Large Tech’s $2 trillion AI gold rush is hiding a structural flaw. Critics say the giants are quietly paying themselves via their very own cloud payments, igniting contemporary AI bubble fears that more and more echo the dot-com period.
Newest company filings present OpenAI and Anthropic alone anchor over half of the roughly $2 trillion in future cloud commitments held by Microsoft, Amazon, Google, and Oracle. This leaves 4 trillion-dollar firms leaning on two unprofitable startups.
The Cloud Loop That Pays Itself
Critics name the mechanism a round-trip funding loop. A tech big writes a billion-dollar examine to an AI startup. The contract then forces that very same cash straight again, within the type of cloud hire. The money by no means leaves the constructing.
Microsoft’s $13 billion stake in OpenAI is the textbook case. The funding landed largely as Azure cloud credit. OpenAI fed these credit into coaching fashions, and Microsoft circled and booked the consumption as contemporary industrial income.
OpenAI’s annual cloud invoice has reportedly ballooned previous $60 billion. The corporate’s precise income sits nearer to $25 billion.
Subscribe to our YouTube channel to look at leaders and journalists present knowledgeable insights
Anthropic performs the identical hand with Amazon. The Claude developer spent $2.66 billion on Amazon Internet Providers in 9 months, roughly each greenback it earned.
“All the AI increase may be constructed on faux income,” remarked analyst Bull Idea.
The sample echoes 2001, when International Crossing and Qwest Communications swapped fiber-optic capability to manufacture gross sales.
Qwest ultimately erased $1.4 billion in fictitious earnings, and International Crossing went bankrupt. The 2026 model stays totally authorized beneath present accounting guidelines.
Paper Earnings Are Doing the Heavy Lifting
The second leg of the loop sits on the earnings assertion. Each contemporary funding spherical for an AI startup lets its Large Tech backer mark up the funding and drop the paper acquire straight into web earnings.
Alphabet posted a document $62.6 billion revenue in Q1 2026. About $28.7 billion of that determine got here from a markup on its Anthropic stake, based on its submitting.
Amazon mirrored the trick. Roughly $16.8 billion of its $30.3 billion in web earnings tracked again to the identical Anthropic income story, based on Fortune’s evaluation.
Behind the headline revenue, Amazon’s free money move cratered 95% to $1.2 billion. The corporate poured $44.2 billion into bodily information facilities in the identical quarter.
Microsoft now carries 49% of its $627 billion future backlog tied to OpenAI alone. Oracle leans tougher nonetheless, with 54% of its $553 billion pipeline using on that very same single buyer.
Actual Firms Are Already Hitting the Wall
The larger downside begins the second AI leaves the protected loop and lands in a finances assembly. Peculiar firms can not recycle infrastructure spending into their very own income, and the invoices are arriving quick.
Uber torched its full 2026 AI coding finances by April after handing Anthropic’s Claude Code and Cursor to hundreds of engineers. Some employees burned $500 to $2,000 in month-to-month API expenses every.
Microsoft, regardless of a multi-billion Anthropic partnership, ordered its personal workers to cease utilizing Claude Code internally after token consumption had change into unsustainable, based on Fortune’s report.
Nvidia’s vp of utilized deep studying Bryan Catanzaro admitted his group now spends extra on compute than on human salaries.
“For my group, the price of compute is way past the prices of the staff,” Catanzaro not too long ago advised Axios.
Cheaper chips could not rescue the maths. Decrease token costs have a tendency to ask heavier agentic workloads, and enterprise AI spending could hold climbing even when {hardware} prices fall sharply.
The AI Bubble Enters Its Show-It Section
The market is not asking whether or not AI can develop. It’s asking whether or not AI will pay for itself.
“The primary firms to truly use AI at scale will not be in a position to afford it,” one analyst remarked.
Index funds and retirement accounts have been dragged deeper into a good cluster of trillion-dollar names whose AI-linked income hinge on a handful of unprofitable startups.
Crypto buyers maintain a direct stake. Bitcoin (BTC) hit a correlation with Nasdaq of 0.75 in January 2026,
This implies any unwind of the Nvidia and OpenAI commerce probably ripples straight into digital belongings. AI tokens, already risky, would really feel the primary blow.
The falling chip costs, agentic adoption, or chilly accounting math profitable the subsequent spherical is now within the steadiness, with the AI increase formally coming into its prove-it section.
Notably, mainstream finance has already taken discover, with Constancy’s personal AI bubble framework itemizing 5 warning checks.
“We expect 5 indicators could provide directional insights into future AI-driven market and financial tendencies,” Constancy listed.
- The speed of combination earnings progress
- Mixture earnings high quality
- Valuations vs. historical past
- The affordability/sustainability of company capex spending, and
- The interest-rate cycle
Large Tech’s Q1 filings already journey two of them, earnings high quality and capex affordability.
The increase could not get the possibility to show something if the warning lights hold multiplying.
The publish AI Bubble Fears Develop as Large Tech Allegedly Pays Itself in Cloud Loop appeared first on BeInCrypto.