Nvidia has announced that its datacentre division achieved record-breaking revenue of $19.4bn in the first quarter of 2025, marking a significant increase of 427% compared to the previous year. The company attributed this success to higher shipments of the Nvidia Hopper GPU computing platform, which is utilized for training and inferencing with various AI applications such as large language models, recommendation engines, and generative AI.
Overall, Nvidia reported total revenue of $26bn for the quarter, representing an 18% increase from the previous quarter and a remarkable 262% increase from the same period last year. The company’s chief financial officer, Colette Kress, highlighted the growing importance of both training and inferencing in driving datacentre revenue, with inference alone accounting for approximately 40% of the total revenue.
Nvidia’s CEO, Jensen Huang, discussed the company’s focus on developing the next wave of growth through its Blackwell platform, which is designed to support trillion-parameter-scale GenAI. Additionally, the networking arm of Nvidia reported revenue of $3.2bn, up 242% from the previous year, with Kress emphasizing the potential for the Spectrum-X product to revolutionize ethernet-only datacentres and enhance networking performance for AI processing.
Looking ahead, Huang expressed the company’s commitment to advancing computing fabrics, including NVLink, InfiniBand, and ethernet networking. He emphasized the importance of bringing new switches, NICs, capabilities, and software stacks to support these computing fabrics and drive innovation in the industry. Kress projected that the Spectrum-X product line will become a multibillion-dollar opportunity for Nvidia within a year, signaling continued growth and expansion in the datacentre market.