentiated competition through the Mini Program ecosystem