El-Erian cites Goldman Sachs chart on AI financing boom, warns on borrowing costs

Mohamed El-Erian sharpened the debate over AI financing by sharing a Goldman Sachs chart that points to huge U.S. funding needs and warning that borrowing costs may climb as demand runs ahead of supply. In his post, he wrote, “My thanks to LS for alerting me to this eye-popping chart from Goldman Sachs,” and said there is “no mechanism as powerful as the U.S. capital markets” when it comes to mobilizing funding.

El-Erian’s concern was not the scale of the market, but the strain inside it. He warned investors to think about a widening imbalance between “the massive appetite for capital market financing - across tech, government, and corporate sectors - and available supply,” a dynamic that could push borrowing costs higher and ripple beyond the United States.

Goldman has spent much of 2026 tracing the same pressure points. On April 2, the firm published “AI Exchanges: Power Problems?” with Brian Singer examining the sources and constraints of power demand growth from AI data centers. On June 1, Goldman followed with “What Is the Forecast for US Data Center Power Demand?” and on June 12 it released “The Outlook for AI-Related Stocks and US Interest Rates,” where Muhammad Qubbaj, co-head of U.S. Interest Rate Products in Goldman Sachs Global Banking & Markets, said he was not yet worried about the rising issuance of bonds by tech companies and the U.S. government.

By August 5, Goldman had pushed the theme further in “How AI Debt Is Reshaping Credit Markets,” saying credit markets were playing a growing role in the buildout of artificial intelligence and that nearly $500 billion of AI-related debt issuance had already been completed so far in 2026. That backdrop matters for Goldman’s bankers, syndicate desks and advisory teams, because the financing wave is not a single trade but a sustained pipeline of data center, power and balance-sheet work that requires constant structuring and distribution.

The pressure is showing up in deal pricing too. One Wall Street Journal report said Meta’s Texas data-center project priced a $12.5 billion offering at a higher interest rate than a similar deal, a sign that investors are demanding more compensation as the AI buildout leans harder on borrowing. For Goldman staff working these transactions, the message from El-Erian’s post is straightforward: the financing machine is still moving, but the cost of keeping it fed is getting higher.