The Artificial Intelligence landscape is suddenly looking quite shaky. In just a few weeks, 2 biggest backers of the AI revolution have hit the brakes. At first, it was Microsoft that quietly began distancing itself from OpenAI, pushing to have internal AI development. Now, Jensen Huang, the CEO of NVIDIA, has confirmed that the chipmaker is pulling back from making further investments in both Anthropic and OpenAI.
For the industry that is built on endless funding rounds and sky-high valuations, all these moves are more than just some realignments. They can, as experts believe, be the first cracks in the bubble that analysts long warned would burst sooner.
NVIDIA makes a quiet exit from OpenAI days after Microsoft
The entire shift started as finally Microsoft steps back. While the company, in its statement (late February), insisted that its partnership with OpenAI has been strong and central, reality on the ground tells a different story. It was recently admitted by Microsoft AI Chief Mustafa Suleyman that the company is pursuing its “true AI self-sufficiency.” He also stated that they are now actively reducing their reliance upon OpenAI while building some internal models like the MA-1 preview.
This pivot comes right after Microsoft poured in approx. $13 billion into OpenAI. Despite it, now they find themselves competing directly with a startup that they helped to fund. Not to mention, Wall Street has also noticed that Microsoft shares tumbled as the investors questioned massive AI spending wisdom, without clear returns.
The retreat of NVIDIA is more dramatic now as experts believe OpenAI finally at a dead end. While speaking at the Morgan Stanley Technology conference, Huang revealed that the recent $30 billion investment of NVIDIA in OpenAI—that is part of a huge $100 billion funding round—would likely be its last one. The organization is also ending the financial support for Anthropic after a $10 billion commitment. The official explanation of Huang points towards upcoming IPOs closing the investment window. However, the industry observers seem not to be buying it.
The tensions now run deeper. Just a few weeks ago, Anthropic was blacklisted by the Trump administration after it refused to let its models power the autonomous weapons. Just a few hours later, OpenAI struck a deal with the Pentagon—a move (mendacious) by the CEO of Anthropic.
The reaction of the public to this entire incident has been swift—downloads for the Claude app by Anthropic skyrocketed past ChatGPT on the U.S. App Store (Apple). Now NVIDIA has found itself awkwardly holding stakes within 2 companies that are pulling in 2 opposite directions—one that is cozying up to the military, while another is taking an ethical stand against the same.
What does this pullback actually mean, and why the AI bubble might finally burst?
All these incidents are not some corporate decisions but a deeper rot within the AI funding model. The logic of circular spending, which propped up valuations, is now collapsing under its weight. When NVIDIA first promised $100 billion to OpenAI last September, a professor from MIT Sloan dismissed it like a “kind of a wash.” It means, NVIDIA invests within OpenAI, then OpenAI uses money to buy NVIDIA chips. Such an incestuous cycle worked only for the time being. Now things are changing, and the math is not adding up.
Some major tech companies have projected to spend approx $600 billion on the AI infrastructure in just 2026. About 75% of it is being tied directly to Artificial Intelligence. Despite it, the returns have been elusive. The capital expenses of Microsoft jumped by 2/3rd YoY, while Azure growth slowed down—investors have punished the stock accordingly. Above all, OpenAI has pledged $1.4 trillion towards infrastructure, despite generating 0 free cash flow. It is totally reliant on the outside investors who might soon lose appetite.
The numbers paint an even bleaker picture and show OpenAI is in the red. As per a MIT study, 5% enterprise AI projects have made it to production with a positive financial impact. Other 95% have shown no value added. An NBER survey revealed 90% of the executive saw no change in their employment or productivity because of AI. Microsoft made some cut quotas on the AI products after its sales force reported a lack of customer demand.
All of it is not dot-com bubble redux, but valuations that have not reached insane levels yet. However, the dynamics feel familiar—massive capital pouring into unproven techs, companies making spends because they fear being left behind, instead of it making economic sense, and also, investors asking uncomfortable questions about when profits would materialize actually.
The exit of NVIDIA now signals that even ultimate insiders see writing on the wall. As an analyst put it, the IPO explanation of Huang, “raises more questions than it answers.” What looks more probable is the made strategic exit from a situation that got expensive, complicated or probably catastrophic. So, the AI party might not be actually completely over, but hosts of them are definitely heading for the door.
