The AI boom has a funding engine, and a good chunk of it appears to be AI companies writing checks to other AI companies.
Bloomberg reporting from October 12-13, 2026, describes a circular funding model that is accelerating infrastructure spending. The same model may also be deepening financial imbalances that look uncomfortably familiar to anyone who remembers 2008.
The basic loop is simple to sketch. Company A invests in Company B. Company B then spends that money buying products or services from Company A. Everyone’s revenue goes up, and everyone’s valuation looks healthier.
What the numbers show
The most detailed data comes from the Bank for International Settlements. BIS Bulletin 137, published on October 1, 2026, examined funding flows across 1,246 AI firms between 2021 and 2025.
Its headline finding: over 55% of incoming investment value to those firms came from other AI firms.
The flows run the other way too. According to BIS, 28.7% of outbound investment by AI firms went to other AI firms.
The supply-chain angle is where things get especially tangled. BIS found that 58% of intra-AI deal value involved commercial supply-chain relationships.
That means investors were frequently also suppliers, customers, or both.
The best-known example involves Nvidia and OpenAI. Their investment ties were initially reported at up to $100 billion, then later restructured to reflect actual commitments.
Nvidia’s financial relationship with CoreWeave, the cloud infrastructure provider, is another case Bloomberg highlighted.
Nvidia CEO Jensen Huang has pushed back on the framing. He has argued that the company’s equity stakes represent only a small fraction of the capital its partners raise.
The foreign capital piece
The circular model has attracted substantial capital from abroad. Savings from surplus economies like South Korea, Japan, and Taiwan have been drawn in to help fund US data centers.
By 2025, approximately $2.5 trillion in foreign capital flows were funding US data centers, per the research findings tied to Bloomberg’s coverage.
Bloomberg compares this pattern to the dynamics that preceded the 2008 financial crisis. Back then, surplus savings from abroad poured into US housing, helping inflate an asset bubble that eventually burst.
What this means for investors and the broader market
The core concern is distorted demand signals. If a meaningful share of AI revenue traces back to investments made by the sellers themselves, those signals get noisy. A company might look like it has booming demand when part of that demand is its own capital coming back as orders.
Heavy interconnectedness is the second risk. The BIS figures suggest AI firms are deeply entangled with one another through equity, supply contracts, and customer relationships.
For investors, the practical takeaway is to look past top-line revenue growth. Disclosures around related-party deals, vendor financing, and equity stakes held by suppliers deserve extra scrutiny. So do restructurings like the one that reshaped the reported Nvidia and OpenAI figures.
On the macro side, the flow of foreign capital into US data centers is worth tracking closely. A shift in savings patterns in South Korea, Japan, or Taiwan, or a change in appetite for US tech assets, could affect funding for the buildout.
The BIS data makes clear that a sizeable part of the sector’s financing has been self-referential. When more than half the incoming investment comes from inside the house, the house’s foundations deserve a closer look.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.