Tencent Turns Down 30% Instant Profit on $53B AI Infrastructure to Build Own Models

7 Sources

Share

Tencent reported Q2 earnings revealing it rejected immediate 30% profits from renting compute capacity, instead investing $53 billion in AI infrastructure. The Chinese tech giant's capital expenditure jumped 176% as it builds proprietary AI models like Hunyuan-4, despite investor skepticism pushing shares down 26% year-to-date.

Tencent Rejects Immediate Returns on AI Infrastructure

Tencent disclosed during its Q2 earnings call that it could achieve more than 30% profit margins by renting out its newly acquired compute capacity, but the Chinese tech giant is deliberately choosing a different path

1

. Company president Martin Lau told analysts the firm has offers for its AI infrastructure "at more than 30 percent profit compared to the price that we paid just a few months ago," representing what he called "a decent return in an immediate timeframe"

1

. Chief Strategy Officer James Mitchell confirmed demand for compute resources is so intense that Tencent could recover its depreciation costs "almost immediately" if it rented its infrastructure

3

. Instead, Tencent is "playing a different game," allocating substantial compute to building state-of-the-art AI models and market-leading AI applications in China

1

.

Capital Expenditure Surge Turns Cash Flow Negative

Tencent's capital expenditure climbed 176% year-over-year to 52.8 billion yuan, approximately $7.8 billion, in Q2 2026

4

. This massive AI infrastructure spending pushed the company's free cash flow into negative territory at minus 13.8 billion yuan, marking a dramatic shift for a company that has generated cash consistently for two decades

4

. Against the previous quarter, capital expenditure rose 65%, illustrating the acceleration of Tencent's AI strategy

4

. Operating costs climbed 22.6% in the quarter as the company prioritized building proprietary AI models over short-term profitability

4

. Martin Lau positioned the infrastructure investment as providing "clear downside protection," noting that in a worst-case scenario, the compute capacity could be rented out at cost recovery

4

.

Q2 Earnings Show Revenue Beat Despite Profit Miss

Tencent posted Q2 revenue of 204.78 billion yuan ($30.36 billion), beating analyst expectations of 202.17 billion yuan and representing 11% year-over-year growth

2

. However, net profit came in at 56 billion yuan, missing the expected 61.82 billion yuan

2

. Stripping out one-time factors and certain non-cash items, adjusted net income reached 68.4 billion yuan, up 9% versus the same period last year

2

. The company's flagship messaging apps, Weixin and WeChat, saw average monthly active users rise seven million to 1.349 billion

3

. Domestic games revenue accelerated sharply to 47.3 billion yuan, up 17% year-over-year, driven by titles including Delta Force and Valorant PC and Mobile

2

. This marked a significant improvement from the 6% growth the division saw in Q1 2026

2

.

AI-Driven Advertising and Cloud Growth

Advertising-related revenue rose 22% as Tencent leveraged AI-driven ad targeting capabilities

3

. Marketing services reached 43.6 billion yuan, with the company crediting AI technology for improved ad performance

4

. The cloud business grew in the low twenties percentage-wise, with Tencent also implementing price increases during the quarter

4

. James Mitchell highlighted that CodeBuddy, Tencent's code generation tool, is accelerating cloud migration projects and creating additional business for Tencent cloud

1

. WorkBuddy, described as an agent swarm that can "plan, execute, and run tasks in parallel, handing back complete deliverables end-to-end in one flow," has become China's most popular AI office tool with 21 million monthly visits in June

4

.

Source: France 24

Source: France 24

Hunyuan Model Development and AI Workloads Strategy

Tencent released its 295-billion parameter open-weight Hunyuan-3 model in July, though Martin Lau described it as "a very small model"

1

. The company is preparing Hunyuan-4, which Lau promised will be bigger and more capable than larger AI models from competitors

3

. Tencent is designing its products specifically to work with Hunyuan-4, with mutual optimization expected to make these products more powerful than if they relied on other models

1

. The company also plans a fifth version of Hunyuan, with Lau stating that Tencent will eventually deliver a state-of-the-art model

3

. The Tencent AI strategy centers on achieving "superior intelligence" through state-of-the-art models and market-leading AI applications to "convert into superior economic returns over the longer term," according to Lau

1

.

Investor Skepticism Despite Operating Strength

Tencent stock fell 26% year-to-date at Wednesday's closing bell in Hong Kong, with the company's share price trending down since the earnings announcement and dipping around three percent following the results

2

3

. Roughly $170 billion of market value has disappeared as investors express concern about rising spending and intense competition from Alibaba, DeepSeek, and Moonshot AI

4

. Prosus, the Amsterdam-listed group holding a large Tencent stake, fell 6% on the earnings day

4

. The market appears conflicted, punishing Tencent both for spending appearing reckless and for not committing to multiyear targets like Alibaba's $50 billion pledge over three years

4

. Tencent has set no multiyear target, saying only that its investment in AI products will double this year

4

. Founder Pony Ma told analysts the company is "making substantial progress toward building a new AI-empowered Tencent," with the critical test arriving when Hy4 launches later this year

4

5

. Tencent is also testing Xiaowei, an AI assistant embedded directly in WeChat, currently in small-scale prototype testing

4

.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved