The 2010s saw low rates and a high equity risk premium (ERP), suppressing multiples. If today's rising rates also cause the ERP to expand back to 2010s levels, it would create a double negative pressure on stock valuations, reversing the last decade's tailwind.
AI data center investments are viable because long-term leases cover initial costs, leaving developers with a valuable future asset. Rising rates increase lease costs and decrease the present value of that future asset, threatening the financial viability of the entire AI infrastructure buildout.
Investors in AI data centers bet heavily on the 'terminal value' after an initial 15-year lease. This ignores the risk that if the AI boom fades, the asset might have to be re-leased at drastically lower, pre-AI rates (like those for Bitcoin mining), destroying the investment thesis.
Investors should scrutinize capital allocation, like large dividends, at founder-led firms. These can be red flags that the CEO is using the company's balance sheet to fund personal ventures or cover outside financial pressures, as allegedly seen with UWMC and Cogent.
Be wary of anointing CEOs as the next great capital allocator. Often, their stellar track record is the result of riding one powerful theme, like vertical software. Their perceived genius is often tied to a factor that may not persist, unlike true multi-industry compounders.