NVIDIA's performance contradicts the AI bubble theory? Analysts: The concern should not be NVIDIA, but the data centers built on debt

Despite Nvidia's performance being widely regarded as an important indicator of the health of the AI industry, several analysts have pointed out that the real risk does not come from the chip manufacturer itself, but from those companies that are heavily indebted to build data centers.

According to an article from Wall Street Insight, the AI chip giant reported revenue and expectations that exceeded market expectations, and its CEO Jensen Huang clearly stated during the earnings call that "the situation we see is completely different," attempting to dispel the AI bubble theory. Nvidia disclosed that its advanced chips have secured $500 billion in orders before 2026, with major clients like Microsoft, Amazon, Google, and Meta clearly planning to accelerate AI chip procurement.

However, on November 20, D.A. Davidson's head of technology research, Gil Luria, warned that concerns about an AI bubble are not Nvidia's problem; the real issue lies with companies that have borrowed heavily to build data centers. He pointed out that data centers are essentially speculative investments that may face liquidation when global capacity reaches saturation in two to three years.

Some analysts noted that Nvidia's impressive performance only proves that spending on AI infrastructure is robust, but does not indicate that the AI economy has truly matured. The more critical indicators are the actual monetization capabilities and customer demand for AI services on platforms like Microsoft and Adobe, rather than just chip sales data.

However, some analysts believe that the demand for AI infrastructure continues to exceed supply, citing Nvidia's $500 billion in orders before 2026 as "just the beginning." Dan Ives of Wedbush Securities stated that this is "a validation moment without an AI bubble, and the AI revolution is still in its early stages."

Strong Chip Demand Cannot Mask Debt Risks

Luria stated that Nvidia's clients have clearly communicated plans to accelerate spending on AI chips, which was confirmed in the earnings report. "Many people will breathe a sigh of relief, but they actually didn't need to worry about Nvidia before the earnings report," he said.

He pointed out that the core issue lies in the financing model of data centers. These facilities, which provide computing power for AI services, are typically owned by specialized operators and large-scale cloud service providers like Microsoft and Google, which are accelerating data center construction through borrowing to meet the growing demand for AI. Luria emphasized:

"Concerns about an AI bubble are not Nvidia's problem, but rather about those companies that are heavily borrowing to build data centers."

He warned that data centers, as speculative investments, may face a cyclical turning point when global operations reach full capacity in two to three years. However, he added, "Nvidia will continue to sell chips regardless."

Disconnection Between Chip Sales and AI Economic Maturity

Billy Toh, head of retail research at CGS International Securities Singapore, clearly pointed out that Nvidia's earnings report is a strong signal of spending on AI infrastructure, but it is not a reliable indicator of whether the AI economy is truly mature. Toh stated:

"To understand the stability of a broader industry, it is more meaningful to focus on the actual adoption and monetization of AI services on platforms like Microsoft, Adobe, and others. Real customer demand and recurring revenue will ultimately confirm whether the AI boom is sustainable."

In addition to concerns about the debt burden on hyperscale cloud service providers, the meager revenues of AI developers like OpenAI relative to their massive expenditures have also left some investors uneasy.

NVIDIA, with its dominant position in advanced chips and chip software, as well as its deeply integrated AI ecosystem, possesses pricing power and profitability demand, and has not been affected by the insufficient revenues of AI companies.

Toh stated:

"Even if many AI startups are struggling, NVIDIA can still sell products to hyperscale cloud service providers, sovereign AI projects, and companies building core infrastructure. This dynamic supports its trillion-dollar market value and is why investors see it as the safest way to gain exposure to AI."

Market Divergence: Infrastructure Boom or the Eve of a Bubble

Rolf Bulk, a stock research analyst at New Street Research, agrees with the view of distinguishing NVIDIA's earnings report from the broader AI market, but he believes this outcome can still alleviate concerns about an AI bubble in the short term.

"This indicates that hyperscale cloud service providers expect demand for computing power to continue to grow strongly in 2026 and beyond. Of course, these GPUs need to be continuously utilized to generate returns for hyperscale cloud service providers and AI companies, which is exactly their bet."

Bulk believes these bets may pay off, as the AI market still has ample long-term growth potential.

"Demand for AI infrastructure continues to exceed available capacity, with OpenAI, Anthropic, Amazon, Google, and others pointing out that customer demand exceeds their ability to provide the necessary computing power."

AI bulls have viewed NVIDIA's earnings report as yet another bullish signal. Ray Wang, chairman of Constellation Research and co-founder of the AI Forum, cited NVIDIA's $500 billion in advanced chip orders before 2026:

"This is not a bubble; it's just the beginning."

Dan Ives also stated, "There is only one chip in the world driving the AI revolution, and that is NVIDIA."

Notably, Jensen Huang himself responded to the AI bubble theory during Wednesday's earnings call: "People have been talking about the AI bubble, but from our perspective, what we see is completely different."