Tencent Executive Dowson Tong Responds to Concerns Over Slow AI Progress
Dowson Tong, CEO of Tencent's Cloud and Smart Industries Group, addressed internal anxieties regarding Tencent's slow AI progress, admitting that compute shortages have hindered the iteration of their Hunyuan large language model. He emphasized that the AI race is a marathon rather than a sprint, where long-term endurance and real-world application scenarios are more critical than starting early. This internal message highlights the strategic challenges faced by Chinese tech giants, particularly compute resource constraints, while revealing Tencent's long-term strategy of prioritizing engineering execution and leveraging its massive ecosystem of application scenarios. Tong noted that while competitors have gained significant visibility in the business-to-business (ToB) market, Tencent's core advantages lie in its engineering capabilities and diverse application scenarios. Tencent's Chairman Pony Ma also previously supported this steady approach, stating that the company aims to avoid rushing into markets unprepared.
## BACKGROUND
Tencent's primary AI foundation model is Hunyuan, which powers various services across its ecosystem, including Tencent Cloud and WeChat. In recent years, global tech companies have engaged in an intense AI race, but Chinese firms have faced unique bottlenecks, particularly in securing high-end GPUs for training large models due to export restrictions.