In Q4, Meta tweaked its depreciation accounting for AI infrastructure, extending it from 4-5 to 5.5 years, cutting expenses by $2.9B in 2025 and boosting profit
Jeran Wittenstein / Bloomberg :
Context & Ripple Effects
This accounting choice sits early in Meta’s AI-buildout arc. Subsequent coverage shows its 2026 capital-spending outlook rising to $125B–$145B and a separate long-term CoreWeave infrastructure commitment, making the assumed useful life of AI equipment increasingly material to reported earnings.
The prior coverage also linked improving advertising growth to capacity for costly AI investment. Extending asset lives does not reduce the cash needed for infrastructure, but it changes when that cost appears in Meta’s income statement.
First-order effects
- Meta records less depreciation expense in 2025—$2.9B less, according to the report—thereby increasing reported profit for that year.
- The expense associated with the affected AI infrastructure is spread over 5.5 years rather than the prior 4–5-year range, shifting a portion of depreciation into later periods.
Second-order effects
- Investors and analysts need to separate operating performance from the earnings benefit of the accounting-life change when comparing Meta’s margins and profit across periods.
- As Meta’s infrastructure budget expands toward the previously forecast $115B–$135B 2026 range and beyond, depreciation assumptions become a more consequential input to its reported cost base.
Third-order effects
- If AI hardware investment remains exceptionally large, asset-life estimates will become a central point of scrutiny in judging whether hyperscalers’ AI-era earnings accurately reflect the economic consumption of rapidly evolving equipment.
- The pattern points toward AI infrastructure being managed not only as a capacity challenge but also as a capital-allocation and accounting-duration challenge; whether longer lives prove appropriate depends on actual equipment utilization and obsolescence.
The trend: AI infrastructure spending is making depreciation schedules and other capital-accounting choices increasingly important to how large platforms present AI investment returns.