The AI Spending Spree
To understand Meta’s financials, you first have to grasp the sheer scale of its investment. The company is locked in an AI arms race with giants like Google, Microsoft, and Amazon. Winning requires a colossal arsenal of specialized hardware—namely, high-powered
servers packed with GPUs (graphics processing units). In the first quarter of 2026, Meta’s capex, which includes these infrastructure costs, hit nearly $20 billion. The company has signaled its spending for the full year could reach $145 billion, a staggering jump from previous years. This spending is the entry fee for competing in AI, but it comes with complex accounting that can make or break an earnings report.
A Quick Primer on Depreciation
When a company like Meta buys a billion dollars' worth of servers, it doesn't count that entire cost against its profits in one go. Instead, it treats the servers as an asset. The cost of that asset is then spread out over its “useful life” via an annual accounting expense called depreciation. Think of it like buying a delivery van for a pizza shop. The van helps generate revenue for several years, so you expense a fraction of its cost each year. The key variable here is “useful life.” Deciding whether the van will last three, five, or seven years dramatically changes the annual depreciation expense and, therefore, the shop’s reported profit each year.
The Crucial Detail: A Server's 'Useful Life'
This brings us to the critical detail for Meta. A few years ago, the company extended the estimated useful life of its server and network equipment from around four years to five and a half. This change had an immediate, positive impact on earnings, reducing depreciation expenses in 2025 by $2.9 billion. But here's the tension: the AI field is moving at lightning speed, with new, more powerful chips emerging every year. Some critics, including noted investor Michael Burry, argue that the true useful life of this cutting-edge hardware might be shorter, not longer, than these estimates. If Meta is too optimistic and has to write down the value of obsolete servers later, it could lead to a significant future hit to its reported profits.
How It Skews the Earnings Picture
This accounting choice creates a major gap between cash flow and reported profit. While Meta's revenue growth has been strong—up 33% year-over-year in the first quarter of 2026—its massive capex spending is causing free cash flow to shrink dramatically. One analyst report projected that the combined free cash flow for five major tech companies, including Meta, could fall by over 90% in 2026, even as their collective net income is forecast to rise. This is the depreciation effect in action: the cash is spent now, but the full expense on the income statement is delayed and spread out. A longer depreciation schedule makes current profits look better, but it may also be masking the true, immense cost of staying at the AI frontier.
