Tencent said it could rapidly recover the depreciation costs of its $53 billion quarterly hardware investment by renting its computing capacity to AI customers, as demand is strong and offers reportedly exceed its recent purchase price by more than 30 per cent. However, the Chinese technology group is prioritising a longer-term strategy of using much of the capacity to develop its own advanced AI models and embed them in its products, aiming for higher returns from AI services and applications.

The company expects revenue from AI tokens and tools such as WorkBuddy and CodeBuddy, which it says can support task automation and cloud-migration work. Tencent released the 295-billion-parameter Hunyuan-3 model in July and plans larger Hunyuan-4 and Hunyuan-5 versions; meanwhile, second-quarter revenue rose 11 per cent to $30.3 billion and net profit increased 9 per cent to $10.3 billion. Despite growth in advertising, gaming and WeChat/Weixin users, Tencent’s shares fell by about 3 per cent after its earnings announcement.