AI's Circular Money Machine: Growth Signal or Bubble Risk?
The hottest AI deals increasingly run in loops. Capital, contracts, and chips are moving between the same few firms - and their valuations are soaring because of it.
Here's the loop in plain terms: one company funds another, which buys compute from a third, which buys chips from the first. It looks efficient on the surface. It can also make demand and revenue look stronger than they are.
The Loop: Who Funds Whom
Nvidia plans up to $100 billion in OpenAI. OpenAI spends heavily on cloud and compute. Oracle is set to spend about $40 billion on Nvidia chips to power OpenAI data centers.
Nvidia holds a stake in CoreWeave, which provides infrastructure to OpenAI and buys Nvidia systems. SoftBank, which owns a multibillion-dollar stake in Nvidia, is working with Oracle and OpenAI on "Stargate," a proposed $500 billion data center project, with Nvidia as a core tech partner.
On top of that, OpenAI announced a deal with AMD to buy its chips and the right to take up to a 10% stake in AMD. Nvidia and AMD are direct competitors, which adds another layer of complexity to how these flows might evolve.
Why Markets Are Concerned
When related parties fund, buy from, and invest in one another, it can inflate perceived demand and mask risk. Analysts warn that if AI productivity gains are slower than expected, a sharp correction in tech could follow.
We've seen a version of this movie. The dot-com era showed how reflexive capital and story-driven growth can decouple prices from fundamentals before a painful reset. Historical context helps frame today's risk.
Why This Could Still Work
AI workloads are compute-hungry and growing. Executives say the revenue potential justifies the spend. The thesis: the winners become multi-trillion-dollar franchises and pay back years of capex with high-margin software and services.
For that to hold, these companies must convert model usage into durable, high-quality revenue fast enough to cover obligations and reward investors. If not, the feedback loop that inflated growth can unwind just as quickly.
Investor Checklist: Separate Real Demand From Circular Revenue
- Map related-party exposure: identify where a vendor, investor, and customer are the same entities across deals.
- Scrutinize revenue quality: look for concentration in a small set of AI counterparties; check disclosures on minimum purchase commitments and take-or-pay agreements.
- Track capex and cash: compare capex and prepayments to operating cash flow; separate stock-based comp from true cash profitability.
- Watch GPU pricing and utilization: falling unit prices, rising discounting, or idle capacity are early stress signals.
- Assess counterparty risk: a weak link in the loop (funding, supply, or demand) can ripple across multiple balance sheets.
- Look for proof of end-customer demand: enterprise expansions, retention, and usage growth beyond AI peers.
- Check backlog vs. delivery: RPO growth is helpful only if it converts to revenue without heavy incentives.
- Note regulatory and antitrust exposure: tighter scrutiny on related-party deals can slow deployments or force unwinds.
Signals the Air Is Coming Out
- Sharp drops in GPU resale prices and rising secondary-market inventory.
- Vendor finance increasing while free cash flow falls.
- Customer concentration rising even as reported demand "broadens."
- Write-downs of prepayments or long-dated purchase commitments.
Portfolio Moves to Consider
- Diversify AI exposure: balance chips, cloud, and software; avoid single-point dependence on one vendor or customer chain.
- Favor cash generators: prioritize firms with positive free cash flow after capex and clear payback periods.
- Prefer multi-cloud and multi-supplier strategies: they're less exposed to one counterparty's stumble.
- Use risk controls: staged entries, covered calls or index puts around key earnings and capex updates.
- Look for "picks-and-shovels" with non-AI demand support (power, networking, memory, facilities) and transparent pricing.
What to Watch Next
- OpenAI's revenue traction and any detail on its AMD stake terms.
- Nvidia's disclosures on the OpenAI investment structure and expected revenue pull-through.
- Oracle's capex cadence and AI infrastructure utilization metrics.
- CoreWeave financing and expansion plans; any changes in Nvidia's stake.
- Progress on Stargate commitments and funding sources.
- The share of the S&P 500 tied to the largest tech names and their AI spend vs. returns. See the "Magnificent Seven" context.
Bottom Line
AI's current growth engine relies on tight loops of capital, chips, and cloud. That can compress timelines but can also exaggerate demand.
Treat every headline deal as a balance sheet question: who gets paid, when, and by whom. If the cash payback is clear, stay the course. If not, protect the downside.
Practical Resource
If you need a fast scan of useful tools for finance workflows, this curated list can help: AI tools for finance.
