If you followed the October earnings announcements from the technology industry’s biggest companies, you would be hard-pressed to find much reason for pessimism about AI. Alphabet reported that Google Cloud revenue more than doubled year over year to $30.7 billion. Microsoft reported Azure growth of 48% and commercial remaining performance obligations of $677 billion. NVIDIA’s latest quarterly revenue reached $96.2 billion, with $89 billion coming from its Data Center business.

These are extraordinary results, and they reflect a very real transformation taking place across enterprise technology. AI adoption is accelerating, demand for computing infrastructure remains strong, and companies are committing unprecedented resources to what many believe will become the foundation of the next computing era.

But after looking at the numbers and the financial relationships supporting them, I found myself increasingly interested in a different question. How much of this activity represents independently generated customer demand, and how much involves the same money moving through an increasingly interconnected AI ecosystem?

That question became the foundation of our new Techstrong Special Report, The AI Money-Go-Round: Behind the Record Earnings, Who Is Really Paying for the AI Boom?

When Everyone Is Everyone Else’s Customer

One of the report’s central findings is that the traditional boundaries separating suppliers, customers, investors and financiers are becoming increasingly blurred.

Microsoft has invested billions in OpenAI while recognizing significant commercial revenue from its relationship with the company. NVIDIA sells GPUs and networking infrastructure to AI cloud providers while also making commitments to purchase computing capacity from some of those same providers. Oracle is using customer prepayments, vendor financing and other arrangements to support its enormous infrastructure expansion.

There are legitimate commercial reasons for all these relationships. Infrastructure requires capital, model developers need computing capacity, and long-term commitments can help make ambitious projects financially viable.

The difficulty comes when we attempt to measure the size of the resulting market. A dollar invested in an AI developer can help fund a cloud contract, which in turn supports equipment purchases and infrastructure construction. Each transaction can generate legitimate financial activity, but adding the announcements together doesn’t tell us how much new money entered the economy or how much independent demand supports it.

The report illustrates this with a hypothetical example in which $10 billion of investment capital can produce $27 billion in announced AI-related financial activity across investment, cloud commitments and equipment purchases. The transactions aren’t fictitious. They simply aren’t interchangeable measures of demand.

Why Enterprise Leaders Should Care

For CIOs and technology executives, this issue extends well beyond Wall Street valuations.

The financial health of the AI ecosystem increasingly influences the availability, pricing and reliability of infrastructure and services that enterprises are building into their own operations. Companies making long-term commitments to model providers, cloud platforms and specialized AI infrastructure operators need to understand the business relationships supporting those providers.

A hyperscaler with substantial cash generation from established businesses can withstand an extended period of infrastructure investment. A specialized AI cloud provider dependent on borrowing and a handful of major customers may face very different pressures if compute pricing changes, financing becomes more expensive or demand develops more slowly than expected.

There is also the question of technological depreciation. New generations of accelerators, networking systems and AI architectures can improve the economics of computing, while potentially reducing the commercial value of infrastructure purchased only a few years earlier. That is good news for enterprise customers in some respects, but it can complicate the financial outlook for the providers those customers depend on.

None of this suggests enterprises should slow their AI adoption. It suggests that vendor financial resilience, contractual flexibility and infrastructure economics deserve a place alongside model capabilities, security and performance in enterprise AI planning.

The Demand Is Real. The Financial Picture Is Incomplete

Our investigation does not conclude that the AI boom is artificial or that companies are improperly reporting revenue. Enterprise and consumer demand is growing, and the financial performance of several leading technology companies is exceptional.

What we found is that publicly available financial disclosures do not allow us to determine reliably how much of the industry’s overall financial activity ultimately depends on independent customer spending versus continued financing and commercial relationships among ecosystem participants.

That leaves important questions about the sustainability of the investment cycle, the quality of enormous contractual backlogs and the concentration of risk among companies increasingly dependent on one another.

The 35-page report examines these relationships in detail, including Microsoft and OpenAI, NVIDIA’s reciprocal cloud commitments, Oracle’s financing structures and the economics of specialized AI cloud providers. It also looks at three possible paths for the industry: continued successful expansion, a period of infrastructure overcapacity and adjustment, or a more disruptive interruption in financing.

I remain optimistic about AI’s long-term potential. But I’ve been around technology long enough to know that an extraordinary technological advance doesn’t guarantee attractive financial returns for everyone investing in it.

The October earnings releases told us how fast the money is moving. Our report examines where it comes from, how it moves through the ecosystem and who could be left holding the risk.

Read the full Techstrong Special Report: The AI Money-Go-Round (PDF)

The report includes 14 financial exhibits and diagrams, four explanatory sidebars, a methodology section and source notes.