AI Has A Hidden Debt Problem

AI Has A Hidden Debt Problem – Prof G Markets

Market views

Bullish

GOOGL— Google is successfully rearchitecting its entire business around AI while maintaining strong growth in core businesses (search up 17%, YouTube up 13%, cloud up 82%). Despite negative free cash flow concerns, the company is well-positioned to emerge with reduced competition and a multi-hundred billion dollar cloud business.

Bearish

TSLA— Tesla faces fundamental issues with an aging product lineup, eroding margins, and declining market share despite growing EV market. The company has lost critical talent that drove past success, and new ventures (robotaxi, Optimus) face numerous unanswered questions about product-market fit, manufacturing capability, and execution without the talent pool that made Tesla successful.
ORCL— Oracle is increasingly relying on debt to finance data centers in a way that is becoming borderline unmanageable, leading to credit rating downgrades and exploding borrowing costs in a downward spiral.
META— Meta is using special purpose vehicles to hide massive amounts of debt (roughly 3x its reported debt) related to AI data center buildout. This hidden debt is backed by data center revenues that may not materialize if AI demand doesn't meet expectations, creating significant risk for the company and its investors.
DATACENTER— AI data center debt is poorly underwritten, virtually uncollateralized, and based on revenue projections that require $1.68 trillion in annual compute demand - more than double the entire global software industry. The industry needs 15x current AI compute demand to service debt obligations, creating massive systemic risk as SPVs own the debt with no recourse to parent companies.
PE— Private credit funds backing AI data centers are conducting minimal due diligence (as little as 10 minutes for multi-billion dollar commitments) and are not rated by agencies like S&P. This shadow banking system is backed by retirement funds and insurance companies, creating systemic risk if data center revenues don't materialize.

Not investment advice.

Mentioned

Market views