In the second quarter of 2025, Nvidia saw a remarkable 154% increase in revenue for its datacentre business, reaching $26.3bn. This contributed to the company’s overall record quarterly revenue of $30bn, up 15% from the previous quarter and 122% from the previous year.
Nvidia’s founder and CEO, Jensen Huang, attributed this success to the transition to accelerated computing by datacentre operators globally. Demand for Nvidia’s Hopper GPU computing platform, specifically for training and inferencing in large language models, recommendation engines, and generative AI applications, drove the growth in the datacentre business.
The company reported that growth in revenue came from both cloud service providers and consumer internet and enterprise companies. Nvidia’s networking products also saw a 16% sequential increase in revenue.
Huang emphasized the importance of accelerated computing in addressing the increasing demand for computational power. He highlighted the cost-saving benefits of GPU-powered servers and predicted that all datacentres would eventually incorporate GPU technology.
Huang’s vision for accelerated computing aims to provide datacentre operators with increased computational power in a sustainable manner, while also reducing the cost of computing and avoiding “computing inflation”. He pointed out the advantages of liquid-cooled datacentres, which offer three to five times the AI throughput compared to traditional methods.