UOB Kay Hian has issued a research report adjusting its price target for TENCENT (HK: 00700) upwards from HK$670 to HK$680 and reaffirming its "Buy" rating.
The formal launch of Tencent's Hunyuan Hy3 model signifies a notable advancement in its foundational model capabilities, attributed to stronger benchmark performance, coding and agent upgrades, and competitive pricing.
Widespread deployment within the Tencent ecosystem aids in gathering real-world usage feedback and accelerates model iteration.
Initial positive momentum, including a 20-fold increase in tokens and a 6-fold rise in users voluntarily opting for Hy3, indicates growing user acceptance and stronger relevance in enterprise productivity and workflow-oriented AI applications.
UOB Kay Hian forecasts that Tencent's Q2 2026 revenue will increase by 11.8% year-on-year to RMB 206.2 billion, with non-GAAP net profit growing 10.8% year-on-year to RMB 69.9 billion, resulting in a net profit margin of 33.9%.
By business segment, the firm expects online games revenue to grow 13.7% year-on-year to RMB 67.3 billion, driven by an 11.7% increase in domestic games and an 18% rise in international games.
Advertising revenue is anticipated to grow 18.8% year-on-year to RMB 42.5 billion, while FinTech and Business Services (FBS) revenue is projected to increase 9.3% year-on-year to RMB 62.7 billion, supported by a 22.0% rise in cloud revenue and a 7.0% growth in fintech revenue.
The brokerage expects gross profit and operating profit to grow 11.3% and 15.5% year-on-year, reaching RMB 116.9 billion and RMB 69.4 billion respectively.
Overall, the firm anticipates Q2 2026 results will largely meet expectations, with potential upside in online games possibly partially offset by a slowdown in FBS growth.
UOB Kay Hian stated that Hy3 significantly enhances the reliability of AI agents.
The most significant breakthrough lies in Hy3's ability to go beyond simple Q&A and complete complex multi-step tasks.
Hy3's agent capabilities can now handle multi-step tasks, directly generating PPT, Word, Excel, PDF, and HTML files based on natural language prompts within Yuanbao.
Its hallucination rate has decreased from 12.5% to 5.4%, making its outputs more trustworthy for enterprise applications.
Simultaneously, enhanced coding and agent capabilities improve performance in complex workflows, including software development, financial modeling, and front-end design.
The firm views this as an early sign that Tencent is strengthening the competitiveness of its foundational models and narrowing the gap with leading frontier models.