By Graham V. Pell, Senior Fellow for Applied Capital Scarcity
There has been considerable complaining lately about the cost of borrowing money.
Treasury yields have surged, financing conditions are tightening, and the weakest U.S. corporate borrowers are now paying spreads exceeding 10 percentage points over government bonds. Default activity among triple-C companies has reached roughly $40 billion this year.
Naturally, business owners are asking an important question:
Why is money suddenly so expensive?
Perhaps because you’re trying to use it for the wrong thing.
While ordinary companies continue requesting capital for outdated purposes such as machinery, inventory, employees and “keeping the business operating,” the artificial-intelligence industry has discovered a much more compelling use for society’s finite savings: constructing enormous refrigerated warehouses full of GPUs.
McKinsey estimates global data-center investment could approach $7 trillion by 2030. Nvidia is helping assemble more than $500 billion of third-party financing for AI infrastructure. ByteDance just secured a $29.6 billion loan largely intended to support its AI expansion.
And frankly, that leaves less money for your regional packaging company.
This is called innovation.
I regularly hear owners of traditional businesses complain that they cannot refinance debt at reasonable rates.
Have you considered installing 40,000 Nvidia accelerators?
One manufacturer recently told me his company employs 600 people, exports useful industrial components and has generated positive cash flow for 34 consecutive years.
I was forced to explain that none of those factories currently contain a large language model.
His financing problem immediately made sense.
Capital markets are not charities. Investors cannot be expected to fund every profitable enterprise merely because it produces goods people purchase.
There are models to train.
Critics worry that this unprecedented AI capital expenditure could crowd out other investment, push borrowing costs higher and leave increasingly elaborate financing structures vulnerable if expected AI revenues fail to materialize.
This misses the larger point.
If AI spending eventually generates enough productivity to justify several trillion dollars of investment, the financing was prudent.
If it does not, we will have created an extraordinary quantity of sophisticated collateral.
Either way, progress has occurred.
The solution for struggling borrowers is therefore not lower interest rates. It is repositioning.
A trucking company is now a distributed physical-inference logistics platform.
A shopping mall is an AI-ready compute campus awaiting electrical upgrades.
Your family restaurant is a human-supervised caloric fulfillment network with tremendous data-center conversion potential.
Even the federal government could improve its credit profile by announcing that the $38 trillion national debt is actually infrastructure financing for an extremely large sovereign AI startup.
Investors love AI startups.
So please stop complaining that artificial intelligence is absorbing extraordinary amounts of electricity, chips, construction capacity and capital.
The market has spoken.
Your business can still borrow money.
It just needs to stop being whatever it is.
Inspired by:
Financial Times - "Treasury sell-off piles pressure on weakest US borrowers" (September 5, 2026)
