Meta's $27B Off-Balance-Sheet AI Deal Reminds Steve Eisman of Enron— Joseph Carlson Disagrees

“Big Short” investor Steve Eisman sees echoes of Enron in how Meta Platforms Inc. (NASDAQ:META) financed a $27 billion AI data center. But YouTube investor Joseph Carlson says the comparison is unfair.

“I don’t think it’s a concern in the slightest,” Carlson said in a Wednesday video, calling Meta stock attractive after its recent selloff.

Why Eisman Sees Enron Echoes

Meta owns 20% of a Louisiana data center venture with funds managed by Blue Owl Capital Inc. (NYSE:OWL). A separate entity issued $27.3 billion in bonds, keeping that borrowing off Meta’s balance sheet.

Meta has agreed to lease the facilities and provide guarantees that could require substantial payments if it exits the leases and the properties lose value.

In September, Eisman raised concerns about arrangements that keep borrowing off a company’s balance sheet while leaving it financially exposed.

“That’s some fancy schmancy footwork that’s reminiscent of bad times past and it doesn’t pass the smell test,” he said.

Eisman also pointed to Ernst & Young flagging the accounting treatment as a critical audit matter in Meta’s 2025 annual report.

“Here it obscures,” he said of the accounting language, suggesting the auditor had relied too heavily on Meta’s representations.

Carlson Says Meta Has Nothing to Hide

Carlson said Eisman was portraying Meta as “a lot more sinister and sneaky than they actually are.”

Unlike Enron, Carlson argued, Meta has publicly disclosed its financing arrangements. He said critical audit matters attract greater scrutiny, not less, and credit-rating agencies already account for Meta’s obligations.

He also pointed to Meta’s profitability and the outside investors financing much of the construction.

However, Meta’s third-quarter filing disclosed $68.4 billion in maximum exposure to loss across its Louisiana venture and a second Texas project, including lease commitments and financial guarantees.

That represents potential exposure, not outstanding debt or expected losses.

S&P Global Ratings also revised Meta’s outlook to negative in October, citing concerns over AI spending and future cash flow.

Prediction-market traders see a broad AI downturn as unlikely in the near term. Polymarket priced a contract-defined downturn at roughly 6% by December 31 and 23% by June 30, 2027, with around $3.6 million traded.

The market requires three specified industry stress signals within 90 days, such as steep semiconductor stock declines or AI-company failures. It measures a broader industry downturn, not the risks of Meta’s financing arrangements specifically.

Image: Shutterstock