Tom Lee vs. AI Debt Trap: What Happens When 10-Year Bonds Fund 2-Year Microchips?
Key takeaways
- Tech firms are borrowing heavily to fund AI hardware that may become obsolete in a few years, turning bond yields into a key test of the AI spending thesis.
- $200 billion debt issued H1 2026
- $80.6 billion capex by Alphabet H1 2026
- $98.2 billion long‑term debt at June’s end
What happened
Tech giants like Amazon, Microsoft, Alphabet, Meta, and Oracle issued about $200 billion of investment‑grade debt in the first half of 2026, nearly double their 2025 total, to fund AI infrastructure. Alphabet spent $80.6 billion on capex in H1 2026 versus $39.6 billion a year earlier, and its long‑term debt was $98.2 billion at June’s end. Oracle plans to raise $45‑$50 billion in 2026 for cloud expansion, while Goldman Sachs projects the five hyperscalers will issue roughly $250 billion of…
Why it matters
Higher debt levels mean AI infrastructure is financed with long‑term bonds, but the hardware’s useful life is only 2‑3 years. If yields stay high, the cost of capital erodes the return on these short‑lived assets, potentially turning AI spending into a debt trap. Investors must assess whether free cash flow and capex…
Short-term and longer-term
Short-term. Over the next few weeks, traders should monitor Treasury yields and the hyperscalers’ quarterly debt issuances.
Longer-term. For investors, the key question is whether the AI infrastructure will generate sufficient cash flow before the debt matures.