Is the conveyor belt of AI-driven capital about to stop?

The US economy is nothing more than a big bet on artificial intelligence. Morgan Stanley investor Ruchir Sharma Recently noticed This money being pumped into AI investments now represents about 40% of US GDP growth in 2025, and AI companies are responsible for 80% of the growth in US stocks. So how bad is it that the latest major deals between AI giants, agreements that have sent stock prices soaring, look like a snake eating its own tail?

In recent months, Nvidia announced that it would do so Investing $100 billion in OpenAIOpenAI announced that it would do so Paying $300 billion to Oracle For computing power, Oracle announced that it would buy Chips worth $40 billion From Nvidia. It doesn’t take a flowchart to feel like these companies are just moving money between each other. But surely that doesn’t happen… right?

Getting confirmation of this is a little more difficult than you might think.

Is it all a round trip?

Many of these agreements, on the face of it, are mutually beneficial. If everything is at the same level, while these trades may be circular, they should move everything forward. Such deals could lead to a “less capacity-constrained world,” which would allow for faster development of models that can produce higher returns on investment, Rishi Galloria, an analyst at RBC Capital Markets, told Gizmodo.

“The better our models are, the more we can realize a lot of AI use cases that have been put on hold simply because the technology isn’t robust enough to handle them yet,” he said. “If that happens, and it can generate a real (return on investment) for customers… that leads to real cost savings, and potentially new revenue generation opportunities, and that creates net benefits from a GDP perspective.”

So, as long as we continue to make breakthroughs in AI, and as long as these companies figure out how to monetize their products, all should be well. In case that doesn’t happen, though?

“If that doesn’t happen, if there’s not real adoption of AI in organizations, it’s all going to be a flop,” Galloria said.

Trailing, in general, refers to the unethical and usually illegal practice of making trades or transactions to artificially prop up a particular asset or company, making it appear more valuable and more in demand than it actually is. In this case, it will be the technology companies that try to show that they are more valuable than they really are by announcing big deals with each other that move the stock price.

What then might indicate whether this money is achieving anything other than acting as hot air in a rapidly inflating bubble? Galloria said he is watching faster model developments, advances in performance, and overall AI adoption. “If this leads to an incremental change in the way an organization adopts and uses AI, that creates a benefit,” he said.

Whether that is currently happening or not is in the eye of the beholder. OpenAI has certainly shown advancements in its technology. The Sora 2 video generation model has been launched Unleash a new hell on the worldused to generate large quantities of Copyright violations and Wrong information. But the latest version of the company’s flagship model, GPT-5, Disappointing It failed to live up to expectations when it was released in August.

Technology adoption rates are also part of the Rorschach test. The company is proud of that 10% of the world Uses ChatGPT, Nearly 80% of the business world She says she is looking into how to leverage technology. But early adopters don’t find much benefit. According to A Survey from MIT95% of companies that tried to integrate generative AI tools into their operations did not produce any return on investment.

These investments generate returns in the stock market. Which, frankly, doesn’t allay concerns about these companies simply boosting each other’s bottom lines.

Take Oracle, for example. Last month, the cloud provider had Rough quarter With all traditional indicators. It failed to meet revenue and profit expectations, and its net income was flat year over year. However, The stock price rose. The reason: The company’s complete list of remaining performance obligations – financial agreements that will provide revenue that have not yet been met. There, the company showed a tremendous amount of growth, up 359% from the previous year, with an expected $455 billion turnover.

This money is not real yet. Neither is he The growth that the company promisedClaiming that Oracle Cloud Infrastructure revenues will grow from less than $20 billion to nearly $150 billion before the early 2030s. But all of that was enough for investors to raise Oracle’s stock price enough to push CEO Larry Ellison into the market. First place on the list of the richest people in the worldbriefly jumped Elon Musk.

OpenAI is either the nexus or the void at the center

Most of this promised revenue will come from OpenAI, which has committed to purchasing $300 billion worth of computing power from the company over five years. The clock on this decade doesn’t start until 2027, but assuming that actually happens, it will be one Biggest cloud computing deals In history.

It’s also one of the most unexpected, just based on the current situation of the companies involved. And in order to deliver the compute that OpenAI promised, Oracle will do just that It is said It needs to generate 4.5 gigawatts of power, equivalent to the power of two Hoover dams. On the other side of the deal, OpenAI would have to pay about $60 billion annually to fit the bill for the agreement. currently It generates about $10 billion in revenueWhich, statistically speaking, is less than $60 billion.

