Nvidia CEO Jensen Huang told investors on Sept. 10 that demand for artificial-intelligence computing is expanding beyond the largest cloud companies, reinforcing the chipmaker’s unusual forecast for roughly 70% revenue growth in fiscal 2028.
Huang made the case at the Goldman Sachs Communacopia + Technology Conference, two weeks after Nvidia disclosed the long-range outlook with its second-quarter results. The company’s official investor notice confirms the appearance, while its earnings-call transcript records the forecast as supply-constrained rather than a ceiling on customer demand.
The forecast followed a quarter in which Nvidia reported $96.2 billion in revenue, more than double the result a year earlier, while data-center sales reached $89 billion. Those results establish the scale behind Huang’s argument, but they do not eliminate the risk embedded in projecting another year of rapid expansion.
The demand argument extends beyond hyperscalers
In remarks reported by TechCrunch, Huang argued that investors can miss much of the market when they focus only on Amazon, Microsoft and Google. Nvidia also sells computing systems and networking to AI laboratories, startups, governments, regional cloud providers and enterprises.
That breadth matters because Nvidia is defending its position as major customers develop custom chips. Huang’s response is that Nvidia is selling a complete computing platform—processors, networking, systems and software—rather than a standalone component. The distinction is central to the company’s claim that demand can remain strong even as alternative accelerators enter the market.
For Nvidia, widening the customer mix also reduces the importance of any single buyer. For customers, however, the same expansion can intensify competition for systems, networking equipment, electrical capacity and specialized engineering talent.
Supply remains the constraint
Nvidia’s official earnings transcript says its fiscal 2028 growth expectation assumes constrained supply. An independent report from The Associated Press also identified memory availability as a limit on how quickly Nvidia can convert orders into revenue.
Nvidia’s quarterly filing puts additional scale around that constraint: supply and capacity commitments rose from $119 billion in the prior quarter to $279 billion as of July 26. The commitments reflect management’s expectations, but they also increase the operational and financial stakes if demand, delivery schedules or component availability change.
The forecast remains forward-looking, not a guaranteed result. Nvidia faces execution risk across manufacturing, memory, advanced packaging and power-hungry data-center deployments. It also depends on customers continuing to find economic value in increasingly large AI investments.
Why it matters for enterprise planning
For business and technology leaders, the message is that AI capacity should be treated as a multiyear infrastructure decision rather than a short experimental budget. If Nvidia’s supply-constrained outlook holds, access to computing, power and implementation expertise may continue to shape deployment schedules and costs over the next planning cycle.
The strategic question is therefore shifting. Companies are moving from deciding whether to test generative AI to deciding which workloads justify scarce computing capacity, which vendors can deliver it and how quickly those systems can produce measurable operating value.
Source note: Nvidia’s conference appearance was announced through its investor-relations site.