MW Nvidia and Oracle are sending similar warning signs about the AI trade
By Britney Nguyen and Christine Ji
Oracle is raising more debt and Nvidia is walking back its OpenAI investment target. Both are signs that the AI trade could be on shaky ground, according to one analyst.
OpenAI, led by CEO Sam Altman, is at the center of financing that has come under increasing scrutiny by investors.
Nvidia and Oracle are at the center of OpenAI's trillion-dollar web of financing, but recent developments at the two companies may be sending cautious signals about the health of the artificial-intelligence trade, according to an analyst.
On Sunday, Oracle $(ORCL)$ announced plans to raise up to $50 billion in 2026 to fund its cloud growth. The money will be raised through a combination of debt and equity to support the infrastructure needs of key customers such as Advanced Micro Devices $(AMD)$, Meta Platforms (META), Nvidia (NVDA), OpenAI, TikTok and xAI.
See more: Oracle amps up its AI bet with a plan to raise as much as $50 billion this year
Meanwhile, the Wall Street Journal reported on Friday that the plan for Nvidia to invest up to $100 billion in the AI startup has "stalled" due to concerns from the chip maker, including some about OpenAI's competitiveness against rival AI model-makers Anthropic and Google. Nvidia said in September that it plans to deploy at least 10 gigawatts' worth of its chips with OpenAI, as well as invest up to $100 billion in the company as it rolls out its AI systems to support data-center and power capacity.
To Richard Windsor, founder of independent research firm Radio Free Mobile, the problem plaguing the two companies is simple enough: "The business model of compute upon which the AI boom is being built is not viable," he wrote on Monday, pointing out that AI infrastructure costs are rising while the revenue per gigawatt that infrastructure providers can earn has remained capped.
OpenAI has multiple agreements to deploy multiple gigawatts' worth of AI infrastructure over the next five years, each gigawatt of which costs $50 billion to build, Windsor said. He added that each gigawatt of capacity has about a five-year lifespan before needing to be replaced.
By his calculations, OpenAI can generate about $10 billion in annual revenue, but "it does not take a genius to figure out that this model leaves no room for profit, paying back debt or equity investors," Windsor said.
The partnerships between OpenAI and AI chip makers, which also include AMD and Broadcom, are predicated on the belief that returns on investment will improve as generations of chips evolve, Windsor said - and that has not been the case so far.
Don't miss: Nvidia's OpenAI investment may not be as big as once hoped - and tech stocks are falling
He pointed to OpenAI, which he said over the last three years has been able to make about $10 billion per gigawatt of compute, but no more than that despite technological improvements.
Investors have picked up on this dynamic and have treated Oracle's debt and equity with increasing skepticism. While shares of Oracle originally soared upon the announcement of its agreement with OpenAI, they've since fallen roughly 50% from their September 2025 peak. Additionally, the issuance of more equity this year will dilute existing shareholders.
Oracle announced on Sunday its plans for a single issuance of investment-grade bonds early this year, as well as issuances of convertible securities and common stock. Even though the company's announcement confirmed Oracle's commitment to maintaining its investment-grade credit rating, Windsor pointed out that the market is still anticipating higher levels of risk on debt for funding data centers.
As a result, Oracle's five-year credit-default swaps, or the cost of insurance on Oracle's debt, are nearly three times above their historic average, Windsor wrote. After the OpenAI deal, they've risen from their 0.05% long-term average to 0.14%.
Nvidia's recently reported investment jitters are another sign to Windsor that many high-profile AI megadeals are no more than nonbinding "statements of intent," meaning that OpenAI isn't guaranteed to secure the funding necessary to pay Oracle for its services. In a quarterly filing in November, Nvidia said: "There is no assurance that we will enter into definitive agreements with respect to the OpenAI opportunity or other potential investments, or that any investment will be completed on expected terms, if at all."
Nvidia CEO Jensen Huang said over the weekend that the chip maker is keeping plans to invest "a great deal of money" in the AI startup. But when asked if that investment could surpass $100 billion, he suggested it wouldn't be in that vicinity.
Although increasingly efficient chips are helping to improve the compute capacity of a single gigawatt, that's making the price of compute tokens - or the units of information processed by large language models - fall. AI model makers like OpenAI determine pricing for customers based on how many tokens are processed, and widespread AI adoption depends on these prices falling. As a result, revenue per gigawatt has remained static.
But the companies still have to make money, Windsor said, or else "the whole proposition will grind to a halt while the market figures out the economic way forward."
To Windsor, that dynamic could prompt a major slowdown in AI investments - and some AI companies would be left for dead. It could also lead to a slowdown in compute supply, which would drive AI-model-usage prices higher. AI companies that are left over could make a profit from selling more expensive compute tokens, he concluded.
"It is from this more stable foundation that the AI industry can develop," Windsor said.
For now, Windsor said, the more money invested in AI means compute supply will continue to rise, while the price falls.
"Eventually, the money will dry up as losing money on trillions of tokens gets pretty annoying after a while, meaning that the cure for low prices of compute is the low price of compute, meaning that the lower it goes, the more likely it is to trigger a reset," Windsor said.
Read on: Big Tech needs a staggering $1.5 trillion to fund the AI boom. This is the complex playbook it's using to get it.
-Britney Nguyen -Christine Ji
This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
February 02, 2026 11:23 ET (16:23 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments