AI's Circular Financing Trap: Why Concentration Risk Is Rising - And Where To Look Instead
AI has supercharged market returns, but the financing behind it is getting risky. The concern: a loop where Big Tech funds AI startups, then books revenue when those startups spend the same money on their cloud or chips. If that loop breaks, it won't just hit a few stocks - it could drag the indices.
What circular financing looks like
Here's the pattern: a large investor writes a big check to an AI startup, and the startup commits to spend heavily on that investor's infrastructure. One high-profile example cited by market watchers is OpenAI reportedly committing to buy 10 GW of Nvidia chips while Nvidia plans up to $100B in investment into OpenAI. That flow can mask cash burn and pull forward demand, making growth look sturdier than it is.
Valuations stay high on the story, not the unit economics. As one portfolio manager warned, if the AI trade "goes south, it's going to take the indices with it." In other words, this isn't a single-name problem - it's a concentration problem.
Macro check: the gap between prices and fundamentals
The economy is cooling. Manufacturing has weakened and layoffs are ticking up, while many AI-linked names still price in flawless execution and aggressive growth. Expect fewer triple-digit gains and more dispersion as liquidity tightens and projects meet real P&L hurdles.
Keep an eye on monthly manufacturing data like the ISM PMI to gauge demand drift. A softer backdrop makes elevated multiples harder to defend, especially for businesses reliant on external funding and prepayments.
A steadier chip option: Broadcom
For those who want chip exposure without the froth, Broadcom offers a different profile. Three pillars stand out: VMware integration (sticky, high-margin software plus silicon), custom chips for large customers, and disciplined dividends. That mix has delivered: shares are up roughly 51% YTD versus about 17% for the S&P 500.
Even so, position sizing matters. Some managers have trimmed after outsized gains to keep portfolio balance intact - a good reminder that risk control beats hindsight.
2026: a broader market could finally show up
If the Fed eases to counter slower growth, leadership may rotate to the "other 493" in the S&P 500. Boring is back: retailers and consumer staples like Home Depot and McDonald's tend to benefit from lower short-term rates and resilient demand. Defensive cash machines - Chevron, AbbVie, Waste Management - can provide income if the AI story leaks air.
Concrete portfolio moves for finance pros
- Cap single-theme exposure. Set hard limits for AI and related suppliers across the book.
- Scrub for circularity. Review related-party disclosures, prepay contracts, and concentration of spend with strategic investors.
- Re-underwrite unit economics. Separate usage-driven revenue from commitment-driven revenue and test sensitivity to price cuts and utilization.
- Run bear cases. Model a 20-30% drop in AI infrastructure spend and test index beta if the top names correct together.
- Favor cash generation. Prioritize FCF yield, recurring margins, and balance-sheet strength over story-driven TAM slides.
- Diversify factor exposure. Add equal-weight or value tilts to reduce mega-cap concentration risk.
- Use income and hedges. Blend in dividend payers; consider covered calls or collars on high-vol names to manage downside.
Bottom line
You don't need to bet against AI. You do need to stop betting the whole portfolio on it. Focus on cash flow, avoid circular revenue loops, keep positions sized, and add exposure to quality names outside the AI core - including steadier chip plays like Broadcom and dependable dividend payers.
If you're evaluating AI tools for finance workflows and want a faster way to map the space, this curated list can help: AI tools for finance.
Your membership also unlocks:
