Even if interest rates rise, AI is unlikely to collapse
- What to look for
- You can examine the productivity and token-demand arguments supporting the claim that AI data center investment is unlikely to stop unless interest rates surge on a sustained basis or are raised consecutively.
- Why it matters
- This helps interpret distinct risks, such as the critical threshold of long-term interest rates approaching 6% and heavily indebted companies, as well as HBM prices and shareholder returns as conditions for a valuation reassessment.
AI summary
The video argues that even if US interest rates rise further, AI data center investment is unlikely to stop unless rates surge on a sustained basis or policy rates are raised consecutively. Its reasoning is that corporate AI adoption reduces labor costs and improves productivity, while token usage grows faster than prices fall, giving big tech companies an incentive to continue investing even while issuing corporate bonds. Positive factors cited for memory include the possibility of supply shortages continuing through 2028, the expansion of long-term supply agreements, and increased demand for enterprise SSDs resulting from HBM shortages. However, it separately identifies risks: a critical threshold at which long-term interest rates approach 6% and expectations of further rate hikes become entrenched, certain heavily indebted AI infrastructure companies, and the process by which the US secures demand for long-term Treasury bonds. It concludes that despite low valuations, foreign capital inflows into Korean semiconductor companies may be limited because of the range of AI investment opportunities in the US and differences in shareholder returns. Confirmation of HBM prices and earnings, along with consistent shareholder returns, are therefore conditions for a valuation reassessment.