NVIDIA $100 billion investment in OpenAI and OpenAI's deal with Oracle: Is the AI cycle a Bubble or a Boom?

In August 2025, during their Q1 financial results, Oracle announced $455 billion in remaining performance obligations (RPOs) following a deal with OpenAI, despite having only $57 billion in annual revenue. This would push Oracle to the 14th-largest company in the world and make its founder, Larry Ellison, the richest man in the world. Oracle is involved in the artificial intelligence market through its data centers, which store data and run computing systems using NVIDIA chips to support and run bots, such as OpenAI's ChatGPT. That is an oversimplified version of the cycle we're seeing in the AI market today.

The Loop

OpenAI requires excess data centers to have the ability to run new LLMs like ChatGPT 5.0, and they rent them out from companies like Oracle. This led to the $300 billion deal that was mentioned earlier. When Oracle starts building these data centers, it will then reinvest that $300 billion earned through its previous deal into NVIDIA to buy the GPUs required to run these LLMs or AI bots. As of recently, the last piece of the cycle, OpenAI, has been put into question. Where does their money to invest in companies like Oracle come from? OpenAI has been consistently looking for ways to finance its trillion-dollar AI plans. Sam Altman, the owner, was even quoted as saying, "There's always a lot of focus on technological innovation. But what really drives a lot of progress is when people also figure out how to innovate on the financial

model... I don't think we've figured out yet the final form of what financing for computing looks like."

Vendor Financing

The solution has now finally been found through NVIDIA's new $100 billion investment into OpenAI. This completes the cycle repeated throughout the industry, or called vendor financing. Vendor financing is when a company sells products but helps the buyer pay for them, often by lending the money itself. For example, NVIDIA sold $40 billion worth of chips to Oracle, but instead of Oracle paying that $40 billion in full up front, NVIDIA offers them installments over multiple years or occasionally with a low-interest loan. This lets Oracle obtain many more chips costing hundreds of billions with almost no upfront expense while only having $57 billion in revenue. Oracle can then utilize the NVIDIA chips it acquired at a low cost to lease its compute capacity to OpenAI, which in turn provides Oracle with the funding to repay NVIDIA. This pattern involves the customer base finding itself financed by the very companies from which it purchases goods. This has not only been shown through the cycle demonstrated earlier between NVIDIA, Oracle, and OpenAI, but also a multitude of different companies funded by NVIDIA. A great example of this of the recent $6.3 billion investment from NVIDIA into CoreWeave, only for CoreWeave to spend that same money on new NVIDIA chips. But because of things like salary and cost of construction, the money received by NVIDIA is not equal to their initial investment. OpenAI has expenses like salaries they must pay, diluting their investment in Oracle. Oracle, when making data centers, must also spend money on land and construction, other than purely the chips they buy from NVIDIA, so in the long run, NVIDIA should be losing billions in these cycles on paper. That is where you and I come in, the investors. As a result of these continued announcements and steady growth throughout these companies, consumer incentive to invest is extremely high, leading to extreme amounts of capital flowing back into every company involved in the cycle, recouping NVIDIA for the partial loss they receive on these large investments.

What could happen?

This repeated cycle can end both badly and well. The ideal situation is that the cycle will continue to repeat itself, fueling artificial intelligence innovation, till eventually one of the top AI companies like OpenAI creates AGI or artificial general intelligence, where AI is better

than a human at every foreseeable task. This would then lead to the profits being so high that the cycle would all pay for itself. The alternative is that the cycle will break once one of the companies cannot make the 100's of billions of dollars in payments, the next company in the cycle can not fund itself, leading to the entire cycle breaking down. OpenAI currently doesn't make any money and is running at a loss, so if investor funding comes to a halt, it would have to default on its payments to Oracle, destroying the cycle and crashing the international market. Both possibilities are entirely plausible, and talk of a so-called AI bubble could lead to a legitimate crash, or we could see one of the biggest booms in not only the stock market but human innovation altogether.