Meta's AI Spending Exceeds Reported Figures Due to Off-Balance-Sheet Commitments

As the surge of AI-generated content continues to reshape social platforms—sparking confusion, distrust, and even backlash—major tech companies are doubling down on their artificial intelligence investments, a trend that promises even deeper AI integration into everyday digital experiences.

On Sunday, the Wall Street Journal published a report revealing that Meta and other AI developers are spending considerably more on their AI initiatives than they publicly acknowledge. The discrepancy stems from back-end agreements that are not reflected on their balance sheets.

According to WSJ, Meta currently holds approximately $693 billion in off-balance-sheet commitments, largely tied to AI expenditure. These costs are separate from the financial liabilities disclosed in Meta’s quarterly reports. In its Q2 update, published late last month, Meta capped its 2026 AI infrastructure spending at $145 billion—a figure that excludes these additional expenses, which will only materialize once the projects are completed.

Hyperion: A Case Study in Off-Balance-Sheet Obligations

WSJ highlights Meta’s “Hyperion” data-center project in Louisiana—a facility roughly the size of 1,700 football fields—as a prime example of how massive obligations can be kept off tech companies’ books. As WSJ explains:

“Meta initially agreed to lease Hyperion for a four-year term starting in 2029, with options to renew for up to 20 years. It guaranteed that it would make bondholders whole if it doesn’t stay the entire two decades. The company doesn’t think payments under that guarantee are probable, so it hasn’t recorded any liability on its balance sheet.”

In practical terms, Meta is not required to list any financial liabilities from the project until 2029. This arrangement is part of an estimated $347 billion in future lease obligations that Meta is currently able to keep off its official accounting.

While Meta has already committed hundreds of billions of dollars to AI development, the true scale of its financial exposure may be even larger than previously reported. The company’s investment strategy rests on the belief that its AI projects will generate substantial ongoing revenue. However, if those anticipated returns fail to materialize, the consequences could be financially devastating.

The Trust Deficit: A Critical Hurdle

Meta’s AI ambitions hinge on widespread consumer adoption and trust—a significant challenge given the company’s track record. As SMT recently outlined, Meta’s poor data-handling and privacy record, coupled with low consumer trust, could derail its expanded AI vision.

Meta is also embroiled in a major court battle related to social media harms, which could result in billions of dollars in compensation payouts. Combined with ongoing efforts to restrict teen social media use, these issues have contributed to a broader negative perception of the company. A 2025 survey from Forrester found that Meta is one of the least trusted corporations globally.

For Meta’s AI vision to succeed, it would need millions of users to grant its AI tools access to sensitive personal data—including health records, financial information, and other private details. In exchange, Meta’s personal AI agents would offer customized advice, daily action plans, and guidance. While the concept is intriguing, it appears unlikely—at least at this stage—that enough people will be willing to place that level of trust in Meta.

If consumer skepticism persists, these massive AI investments could come back to haunt the company within three years. The bottom line: Meta must make personal superintelligence a reality, or it risks paying trillions for a technology that fails to gain traction.

So, will Meta’s colossal AI bet ultimately pay off? The answer may depend less on technological prowess and more on whether the company can overcome its trust deficit—a challenge that seems increasingly formidable.