After the company's founder called its AI effort a leaky ship, Tencent's cloud boss is betting on enterprise tools over headline-grabbing models
Tencent built one of the most dominant consumer franchises in tech history with WeChat. Then the post-ChatGPT era arrived, and the company found itself chasing rivals like DeepSeek instead of setting the pace.
This summer, Tencent’s billionaire founder reportedly described the company’s AI effort as “a leaky ship.”
One of his key lieutenants has now pitched a fix. Dowson Tong, CEO of Tencent Cloud and Smart Industries Group, wants the company to retool its AI push around businesses rather than trying to win a race for the biggest foundational model.
The marathon pitch
Tong laid out his thinking during a fireside chat on June 5, then expanded on it in an internal publication. His central framing: AI is a “marathon,” not a sprint.
Instead of measuring success by benchmark scores on one flagship model, Tong argues for scenario-driven applications and strong engineering execution.
He is also pushing a multi-model approach on Tencent Cloud. Customers get to choose among models rather than being locked into a single in-house option, with openness and innovation pitched as priorities.
The receipts so far
Tong’s argument rests partly on internal results. Tencent’s coding assistant, CodeBuddy, is used by more than 12,000 engineers inside the company.
The tool now generates more than 50% of new code, according to the company’s figures. It has also cut coding time by more than 40%.
Tencent has not abandoned model development either. The company has shipped more than 30 Hunyuan model releases over the past year, according to the research findings.
Tencent’s 2025 R&D spending reached record levels, supporting both self-developed models and the computing resources needed to run them.
How Tencent got here
The restructuring that set up Tong’s pitch began in December 2025. That month, Tencent appointed Shunyu Yao as Chief AI Scientist.
Yao’s arrival triggered a reshuffle of the company’s AI organization. Tencent merged its AI Lab with the Hunyuan team and created specialized departments focused on designing models and products together.
Tong’s remarks have been described as reflecting a turnaround following Yao’s appointment.
What this means for Tencent and its rivals
For investors, the most important signal is a shift in how Tencent wants its AI spending judged. Tong is steering the conversation toward tangible business value and return on investment, rather than raw model capability.
Heavy capital expenditure is easier to defend when you can point to engineers saving more than 40% of their coding time.
The model-agnostic cloud strategy could also widen Tencent’s addressable market in enterprise AI. Businesses that want flexibility may find a cloud provider offering multiple models more appealing than one pushing a single in-house option.
What to watch next: whether enterprise customers sign up for Tencent Cloud’s AI offerings at scale, how the reorganized Hunyuan and AI Lab teams perform after the merger, and whether R&D spending keeps climbing from its 2025 record.
