MUMBAI: Nvidia is putting its money where its chips are. CEO Jensen Huang has pushed back against criticism that the chipmaker is financing AI companies only for that money to circle back into Nvidia’s own chips, arguing that the economics of frontier AI are fundamentally different from those of earlier technology businesses.
Speaking to CNBC’s Jim Cramer on Mad Money, Huang said frontier AI companies require billions of dollars just to get started and potentially tens of billions more before they can reach profitability. Nvidia’s investments, he argued, are therefore a response to the extraordinary capital demands of building AI infrastructure rather than an effort to manufacture demand for its processors.
“I think they’re missing a very big point,” Huang said, defending Nvidia’s growing role as both an investor and strategic partner to AI companies. He said the chipmaker wants leading AI businesses to build their technology ecosystems around Nvidia’s platform while helping them bridge the financing gap.
Nvidia has invested across several parts of the AI ecosystem, including model developers such as OpenAI and Anthropic, as well as so-called neocloud companies that provide access to Nvidia-powered computing. Its involvement has also expanded into financing arrangements covering AI infrastructure and data centres.
One major commitment centres on a planned compute campus in Ohio, where OpenAI is expected to be the main tenant. Nvidia has committed up to $105 billion in support for the project. Separately, the company has partnered with major Wall Street firms on arrangements designed to mobilise as much as $500 billion for data centre financing.
That scale has fuelled concerns about a potential circular flow of money: Nvidia provides capital to customers, customers use that capital to purchase Nvidia’s chips, and the resulting sales help support Nvidia’s growth.
Huang rejected that comparison, saying frontier AI companies currently lack the financial track records and investment-grade status needed to raise capital cheaply on their own. Nvidia, he argued, can help fill that gap as these companies scale their AI services and move towards profitability.
He also played down concerns about what happens if one of Nvidia’s backed AI companies runs into financial trouble. According to Huang, the computing infrastructure being deployed is not tied to a single customer or application and can be redeployed elsewhere, giving Nvidia flexibility if individual projects or businesses falter.
The chipmaker’s latest financial results have also given Huang plenty of fuel for his optimism. Nvidia reported $96.2 billion in quarterly revenue for the second quarter of fiscal 2027, more than double the figure recorded a year earlier.
Its data centre business was the biggest engine, with revenue climbing 117 per cent to $89 billion. Nvidia has also projected approximately 70 per cent revenue growth for fiscal 2028, reinforcing expectations that AI infrastructure spending remains far from cooling off.
Nvidia shares rose around 4 per cent in extended trading following the results. Yet despite the strong performance, the stock was up roughly 12 per cent for the year, reflecting continuing investor questions about how long the broader AI boom can sustain its current pace.
Huang remains confident that demand is being driven by genuine adoption across hyperscalers, neoclouds, sovereign clouds and enterprises, rather than financial engineering. He also argued that AI infrastructure is increasingly becoming productive infrastructure because Nvidia’s systems can serve multiple stages and applications of AI.
For Huang, then, Nvidia’s investments are less about creating a demand loop and more about helping build the very ecosystem that could keep AI spending growing. Whether investors see that as strategic foresight or a particularly expensive feedback loop remains one of the bigger questions hanging over the AI trade.