Nvidia’s US $96 Billion Quarter Shows AI Demand Is Still Outrunning Supply

Nvidia has spent the past three years making extraordinary growth look increasingly routine, and its latest quarter was no exception. The company more than doubled revenue from the same period last year to US$96 billion, with US$89 billion generated by its data center business, while guiding the October quarter to US$108 billion, ahead of estimates.

According to Josh Gilbert, etoro Lead Analyst, APAC & Middle East, the results underline the scale of AI infrastructure demand and suggest Nvidia’s immediate challenge is no longer finding customers, but securing enough supply to meet them.

“For a company of this size to still be compounding at this rate is staggering. Blackwell Ultra drove the quarter, while Vera Rubin is already in full production. Nvidia is bringing out its next generation before customers can get enough of the last one,” Gilbert said.

Nvidia’s future supply commitments more than doubled during the quarter, rising from US$119 billion to US$279 billion, with much of that increase focused on securing memory supply. Gilbert said this illustrates how the bottleneck in AI infrastructure is shifting from customer demand to the availability of critical components.

That pressure is also reflected in Nvidia’s margin outlook, which now bottoms at around 71% to 72% before recovering as price increases flow through.

Free cash flow declined to US$21 billion, from approximately US$48 billion three months earlier, but Gilbert said the fall needs to be viewed in the context of Nvidia’s investment in securing future supply and building inventory for Rubin.

“This is the cost of getting hold of parts, not a business going backwards,” Gilbert said.

Nvidia CEO Jensen Huang indicated demand is running close to 100% growth, while the company is only committing to around 70% because of what it can realistically supply. For Gilbert, this is an important distinction as investors assess where Nvidia and the wider AI cycle go next.

“This is a company turning away business, not a company with a demand problem. Cycles roll over when supply catches up and companies start discounting to shift stock. Nvidia is doing the opposite and putting prices up by more than 15%,” Gilbert said.

Big Tech continues to underpin much of Nvidia’s growth, with hyperscalers accounting for around 55% of data center revenue during the quarter. However, AI cloud providers, industrial companies and enterprise customers now account for the remaining 45%, giving Nvidia another growth engine alongside the major technology platforms.

Gilbert added that Nvidia’s position has also evolved beyond that of a traditional semiconductor company, with its technology increasingly forming a core infrastructure layer for the global AI build-out.

“Amazon has committed to another 2 million GPUs over the next two years while building its own chips at the same time. That tells you the competition is real, but nobody has found a way around Nvidia yet,” Gilbert said.

Looking ahead, Gilbert expects the AI investment cycle to continue, while opportunities broaden across the companies supplying the infrastructure required to support it.

“The AI boom still has plenty of runway, but the list of winners is going to get longer, particularly at the supply end. The question is no longer whether Nvidia can sell the chips, but whether the rest of the industry can build fast enough around them,” Gilbert concluded.