AI, China and Data Centres: Nvidia’s Revenue in Context
Nvidia reports quarterly revenue of US$46.7bn, a 56% year-on-year increase, with its data centre division contributing US$41.1bn during the three months to July. That performance, fuelled by accelerating demand for AI infrastructure, comes despite mounting trade restrictions and investor caution tied to China.
The company’s data centre business, which supports major AI operations run by Microsoft, Meta and OpenAI, remains central to Nvidia’s revenue growth.
However, share prices fell in after-hours trading, as the market responded to the firm’s outlook and the ongoing uncertainty over export restrictions on advanced chips.
In a call to investors following the report's release on Wednesday, Nvidia CEO Jensen Huang said spending by the largest technology companies has reached new heights.
“Over time, you would think that AI would accelerate GDP growth. Our contribution to that is a large part of the AI infrastructure.”
AI chip demand powers data centre growth
Meta, OpenAI, Microsoft and other tech giants are all scaling up AI capabilities, pushing demand for Nvidia’s high-end processors designed to run and train complex machine learning models. That demand is helping drive rapid expansion across hyperscale data centre infrastructure.
While revenue growth remains strong, geopolitical tensions are putting pressure on operations. Since US President Donald Trump returned to the White House, his administration has moved to tighten chip exports to China, citing national security concerns.
The decision directly affects Nvidia, whose chips are used across both commercial and research applications in China. Though Nvidia has designed specific products to navigate the restrictions, its exposure to shifting US policy continues to raise concerns.
Investor Eileen Burbridge, Founding Partner of Passion Capital, told the BBC that the share price wobble reflects the data centre division: “Not posting results as strong as it was hoping.”
She adds the broader trend remains hard to ignore and the growth has been “unbelievable”, but notes there has been “so much capital that’s gone in that I don’t think it’s unfair to say there’s been maybe too much exuberance or a bit of a bubble”.
H20 and Blackwell chips under review
In July, Nvidia said it would resume selling its H20 processors in China after gaining temporary approval from the White House. The H20 chips were designed to comply with export controls while still serving Chinese customers.
“The AI race is now on.”
Even so, US authorities began reviewing export licences for these chips late last month. While some China-based firms have now received licences, no shipments have been made to date.
Executives also reveal that the US government wants a 15% share of revenue from any licensed H20 sales. This adds further pressure, as Nvidia also seeks approval to export its Blackwell chips, the next generation of AI hardware, to the Chinese market.
The Blackwell series is central to Nvidia’s roadmap, offering increased power for large-scale AI workloads. Approval delays risk cutting off access to the world’s largest semiconductor market, just as AI demand is peaking.
Despite these challenges, Nvidia projects revenue of US$54bn for the current quarter, above Wall Street expectations. The company hit a US$4tn market valuation in July, becoming the most valuable firm globally.
AI infrastructure spending raises questions
Major cloud computing platforms like Amazon Web Services (AWS) and Microsoft Azure are investing heavily in AI data centres and cloud-specific infrastructure. They require chips that can handle large-scale training and inference workloads with speed and efficiency.
Social media companies, meanwhile, are integrating AI to improve content delivery and moderation. Traditional enterprise software providers are also shifting strategies to embed AI across platforms.
Jensen’s message to investors during the earnings call on Wednesday emphasised Nvidia’s pivotal role in these transformations.
The takeaway? Nvidia’s chips don’t just power individual AI applications – they’re becoming the basis for entire digital economies built around AI.
However, some analysts warn that infrastructure investment may be outpacing practical application. There are concerns that hyperscalers and enterprises are purchasing more AI capability than they can immediately deploy or monetise.
For now, capital continues to flow, driven by the promise of AI to reshape cloud services and computing models.
But as export controls, licensing delays and market saturation challenges persist, Nvidia’s ability to maintain momentum remains closely tied to both policy and performance.
