Oracle opened the day higher on plans to raise $50 billion for AI infrastructure. It closed lower after reminding investors who that infrastructure is for.

The company said Sunday night that it planned to raise up to $50 billion in debt and equity during the 2026 calendar year to fund additional data center capacity for its cloud customers. The market’s initial reaction was favorable, with Oracle shares rising about 2% in early trading, as investors took the announcement as confirmation that demand for AI infrastructure remained strong and contracted. The market seemed to feel confident that Oracle actually had a plan to address its roughly $100 billion debt load.

As Oracle’s price wavered slightly at $168, its social media team filled out the narrative.

“The Nvidia-OpenAI deal has zero impact on our financial relationship with OpenAI,” the company posted on X. “We remain highly confident in OpenAI’s ability to raise funds and meet its commitments.”

The market’s reaction was swift and brutal. Rather than projecting the confidence it intended, the post served as a negative signal for investors already angsty about Oracle’s debt.

“This is literally bank-run language,” venture capitalist Alex Kolicich wrote on X.

Within minutes of the post, Oracle’s stock began to tumble, closing down 2.79% at $160.06. By trying to prove its independence, Oracle instead reminded everyone just how exposed it is, and how far it is sticking its neck out.

To be fair, Oracle’s five-year credit default swaps also fell 17%, a sign that investors feel more confident in the company’s ability to manage its debt and avoid a credit downgrade. The question is why equities tumbled as well.

Microsoft and Nvidia have both seen stock movements downward in relation to their OpenAI exposure as investors send the message that they’re bullish about AI but not necessarily the ChatGPT-maker.

Nvidia had been expected to make a major equity investment in OpenAI, potentially committing up to $100 billion as part of OpenAI’s next funding round. But reporting over the weekend indicated that the deal has stalled and was never in fact binding, with CEO Jensen Huang adding credence to the reporting by emphasizing the funding was “never a commitment,” only reaching the letter of intent stage. Every investment by Nvidia in OpenAI would be decided in stages, he said.

Huang reiterated that Nvidia would “absolutely be involved” in OpenAI’s funding round, in what could be Nvidia’s “largest investment,” but nothing to the tune of $100 billion. Microsoft saw a $360 billion stock wipeout last week as investors blanched at its level of AI spend. Even though Microsoft beat expectations considerably, the selloff seemed to punish its disclosure that 45% of its $625 billion commercial backlog—nearly $250 billion—was tied to OpenAI. Meanwhile, Microsoft’s revenue growth from its AI cloud compute was stalling, a sign that perhaps there wouldn’t be the cliff of revenue needed to finance Microsoft’s own debts after all.

How to price a private company in public markets

The evidence is mounting that OpenAI, once treated as an engine for growth, is now being priced in like a source of inherent risk. For months, investors rallied on any announcement of OpenAI and a big number: bigger data centers, bigger chip orders, bigger contracts. Amazon, Microsoft, Google, and Nvidia all got big boosts based on the simple assumption that if OpenAI needed it, demand must be worth funding. Even though detractors would complain about the deals’ “circular funding,” the prevailing assumption was that everyone would get paid eventually, either through the force of their own value inflation or through revenue proper.

That assumption is starting to crack. The problem is that OpenAI, a private company, is dealing with members of the Magnificent Seven without any of the disclosures that markets rely on to price risk. And investors are starting to get spooked.

OpenAI has already committed to roughly $1.4 trillion in spending on compute, power, and infrastructure, while generating just over $20 billion in annualized revenue. The idea is that the gap will be bridged by continuous fundraising; larger rounds, at larger valuations, from an increasingly narrow pool of investors that also benefit from OpenAI’s growth. But now that model is being scrutinized with high sensitivity.

Nvidia has only added to that unease. When Huang emphasized that Nvidia’s mammoth investment in OpenAI was nonbinding, it raised a question that extends far beyond Nvidia: If OpenAI’s financing is contingent, or delayed, what happens to the infrastructure that has already been built to match the supposed demand?

That question matters the most by far to companies like Oracle or Microsoft that have already taken on leverage to meet that exact demand.

Oracle is not waiting to see whether OpenAI raises its next round. It has already borrowed, already built, and already committed to spending years ahead of cash flow, and if it doesn’t work out it could be caught holding the hot potato. That’s why, when a company feels compelled to publicly assert that a counterparty can “raise funds and meet its commitments,” investors hear desperation.