Tencent is slowing down its deployment of GPUs.
During the company's recent Q4 2024 earnings call, chief strategy officer James Mitchell explained that with the Chinese tech giant's implementation of DeepSeek, it is getting more out of its GPUs.
As highlighted in a report from The Register, the company has restrategized following the breakthrough made earlier this year by DeepSeek.
In January 2025, DeepSeek claimed that it had managed to train a Large Language Model, performant at a level comparable with those developed by OpenAI and Meta, but with less advanced chips and for a fraction of the cost - approximately $5.6 million.
While there are caveats to that claim, including the fact that the number only counts the single training run, not all the research runs, and investigations are ongoing as to whether the company accessed more advanced GPUs than it said, the news created a shift in thinking when it came to training workloads.
Mitchell noted in the earnings call that, last year, the company believed that each new generation of LLMs needed more GPUs.
“That period of time ended with the breakthroughs that DeepSeek demonstrated,” he said. “And now, the industry and we, within the industry, are getting much higher productivity on a large language model training from existing GPUs without needing to add additional GPUs at the pace previously expected.”
The company saw a capex in 2024 of around $10.6 billion, more than triple that of the year prior, with company president Martin Lau noting they had stepped up the purchase of GPUs.
Capex for 2025 is expected to be in the low teens percentage of the company's revenue. FY 2024 revenue was $91.1 billion, up eight percent year over year.
In the US, Microsoft has estimated that it will spend $80 billion on AI data centers this year which is significantly more than Tencent's spend for 2024, or likely spend in 2025. Meanwhile, Alibaba Group is expecting to spend around $53 billion in cloud and AI infrastructure in the next three years in what will be the company's "most concentrated and highest level of investments."
Mitchell said of this: "That's because the Chinese companies are generally prioritizing efficiency and utilization - efficient utilization of the GPU servers. And that doesn't necessarily impair the ultimate effectiveness of the technology that's being developed. And I think DeepSeek's success really sort of symbolizes and demonstrated that reality.”
Tencent's profit for the year was $31.6 billion, up 40 percent. The quarter saw a revenue of $24 billion, up 11 percent, while profit reached $12.6 billion, up 17 percent. This is across the whole company and not exclusive to cloud and AI income.
Chinese competitor Alibaba Cloud's most recent quarter saw revenue growth of 13 percent year-over-year, to $4.349 billion, while AI-related product revenue achieved triple-digit growth for the sixth consecutive quarter.
