🚨 THE SIGNS THAT THE AI BOOM IS BUILT ON FAKE REVENUE ARE NOW HERE. Here is what is really happening 🧵

🚨 THE SIGNS THAT THE AI BOOM IS BUILT ON FAKE REVENUE ARE NOW HERE. Here is what is really happening 🧵

Yesterday, Oracle fell 5.5% to $135.69. The FT had reported that OpenAI told investors its annualized revenue was approaching $50 billion at the end of September. The figure reported last month was about $70 billion.

So why would Oracle fall on a report about OpenAI? Because about half of Oracle’s $664 billion backlog is estimated to come from OpenAI. Oracle does not disclose the exact number. Backlog is the money customers have signed contracts to pay in the future. Companies report it to investors as future revenue. So roughly $300 billion of Oracle’s future revenue is a promise from one startup.

But How does OpenAI pay for that promise? From its own revenue, and from money it raises from investors. If its revenue is lower than investors thought, it has less to pay with and a harder time raising more money at a high valuation. It is in talks to raise at about $1.4 trillion.

And this is where the loop shows up. Some of those investors are the same companies selling OpenAI the servers. So Oracle’s future revenue depends on OpenAI’s revenue, and OpenAI’s ability to pay depends on investors who are also sellers. This is a circular loop.

Here is how the loop works. The numbers are made up. A tech giant invests $10 billion in an AI startup. The startup signs a contract to spend $10 billion renting servers from that same giant. The giant reports the contract as future revenue. When the startup uses the servers, the giant records the payment as cloud revenue. The money went out as an investment and came back as sales.

And then it repeats. The sales make the giant’s business look bigger and push its stock price up. The startup shows its fast growth to investors and raises more money. That money pays for more servers. That is why one weak revenue number was enough to hit Oracle.

This is not just an example. Look at what happened in Q1 2026. Amazon invested $50 billion in OpenAI. In the same quarter, OpenAI added $100 billion to its commitment to spend on AWS. Its total AWS commitment is now $138 billion. Amazon puts in $50 billion. OpenAI promises $138 billion back.

Anthropic follows the same pattern. Amazon put $5 billion into Anthropic. Anthropic committed more than $100 billion to AWS. It also committed about $200 billion to Google Cloud. And Nvidia took a $30 billion stake in OpenAI, which buys Nvidia chips.

The latest earnings reports show how big this loop has become. Microsoft’s backlog is $678 billion, up 84% in a year. Excluding OpenAI, it grew only 25%, per Microsoft’s CFO Amy Hood. Without OpenAI, the growth is much smaller. Google Cloud’s backlog went from $106 billion to $514 billion in a year. That is an increase of $408 billion. Anthropic’s $200 billion commitment alone is about half of that.

AWS’s backlog grew by $132 billion in a single quarter. In the same quarter, Anthropic added more than $100 billion to its AWS commitment. Add it all up. OpenAI and Anthropic make up about half of the cloud backlog held by Microsoft, Oracle, Google and Amazon, per The Information. That backlog is now over $2.3 trillion. So the backlog is growing because of two customers who lose money or barely make a profit.

How big are these promises compared to what the startups earn? OpenAI has committed $300 billion to Oracle and $138 billion to AWS. That is at least $438 billion to two companies. OpenAI’s annualized revenue is about $50 billion. Its operating loss in Q2 alone was $12.3 billion. That is why a $20 billion gap in OpenAI’s revenue number mattered so much to Oracle.

Back to Oracle. It has to build first and get paid later. The OpenAI contract is reported at $300 billion over five years. That is about $60 billion a year, starting around 2027. Last quarter Oracle spent $28.5 billion on capex. Its free cash flow was negative $5.4 billion. For the full fiscal year, it was negative $23.7 billion.

So how is Oracle paying for it? With debt and stock. It raised $43 billion in debt last fiscal year and sold $19.9 billion of its own stock this quarter. Total debt is now $125 billion, and interest expense is up 55%. It also has $288 billion of data center lease commitments that are not on its balance sheet.

The market is already pricing in the risk. S&P cut Oracle’s credit rating to BBB-, one notch above junk. The cost of insuring Oracle’s debt hit a record. And the stock is down about 59% from its peak.

There is also a second part of the loop. Paper profits. Every time an AI startup raises money at a higher valuation, the investor raises the value of its stake on its books and counts the increase as income. No cash comes in. The profit is on paper.

Look at Alphabet in Q2 2026. Net income was $112.1 billion. $99.0 billion of it came from gains on investments, driven by SpaceX and Anthropic. Its operating profit was $40.8 billion. So the investment gains were more than twice its operating profit.

Amazon shows the same. Net income was $62.6 billion. $53.4 billion of it came from investment gains, mainly from Anthropic. Its operating profit was $27.5 billion. So the Anthropic gain was almost twice its operating profit.

And while Amazon reports record profits, its cash tells a different story. Free cash flow over the last twelve months was negative $7.6 billion. A year earlier it was positive $18.2 billion. It is spending more cash than it brings in.

This has happened before. In 2001, Qwest and Global Crossing swapped network capacity with each other to book sales. Qwest had to reverse $950 million of that revenue and later paid a $250 million settlement to the SEC. Global Crossing filed for bankruptcy in January 2002. Those swaps were found to be improper. Today’s deals follow current accounting rules.

This also reaches ordinary investors. These stocks are the largest holdings in index funds and retirement accounts. When prices rise, those funds have to buy more of them. So the loop feeds itself. Investments become sales, sales become backlog, backlog lifts stock prices, and higher stock prices let the startups raise more money to invest again.

So what are the signs? The backlog is growing because of two startups that lose money or barely make a profit. Oracle is borrowing, selling stock and spending more cash than it earns to build for a contract that starts around 2027. Alphabet and Amazon are reporting profits that come mostly from paper gains on the same startups. And one report that OpenAI’s revenue was lower than investors assumed was enough to drop Oracle 5.5% in a day.

This does not mean AI revenue is completely fake, or that the technology is not real. But a large part of the sales and profits these companies report depends on money that investors, including the tech giants themselves, keep putting into OpenAI and Anthropic. If that funding slows, the backlog, the sales and the paper profits all weaken together.