Nvidia has reported a record quarter for its data centre division as it continues to profit from the boom in popularity of generative AI systems. The US GPU maker’s chips are routinely used to train models like OpenAI’s GPT-4, the technology behind ChatGPT, and it anticipates strong demand continuing, forecasting a $4bn revenue bump in the next quarter, well ahead of market expectations.
Overall revenue for Nvidia was down year-on-year, with its income for Q1 dropping from $8.2bn last year to $7.1bn in the last three months. But against a backdrop of massive losses and job cuts at other tech giants, which, like Nvidia, are contending with difficult market conditions, CEO Jensen Huang sounded a characteristically optimistic note about the outlook for the business.
Nvidia data centre revenue on the rise
Huang’s optimism would appear to be well-founded. Nvidia made $4.3bn from its data centre division in the last quarter, a 13% year-on-year jump and 18% up on the previous quarter. This countered a big dip in revenue from the company’s gaming division, which saw income drop 38% year-on-year.
Data centre income has been built on the success of Nvidia’s flagship AI GPU, the A100, which sells for around $10,000 a chip. It is the hardware of choice for most of the systems that train large language AI models, and these supercomputers can use thousands of the chips at a time. The AI supercomputer constructed by Microsoft for OpenAI to use to train its models features 10,000 of Nvidia’s GPUs. It means Nvidia, traditionally a gaming company, now derives most of its income from enterprise.
The company is not resting on its heels, launching a replacement for the A100, the H100 this year, and building out its service offering with a cloud-based AI supercomputer platform through which businesses can access one of its DGX AI supercomputers.
It has also been dipping its toe into quantum computing, with DGX Quantum, a cloud platform for developers working on early quantum computing systems. As reported by Tech Monitor, Nvidia has been working with British quantum company Orca Computing to marry up quantum and classical machines in a bid to speed up machine learning.
The outlook is sunny for Nvidia
Nvidia is forecasting for the current quarter to be around $11bn, well ahead of expectations of market analysts, which had suggested the company would make just over $7bn. Unsurprisingly, this sent its share price soaring.
“The computer industry is going through two simultaneous transitions — accelerated computing and generative AI,” said Nvidia CEO Huang. “A trillion dollars of installed global data centre infrastructure will transition from general-purpose to accelerated computing as companies race to apply generative AI into every product, service and business process.
He added: “Our entire data centre family of products is in production. We are significantly increasing our supply to meet surging demand for them."