You can see a similar pie shape for OpenAI’s recent deal with Nvidia’s competitor AMD as well. The exact details of the agreement were not reported, but the chip manufacturer AMD It is expected to generate tens of billions of dollars Over the next half-decade as it sells its AI chips to OpenAI. As part of the agreement, OpenAI will receive a block of shares in AMD, with options to buy up to 10% of the company. Fortunately for OpenAI, there’s no better time to pick up some AMD stock before a big AI-related deal is announced. Company The stock price rose by approximately 35% After this announcement.

With these two recent deals, OpenAI has agreed to do just that Computing deals worth over $1 trillion So far this year. That’s a lot of money for any company to spend, but it’s a particularly large amount for a company that remains private and reports only $10 billion in expected revenue Until 2025. Even in the latest funding rounds, the company as a whole is currently valued at around $500 billion.

Most of these deals have contingencies attached. For example, Nvidia’s investment in OpenAI is not actually $100 billion, but an initial $10 billion per 1 gigawatt of data center capacity with the potential for $100 billion if 10 gigawatts is eventually achieved. But stock prices and valuations certainly treat these deals as if they were fixed. And OpenAI seems to work this way too. The company claims it will increase its revenue more than 10-fold in the next few years, and the projects it will achieve $129 billion annually by 2029.

Capital conveyor belts

These kind of inflated potential revenue numbers are the thing that makes some people think of the dot-com bubble of the early 2000s, where we saw… Companies like Commerce One Obtaining a $21 billion valuation despite having almost no revenue. But Peter Atwater, an assistant professor of economics at the University of William & Mary and president of consulting firm Financial Insyghts, sees a different reflection in the AI ​​bubble: a housing market collapse.

“What we saw at the top of the mortgage market was all these conveyor belts of capital, money flowing from one end to the other end to the other end,” he told Gizmodo. “And what you started to see was that there were multiple points of relationship such that any participant in the system was then dependent on every other conveyor belt in the system working simultaneously to keep the system going.” “In many ways, we are seeing the same developing network of capital flows across the AI ​​space.”

This creates some obvious problems. Circular trades, in theory, are wheels that move everything forward, they should keep turning. If any one of them stops, the whole thing stops, because they are all so interconnected that failure cannot truly be isolated.

The types of big, conditional deals that have been dominating AI headlines aren’t that different from what was happening in the mortgage industry in 2007, where certain financial obligations required mortgages to meet certain conditions, Atwater said.

“In the madness of a bubble, everyone overcommits,” he explained. “The purpose of overcommitting is to demand what you think will be a very scarce commodity in the future. So, you have buyers overcommitting, and sellers agreeing to oversupply as a result.” “What we find over and over again is that commitments are some of the first to be made as soon as circumstances change, as soon as trust begins to decline.”

At the moment, there is preparation for those commitments. This cannot be guaranteed in the future if all these promised returns on investment are not achieved. Atwater said the market requires credit markets to be prepared to continue offering huge sums of money to cover concluded agreements, equity markets to value these transactions at an “extraordinary multiple,” and suppliers able to deliver the promised products. There is no guarantee that all of these factors will hold up.

Math is already very difficult. As did technology commentator Ed Zitron He pointed outMajor companies like Microsoft, Meta, Tesla, Amazon, and Google have invested about $560 billion in AI infrastructure over the past two years. They have generated a total of $35 billion in AI-related revenue. OpenAI’s liabilities are even larger, with lower returns.

The company’s development and expansion of its services will depend largely on large data center projects, which will require the same amount of energy to operate. New York City and San Diego combined– The energy that Currently not even available. And again, there’s no guarantee that the final product, once all that energy is spent and the data centers are built, will actually generate revenue.

“Ultimately, if you don’t have a consumer for the product, there’s no space for AI because these companies can’t keep doing it for nothing. Listening to a lot of the calls in the last couple of weeks, there’s a clear open question about how these companies are going to make money from this,” Atwater said.

For now, everyone sees green, the eternal fountain of hope. As long as this is the case, no one will ask where the revenue comes from. “Right now, the AI ​​sector operates with an eternal mindset,” Atwater said. “They act as if they have a very long period of time in which they can figure it out and make money.” “As long as confidence is high, this entire ecosystem can deliver fantasy. When confidence is low, they are expected to deliver real performance in a very short time frame.”

Unfortunately, if that happens, these companies will not alone bear the brunt of failure. “You have to look at this as a larger ecosystem. To talk about AI today, that means we have to talk about the credit market, we have to talk about the credit market. Wall Street and AI are one beast,” Atwater said, warning that a very small number of companies currently have a significant grip on the entire US economy.

Many investors are congregating in the AI ​​space, fearful of missing out on a market that looks like it can only go up. But few look at why these valuations and stock prices keep rising, and show little curiosity about what would happen if all that money were diverted, artificially inflating the actual value of the companies they are betting on.

“Why,” Atwater said, “is the last question to be asked in a bull market.”