Tencent says it could make instant profits on $53bn hardware splurge by renting it for AI workloads
Plans to build models instead because it thinks selling tokens will prove more lucrative in the long term
Tencent, the Chinese tech giant, has invested $53 billion in new hardware for AI development and has promised to make profits from it by renting out its infrastructure to AI workloads. At its Q2 earnings call, the company's chief strategy officer, James Mitchell, stated that if Tencent rented its compute capacity, it could recover depreciation costs "almost immediately" given the strong demand for computing resources.
Company president Martin Lau emphasized that Tencent is not focused on instant profits but rather on building its own models to state-of-the-art status and deploying AI applications to market leadership in China. By providing superior intelligence through state-of-the-art models and market-leading AI applications, Tencent believes it can achieve superior economic returns in the long run.
Tencent's new AI agents, WorkBuddy, and CodeBuddy, are designed to accelerate cloud migration projects and create more business for Tencent cloud. The company recently released its largest model, Hunyuan-3, and plans to release Hunyuan-4, which will be even bigger and more capable than larger models from other companies. With Q2 revenue growing 11 percent to $30.3 billion and net profit rising 9 percent to $10.3 billion, investors are unsure about the implications of Tencent's strategy.
Despite this, Tencent's flagship messaging apps, Weixin and WeChat, saw a 7 million rise in average monthly active users to 1.349 billion, and advertising-related revenue increased by 22 percent.
Written by urgent.news from The Register's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.