Bloomberg reports that the biggest players in AI — Microsoft, OpenAI, and Nvidia — as explored in the economics of AI compute infrastructure — — have built a $100 billion circular economy where every dollar paid out comes back in through another door. The question nobody is asking: is this a flywheel or a financial hall of mirrors?
The Circle of AI Capital
Follow the money and you’ll get dizzy.
Microsoft invested $13 billion in OpenAI. OpenAI spends billions of that on Nvidia GPUs. Nvidia just invested $30 billion in OpenAI’s latest round. Microsoft buys Nvidia GPUs for Azure. OpenAI runs on Azure. The money doesn’t flow — it orbits.
This isn’t a supply chain. It’s a financial centrifuge where the same dollars get recycled through different corporate bank accounts, each time generating a new press release about “strategic investment” and “deepening partnerships.”
Everyone Is Everything to Everyone
The traditional categories of business relationships have collapsed entirely in AI:
- Microsoft is OpenAI’s biggest investor, biggest customer (via API licensing), and biggest competitor (via Copilot). It’s also Nvidia’s biggest cloud customer.
- OpenAI is Microsoft’s portfolio company, Microsoft’s technology supplier, and increasingly Microsoft’s rival. It depends on Nvidia hardware and now takes Nvidia money.
- Nvidia is the arms dealer selling to everyone — and now investing in everyone too. Beyond OpenAI, it has backed Anthropic and Elon Musk’s xAI. It has signed seven-plus multibillion-dollar investment deals in 2025 alone.
In no other industry does a single company simultaneously act as supplier, customer, investor, and competitor to the same partners. AI has invented a new corporate geometry.
Nvidia’s Investment Blitz
Nvidia’s strategy deserves special attention. With $30 billion flowing into OpenAI and additional billions into Anthropic and xAI, Jensen Huang is doing something unprecedented: the biggest hardware supplier in AI is becoming the biggest financial backer of its own customers.
Why? Because every dollar Nvidia invests in an AI lab is a dollar that eventually flows back to Nvidia as GPU purchases. It’s vendor financing dressed up as venture capital. Nvidia doesn’t need these companies to succeed as businesses — it needs them to keep buying chips.
The Flywheel Argument
Bulls see a virtuous cycle. Microsoft’s investment makes OpenAI better. Better AI drives Azure demand. Azure demand means more Nvidia GPU purchases. Nvidia’s profits fund more AI investments. Each node strengthens the others. The whole system accelerates.
The Bubble Argument
Bears see something different: three companies inflating each other’s valuations with money that never leaves the circuit. Microsoft’s $13 billion investment in OpenAI looks brilliant when OpenAI is valued at $300 billion. But that valuation is partly based on Microsoft being a committed customer. Which is partly based on Nvidia GPUs that Nvidia helped finance.
Pull one thread and the whole tapestry unravels. If OpenAI’s revenue growth slows, Microsoft’s investment looks worse, Azure AI demand softens, Nvidia sells fewer GPUs, and Nvidia has less capital to invest in the next round. The flywheel becomes a doom loop.
What History Tells Us
Circular economies in tech aren’t new. The dot-com era had companies buying ads from each other to inflate revenue. The crypto world had protocols investing in each other’s tokens. In both cases, the music eventually stopped.
The difference: AI is generating real revenue. Microsoft’s AI business is a genuine profit center. Nvidia is producing actual chips. OpenAI has real users paying real money. The question isn’t whether value exists — it’s whether $100 billion of circular capital accurately prices that value or leverages it three times over.
The Strategic Takeaway
The Microsoft-OpenAI-Nvidia triangle reveals a new pattern: the circular moat. Instead of building walls to keep competitors out, these companies have built financial pipelines to keep capital in. The money can’t escape the system because every exit is someone else’s entrance.
It works — until external capital stops flowing in. The moment new outside money dries up, the circle has to sustain itself on actual profits. That’s the real test. Watch the ratio between fresh outside capital entering the system and recycled internal capital. When recycled exceeds fresh, the circular economy becomes a closed loop — and closed loops eventually run out of energy.
Read the full deep dive on Business Engineer →
Free tools: Map of AI | Try Business Engineering








