Amazon and Alphabet profits expose the AI boom's circular financing

The same capital can appear in several success stories

A technology company can finance an AI laboratory, provide the infrastructure that laboratory purchases, and hold an equity stake whose value rises in later funding rounds. Cloud sales and investment gains then appear through different lines while depending on a connected relationship.

The accounting can be legal and disclosed while still demanding careful interpretation. A dollar of operating profit, a dollar of cloud revenue, and a dollar of unrealized investment gain do not carry the same durability or risk.

Circular does not automatically mean fictional

AI systems serve real users, data centers contain real hardware, and companies outside the financing network buy real services. Investment relationships can also help laboratories scale infrastructure faster than ordinary financing would allow.

The concern is not that every transaction is fake. It is that a small set of firms can become simultaneously investors, suppliers, customers, and valuation anchors, making independent demand and risk harder to see.

Stress-test the network, not only each company

If private valuations stop rising, unrealized gains can reverse even while core operations remain sound. If a laboratory cuts infrastructure spending, cloud-growth expectations and the value of the investor's stake can weaken together.

Boards, investors, and regulators should therefore map related financing, separate paper gains from operating results, and test what happens when one large counterparty slows.

• Report operating profit separately from unrealized AI investment gains.
• Disclose material investor, supplier, and customer overlaps.
• Measure end-user demand independently from funded cloud commitments.
• Stress-test simultaneous valuation and infrastructure-spending reversals.