← 返回 Avalaches

Tencent Holdings Ltd. 截至3月的季度营收年增9%,达1965亿元人民币(约28.9 billion美元),为近6季最慢增速;净利增至581亿元人民币并符合预期。营收偏弱部分归因于今年农历新年较晚,削弱其对线上娱乐消费的传统拉动效应,且国内游戏收入走弱凸显核心业务成长动能老化的疑虑。

市场焦点转向AI变现与成本压力:Tencent 表示将今年AI投资至少加倍至逾360亿元人民币,但在用户采用与先进大型语言模型进展上落后 ByteDance Ltd. 与 Alibaba Group Holding Ltd.;同时基础设施支出扩张挤压利润率。相较之下,Alibaba 提出未来3年累计逾50 billion美元的AI支出目标,而美国 Amazon.com Inc.、Alphabet Inc.、Meta Platforms Inc.、Microsoft Corp. 今年合计计划投入逾700 billion美元,形成明显量级落差。

资本市场反应偏负面:Tencent 今年市值约蒸发160 billion美元(约-23%),跌幅大于同日公布财报的 Alibaba;投资人也转向纯AI模型公司如 Zhipu、MiniMax Group Inc.,其估值自1月亮相后倍增。技术面上,Tencent 升级基础模型 Hunyuan(HY),由 OpenAI 前员工 Yao Shunyu 主导重整后首度高压测试,但在程式能力与语言细腻度等仍落后 Moonshot、DeepSeek;同时其主要聊天机器人采用 DeepSeek,Nomura 的 Jialong Shi 指出此合作可在HY未达顶尖时提供备援。

d0600da12c18.png



Tencent Holdings Ltd. posted its slowest revenue growth in six quarters: March-quarter sales rose 9% year on year to CNY 196.5 billion (about USD 28.9 billion), while net income increased to CNY 58.1 billion, in line with estimates. The softer top line was partly linked to the later timing of Lunar New Year, and it coincided with weaker domestic game sales, reinforcing concerns that core engines such as gaming and advertising are aging.

Investor attention is now on the pace and payoff of AI spending as infrastructure costs pressure margins. Tencent plans to at least double AI investment to more than CNY 36 billion this year, yet it trails ByteDance Ltd. and Alibaba Group Holding Ltd. in user adoption and frontier large-language-model progress. By comparison, Alibaba has set a rolling three-year AI spending target above USD 50 billion, while Amazon.com Inc., Alphabet Inc., Meta Platforms Inc., and Microsoft Corp. together plan to deploy more than USD 700 billion this year, underscoring the gap versus US hyperscalers.

The market response has been negative: Tencent has lost roughly USD 160 billion of market value this year, or 23%, a steeper drop than Alibaba, and investors have rotated toward pure-play model makers such as Zhipu and MiniMax Group Inc., whose valuations have multiplied since January debuts. Operationally, Tencent upgraded its Hunyuan (HY) foundation model under OpenAI hire Yao Shunyu, but it still lags Moonshot and DeepSeek on attributes like coding and linguistic nuance; Tencent also powers its main chatbot with DeepSeek, which Nomura’s Jialong Shi framed as a fallback if HY fails to reach state-of-the-art performance.
2026-05-15 (Friday) · a5ce1cab62835c837b5a292005c856aba166b92e