Nvidia reported strong earnings and forecasts on Wednesday, which analysts saw as a clear sign that spending on AI infrastructure will continue. However, it is less clear whether this result will dispel concerns about an AI bubble in the market. Concerns have grown in recent months that big tech companies’ huge investments in AI could outweigh their realistic returns, with some industry observers and analysts predicting an AI bubble. Nvidia’s profits are widely seen as a key indicator of the health of the AI industry, but some analysts warn that its performance doesn’t tell the whole story. “I think a lot of people will be relieved, but they didn’t have to worry about Nvidia going (to earnings) anyway,” Gil Luria, head of technology research at DA Davidson, told CNBC on Thursday. Concerns (about the AI bubble) are not Nvidia’s problem. The concern is that companies are raising large amounts of debt to build data centers. DA Davidson Gil Luria Head of Technology Research The analyst noted that Nvidia’s customers, including Microsoft, Amazon, Google and Meta, have already communicated plans to accelerate spending on AI chips, and that is reflected in Nvidia’s results. The strong demand has also provided a tailwind for Nvidia-related chip stocks, with major Asian suppliers rising on Thursday. However, Luria said, “The concern[about the AI bubble]is not Nvidia’s problem. The concern is that companies are raising large amounts of debt to build data centers.” Nvidia’s AI chips, also known as graphics processing units, are used in data centers to provide the computing power needed to train and run AI services. These data centers are often owned by specialized operators or large technology companies such as Microsoft and Google, known as hyperscalers. As these companies prepare to meet growing AI demand, they are borrowing money to fund data center deployments. “(NVIDIA’s revenue) certainly alleviates concerns about Nvidia, but that doesn’t mean we don’t need to be on the lookout for companies borrowing and lending to build data centers,” Luria said. The analyst said data centers are inherently speculative investments that could be in trouble two to three years from now when the world reaches full capacity and the cycle reverses. Still, he added, “NVIDIA will continue to sell chips in some form.” AI Chips and the Promise of AI Other analysts who spoke to CNBC drew a clear line between AI chip companies like Nvidia and downstream companies such as hyperscalers and companies like Chat-GPT maker OpenAI that are actually building AI models. “While NVIDIA’s earnings are a strong signal of spending on AI infrastructure, they are not a reliable measure of whether the AI economy has truly matured across the industry,” said Billy To, regional head of retail research at CGS International Securities Singapore. “To understand broader industry stability, it is more meaningful to look at the actual adoption and monetization of AI services at companies like Microsoft, Adobe, and other enterprise platforms, where real customer demand and recurring revenue will ultimately confirm whether the AI boom is sustainable,” he added. In addition to concerns about hyperscalers taking on debt, the low returns of AI developers such as OpenAI compared to their high spending is also a cause for concern for some investors. The lack of revenue for AI companies is not being felt by Nvidia, which dominates advanced chips and chip software and is deeply integrated into the overall AI ecosystem, giving it pricing power and profitable demand. “Even though many AI startups are struggling, Nvidia is still selling to hyperscalers, sovereign AI initiatives, and companies building core infrastructure,” To said. “This dynamic helps justify the company’s multi-trillion-dollar market cap and why investors see it as the safest way to gain exposure to AI,” he said, but said that protection will wane as AI progresses slower. Parade of bulls Rolf Bulk, equity research analyst at New Street Research, agreed with the distinction between Nvidia’s earnings and the broader AI market. But he still sees Nvidia’s performance as a factor that could calm fears of an AI bubble in the short term. “This shows hyperscalers expect computing demand to continue to grow strongly in 2026 and beyond,” he told CNBC. “Of course, to make money for hyperscalers and AI companies, these GPUs need to continue to be leveraged well, and that’s their bet.” According to Bulk, those bets can pay off, and there is plenty of room for long-term growth in the AI market. “Demand for AI infrastructure consistently exceeds available capacity, with OpenAI, Anthropic, Amazon, Google, and others all pointing out that customer demand exceeds their ability to provide the computing they need,” he said. Meanwhile, strong believers in AI, who have already dismissed bubble concerns, are likely to see Nvidia’s gains as a further bullish sign for the industry as a whole. “This is not a bubble. It’s just the beginning,” said Ray Wang, chairman of Constellation Research and co-founder of AI Forum, citing $500 billion in reservations for Nvidia’s advanced chips by 2026. Dan Ives of Wedbush Securities echoed that sentiment in an email to CNBC, saying Nvidia’s performance “isn’t an AI bubble; it’s an AI forum. “This was a moment that confirmed that we were in the early stages of a revolution,” he said. “There’s one chip in the world that’s powering the AI revolution, and that’s Nvidia,” Ives added. Nvidia CEO Jensen Huang himself dispelled concerns about AI during Wednesday’s earnings call. “There’s a lot of talk about an AI bubble,” he says. “From our perspective, we see something completely different.” — CNBC’s Martin Soong contributed to this report
