Major technology hyperscalers have accumulated approximately $662 billion in data center lease obligations that currently remain off their balance sheets under Generally Accepted Accounting Principles (GAAP), according to a recent report. These leases, which primarily support artificial intelligence infrastructure development, are expected to transition onto corporate balance sheets in the coming years as accounting standards evolve and lease terms commence.
The anticipated shift could materially impact key financial metrics for the sector's largest companies. When these obligations are recorded on balance sheets, over half a trillion dollars in previously unreported assets will require disclosure, potentially affecting debt-to-equity ratios, return on assets, and other performance indicators that investors and analysts use to evaluate financial health. The timing of these balance sheet adjustments coincides with accelerating AI infrastructure buildouts across the industry.
Some market observers have raised questions about the accounting treatment of these arrangements, specifically regarding the useful life assumptions applied to hardware assets and the classification of lease obligations. These concerns center on whether current accounting practices adequately reflect the true cost of AI infrastructure investments and whether companies may be overstating equipment longevity to reduce near-term depreciation expenses.