Amid the AI arms race led by global tech giants, the authenticity of balance sheets is facing unprecedented scrutiny. Moody's, an international rating agency, recently issued a warning that the current U.S. accounting standards (GAAP) have significant "gray areas," allowing tech giants such as Microsoft, Oracle, and Meta to effectively "disappear" billions of dollars in potential liabilities from their balance sheets related to data centers.
The core of this accounting "magic" lies in opaque financing models. Given the astronomical amounts of capital required to build AI data centers, tech giants are increasingly using special purpose vehicles (SPVs) to attract external investment and then lease back these facilities. Although long-term lease contracts are essentially debt in the eyes of credit analysts, tech companies cleverly structure lease terms so that these expenses are nearly "invisible" on their financial statements.
In its report, Moody's analysts stated that current disclosures "fail to show the full picture." For example, companies often sign short-term lease agreements while committing to pay large compensation when assets depreciate due to non-renewal, a potential risk that is frequently overlooked in current debt calculations. This means that as AI investments are expected to surpass $3 trillion within the next five years, investors may find it difficult to assess the true financial health of these giants through their financial reports.
As concerns about an AI bubble grow, Moody's report sounds a clear alarm. AIbase believes that while tech companies have substantial cash flows as a moat, the tendency to keep high-depreciation assets like data centers off-balance-sheet is undermining the transparency of capital markets. During the peak of AI construction, this accounting limitation could lead to potentially serious financial risks being severely underestimated.
