Case Details

Meta Platforms, Inc. (NASDAQ: META), extended the estimated useful life of its servers and networking assets, from 4-5 years to 5.5 years. This change reduced depreciation expenses by around USD2.9bn, increasing reported profit. The Big Short investor Michael Burry alleged that major technology companies including Meta were understating depreciation by assigning chips unrealistic longer useful lives. Meta’s CFO Susan Li, maintained the company was making efficiency gains by extending their useful lives. Once the artificial intelligence (AI) chips and assets required replacement sooner than planned, depreciation costs would increase in future periods, thus reducing the profits in subsequent years.

Simultaneously, Meta aggressively expanded into AI data centers and began construction of a massive AI data center in Richland Parish, Louisiana, US, and planned to invest over USD10bn for a 20% holding in a joint venture Hyperion. The project was financed primarily by Meta’s joint venture partner through USD27.3bn in debt financing, the data center’s design was controlled by Meta as sole tenant. Neither the debt nor the project’s cost appeared directly on Meta’s balance sheet. Meta decided to use a variable interest entity (VIE) structure for Hyperion, reporting the investment under the equity method, often known as the off balance-sheet approach. Analysts questioning whether Hyperion’s governance structure, the transparency of its financing, and Meta’s residual guarantee provided meant that consolidation was required.

This case illustrates how accounting choices, including depreciation schedules and VIE structures, can significantly alter the financial picture of large-scale AI investments. It highlights the challenges of rapid technology hardware cycles, regulatory gaps, and market skepticism. Students are invited to consider technical accounting rules involved but also the strategic implications of how companies present their AI capital expenditures to stakeholders.