A new supply crunch is closing in on the AI boom as Wall Street starts to reach the limits of how much capital it’s willing to hand over to Silicon Valley.
Hyperscalers have already been grappling with shortages of chips, materials and other physical inputs needed for AI infrastructure. They are also running out of political capital as voters push back on the aggressive data center build-out.
Now the spigot of money is starting to dry up. OpenAI slashed the target on its latest funding round to $20 billion from $30 billion, sources told the Financial Times. That’s after securing about $11 billion in commitments, less than half of its original goal, according to the Wall Street Journal.
A top credit rating agency also dented Silicon Valley’s appeal to debt markets. S&P Global cut its credit outlook on Meta to “negative” from “stable” late Friday, citing negative cash flow and uncertainty around capital expenditure plans.
Meanwhile, the head of technology, media and telecom lending at a large European bank told Bloomberg last week that the bank is approaching the limit of what it can lend to the sector given its sizable exposure.
But the latest earnings reports indicate the sector still plans to spend heavily, sending their stocks tumbling. A handful of hyperscalers are expected to invest $800 billion this year and more than $1 trillion next year.
“Now, another concern is moving to the forefront: Does enough capital even exist to finance everyone’s ambitious AI spending plans?” Wall Street veteran Ed Yardeni asked last week.
In a note last week, he pointed to a series of red flags: Broadcom CEO Hock Tan warned capital could become the biggest constraint on AI; BlackRock CEO Larry Fink said that competition for capital was the fiercest he had witnessed in his career; and Nvidia CEO Jensen Huang observed that neocloud demand is “off the charts” but capital is scarce.
Trillions more in financing will be in demand as AI businesses fall short on self-sufficiency. Market research firm Gartner estimated data center owners need $8.5 trillion in cumulative revenue by the end of the decade to justify their investments. But the data centers are expected to generate no more than $3.6 trillion of enterprise and consumer AI revenue by then.
As a result, tech companies keep going to the well. On Monday, Oracle said it plans to raise $20 billion by issuing stocks and bonds. That’s after Elon Musk said SpaceX closed $40 billion in debt financing, Meta announced plans to raise $95 billion in equity, and Microsoft raised $25 billion in debt.
For its part, the sector recognizes that the availability of money is tighter. Meta nodded to the “increasingly challenging environment for raising capital” in a recent filing and suggested it may have to speed up fundraising instead of slowing it down.
“The Special Committee considered that, while the Company presently has sufficient capital resources for its current needs, in order to finance its funding needs going forward it would be prudent for the Company to do so before even more companies have absorbed available capital in the market,” it said.
To be sure, Wall Street may also turn more selective about which AI companies to fund, allowing well qualified borrowers to obtain financing while shutting out shakier ones.
Analysts at Bank of America said last week capital remains ample for investment grade firms and projects backstopped by suppliers, though such “circular” deals have also drawn more scrutiny. Still, BofA sees positive leading indicators for AI chipmakers into next year.
But tech companies aren’t the only ones tapping Wall Street for cash. The federal government ran a deficit of nearly $2 trillion last fiscal year with no end in sight to its profligacy, putting more pressure on the bond market.
“All those things combined basically means that at the moment we have a shortage of capital, and we have an abundance of good ideas that we need to invest in,” Apollo chief economist Torsten Slok told CNBC last month.
This story was originally featured on Fortune.com