Nvidia drives forward accelerated computing advantage

Nvidia has posted a 154% increase in revenue for its datacentre business in the second quarter of 2025, compared with the same period a year ago, generating $26.3bn. 

Overall, the company posted what it said was “record quarterly revenue” of $30bn, up 15% from the first quarter and up 122% from a year ago.

Nvidia founder and CEO Jensen Huang said the results showed that datacentre operators were transitioning to accelerated computing. “Nvidia achieved record revenues as global datacentres are in full throttle to modernise the entire computing stack with accelerated computing and generative AI [artificial intelligence],” he said.

Prepared remarks accompanying the financial results show that Nvidia’s datacentre numbers were driven by demand for its Hopper graphics processing unit (GPU) computing platform, which is being used for training and inferencing in large language models, recommendation engines and generative AI (GenAI) applications.

The company said the sequential growth was driven by consumer internet and enterprise companies. The results also show a roughly even split between the spending by cloud service providers (45%) and the technology investments being made by consumer internet and enterprise companies (50%). The remaining growth in Nvidia’s datacentre business came from its networking products, which saw a sequential rise in revenue of 16%.

According to a transcript of the earnings call, posted on Seeking Alpha, when Huang was asked about how datacentre operators justified the return on investment for GPU-powered servers, based on Nvidia technology, he described a transition from general-purpose computing to accelerated computing.

“CPU scaling has been known to be slowing for some time. It is slowing to a crawl,” he said. “And yet the amount of computing demand continues to grow quite significantly. You could maybe even estimate it to be doubling every single year, so if we don’t have a new approach, computing inflation would be driving up the cost for every company, and it would be driving up the energy consumption of datacentres around the world.”

Over the past year, Huang has been pushing the idea of accelerated computing, powered by datacentre GPUs. “It’s not unusual to see someone save 90% of their computing cost. And the reason for that is, of course, you just sped up an application 50 times, [so] you would expect the computing cost to decline quite significantly,” he said.

Due to the need to provide more computing demand, Huang predicted that every single datacentre would incorporate GPU technology. “The world builds about $1tn worth of datacentres – $1tn worth of datacentres in a few years will be all accelerated computing.”

Huang is betting that his vision for accelerating workloads with Nvidia GPUs will offer datacentre operators a way to increase computational power in their datacentres in a sustainable way while also driving down the cost of computing, avoiding what he calls “computing inflation”.

He claimed that a liquid-cooled datacentre offers three to five times the AI throughput compared to the past: “Liquid cooling is cheaper and allows you to have the benefit of this capability we call NVLink, which allows us to expand it to 72 Grace Blackwell packages, which has essentially 144 GPUs.”

Read more on Clustering for high availability and HPC

• 

  An era of post Moore’s Law computing?

  

  By: Cliff Saran

• 

  Singtel expands GPUaaS offering to Southeast Asia

  

  By: Aaron Tan

• 

  Nvidia sees massive datacentre compute growth, says networking is next

  

  By: Cliff Saran

• 

  GTC 2024: Storage suppliers queue up to ride the Nvidia AI wave

  

  By: Antony Adshead