Are we in an AI bubble? Felix Prehn of Goat Academy explains five clear warning signs that suggest the current AI boom may be fragile. The goal is not fear. It is clarity. Readers should understand how funding works behind the scenes, where the pressure points are, and why history might rhyme.
- Vendor financing loops
Vendor financing is when a seller helps the buyer pay for the product. This is normal if the seller has strong cash flow. Many car makers and phone brands do this. Today, parts of the AI stack use a more risky loop.
Simple version: An AI lab signs a huge cloud deal. The cloud provider buys chips. The chip maker then invests back into the AI lab or helps finance it.
Why this matters: Money circulates in a tight circle, not from real profits. If any link weakens, the loop can fail.
Term explained: Vendor financing is seller-backed lending. It becomes risky when it replaces true customer demand and cash generation.
- Debt-heavy balance sheets vs. cash flow
Some providers announcing very large multi-year AI infrastructure deals carry high debt relative to equity. Debt-to-equity shows how much a company borrows compared to what owners have invested.
Example dynamic: A provider commits to tens or hundreds of billions for data centers but must borrow large sums each year to build them.
Why this matters: If revenue growth lags the plan, interest costs and refinancing needs can become a strain.
Term explained: Debt-to-equity ratio compares borrowed money to owner capital. Higher numbers mean more leverage and more risk in a downturn.
- The AI funding gap
Building AI infrastructure is capital intensive. Analysts estimate multi-trillion spending this cycle, with annual needs near the size of total capital spending by all S&P 500 firms in a typical year.
Sources of funds: Big Tech cash flow, bonds, asset-backed financing tied to data centers, and
private capital.
The problem: Even after these sources, estimates still show a large gap, often cited around $800 billion or more.
Why this matters: If the gap is not filled, build-out plans may slow, be re-scoped, or shift risk to new players.
- Stress in private credit
Private credit firms raise money from institutions and lend to companies at higher rates. This market was expected to fill much of the AI funding gap.
Recent signs: Some large private credit vehicles and managers have underperformed while AI equities surged.
Contributing issues: Rising non-performing loans in consumer credit exposure, tougher refinancing at higher rates, and fundraising fatigue across many funds at once.
Why this matters: If private credit struggles, the missing capital must come from elsewhere, or projects get delayed.
- Classic capital cycle pressures
History shows a pattern in major build-outs:
Stage 1: Revolutionary tech, early outsized returns.
Stage 2: Capital rush, rapid capacity build.
Stage 3: Competition, overcapacity risks.
Stage 4: Returns fall, debt bites, weak players fail.
Stage 5: Consolidation; the technology wins long-term, but many investors lose money.
The dot-com build-out offers a useful parallel. The internet was real. The capacity got built. But several firms failed because they scaled too fast with too much debt before revenue arrived.
What could happen next
Scenario A: Slower capex. Data center and chip demand growth cools, pressuring high-valuation names.
Scenario B: More leverage at the largest platforms. This raises future balance-sheet risk.
Scenario C: Public support. Subsidies and guarantees extend the cycle but shift risk to taxpayers and can create moral hazard.
How to think about risk in simple terms
Separate technology progress from financing risk. The tech can be real while the financing is fragile.
Watch cash flow vs. promises. Fast revenue growth must convert to cash to support debt.
Track the lenders. If lenders struggle, borrowers may need to cut plans or pay more to refinance.
Expect uneven outcomes. Leaders with strong cash flow and low debt are more resilient than those depending on debt and vendor loops.
Key terms simplified
Vendor financing: The seller helps the buyer pay, often by lending or investing. Safe when backed by strong cash flow; risky when it replaces real demand.
Debt-to-equity: A measure of leverage. Higher means more borrowed money relative to owner capital.
Private credit: Non-bank lenders who fund companies at higher interest rates. Sensitive to defaults and funding conditions.
Capex (capital expenditure): Money spent to build long-lived assets, like data centers and chip facilities.
Bottom line
The AI future can be bright and still have a bumpy financing path. These five warning signs—vendor loops, high leverage, a large funding gap, stress in private credit, and classic capital cycle patterns—suggest caution. For investors, the lesson is to respect balance sheets, follow the cash, and remember that the capital cycle eventually turns—even when the technology story remains strong.
For more about Felix Prehn and Goat Academy, see the background here: Felix Prehn Goat Academy.