The Physical World Is Choking on the Cost of Capital
If you want to understand what the bond market is doing to the real economy, you have to look at the two most capital-intensive industries in the world: housing and data centers. On the surface, they have nothing in common. One shelters families; the other houses GPUs. But on a balance sheet, they are identical. They are both highly leveraged, physical infrastructure projects whose viability depends entirely on the cost of debt.
This week, the rising cost of that debt—driven higher by geopolitical tensions and a sticky long end of the curve—snapped both markets at once. The divergence between the physical cost of building the AI boom and the liquid wealth it generated has never been wider.
The 15% Yield on the AI Revolution
Neoclouds are the companies that buy Nvidia GPUs with borrowed money and rent them out. CoreWeave is the pioneer of this model. For 18 months, they were the darlings of the AI boom, reaching a $90 billion valuation. Today, their bonds are sinking into distressed territory, with yields shooting up to 15%.
They are not alone. Elice Inc., a $750 million data center developer, tanked 32% on its debut in Seoul. Firmus and AirTrunk have delayed their public offerings. Nscale and SB Energy are facing intense investor pushback on valuations. The market is suddenly demanding cash flow instead of promises.
The problem is not demand for AI. TSMC and ASML are still reporting massive orders. The problem is the cost of capital required to build the physical structures to house those chips. When the cost of capital spikes, the math on building a multi-billion dollar data center breaks down.
Main Street Meets the Neocloud
That exact same math is breaking the US housing market. Pending sales of existing US homes just tumbled 7.3% to a record low—the lowest reading since records began in 2001. The decline was steeper than any economist projected.
A prospective homebuyer looking at a 7% mortgage and a neocloud CEO looking at a 15% junk bond yield are experiencing the exact same macroeconomic constraint. The physical world is incredibly sensitive to the cost of debt. When rates stay high, concrete doesn't get poured, copper doesn't get laid, and houses don't get bought.
The Bifurcation of Wealth
The most surreal part of this dynamic is the parallel universe operating right next to it. While the physical infrastructure of the AI boom struggles for financing, the liquid wealth generated by the equity boom is sloshing around Silicon Valley in unprecedented volumes.
Jeff Wang, a former Sequoia stock-picker, just raised over $1 billion for a new tech hedge fund in one of the toughest fundraising environments in years. Meanwhile, VIP matchmakers in Menlo Park are charging Anthropic and Perplexity employees upwards of $500,000 for dating services. The liquid wealth is completely insulated from the physical constraints of the real economy.
This is the cost of capital collision. The digital economy can scale infinitely on a spreadsheet, but eventually, it has to touch the physical world. And right now, the physical world is choking on the cost of debt.