Why hyperscalers are increasingly using finance leases for data center shells?

Hyperscalers are increasingly leveraging finance leases for data center shells to rapidly expand AI infrastructure, a trend Morgan Stanley estimates could comprise up to 20% of the projected $2 trillion in capital investment over the next three years, with $388 billion already committed. This strategy, exemplified by Oracle's substantial lease commitments, enables accelerated buildouts and preserves liquidity for critical compute hardware like GPUs, but it also creates significant long-term contractual obligations and impacts balance sheet metrics such as ROIC and leverage. The varying accounting treatments for these finance leases across major players, including differences in free cash flow reporting by Amazon versus Google and Microsoft, further complicates financial comparability and reshapes how capital intensity is assessed within the tech sector.

Analysis

Hyperscalers are increasingly utilizing finance leases for data center shells to accelerate AI infrastructure expansion, a strategy Morgan Stanley estimates could account for up to 20% of the projected $2 trillion hyperscaler capital investment over the next three years, with $388 billion already committed. This approach enables companies to conserve liquidity for critical compute hardware like GPUs, which constitute over half of total outlays, while securing long-term control of powered land and shells.

These finance leases, exemplified by Oracle's over $100 billion commitment including a $15 billion liability for Abilene Phase 2 by mid-2026, represent significant long-term contractual obligations with typical 15-year terms and 7% implied interest rates. The structure expands balance sheets with right-of-use assets and lease liabilities, impacting key metrics such as return on invested capital (ROIC), leverage ratios, and cash generation due to front-loaded depreciation and interest expenses.

A significant challenge for investors is the lack of consistent financial reporting for these leases across hyperscalers. Morgan Stanley highlights that Amazon includes finance lease assets in its free cash flow (FCF) calculations, while Google, Microsoft, and Oracle do not, and Meta and Amazon include principal payments. This divergence in FCF presentation and capital expenditure treatment limits comparability and complicates the assessment of capital intensity and financial flexibility within the sector.

Key Decisions for Investors

• Scrutinize hyperscalers' balance sheets for finance lease liabilities to accurately assess true leverage and capital intensity, as these represent significant long-term obligations.
• Adjust free cash flow (FCF) models to normalize for varying accounting treatments of finance leases across companies, improving comparability for capital intensity and financial flexibility assessments.
• Monitor the impact of front-loaded depreciation and interest expenses from finance leases on reported profitability and Return on Invested Capital (ROIC) in the initial years of these contracts.