"The Big Short" switches to "battle mode"! Burry: Tech giants "underestimating depreciation leads to inflated profits," Oracle's profits inflated by 26.9% in 2028, Meta inflated by 20.8%

Michael Burry, known for predicting the 2008 financial crisis, has once again turned his attention to tech giants.

On November 11, Michael Burry posted on his social media platform that tech giants are underestimating depreciation by extending the "useful life" of assets, artificially inflating earnings. He estimates that from 2026 to 2028, large tech companies will inflate profits by $176 billion due to underestimated depreciation.

Burry promised to disclose more details on November 25. He specifically pointed out that by 2028, Oracle's profits could be overstated by 26.9%, while Meta's profits could be overstated by 20.8%.

Notably, he had previously disclosed holding put options on NVIDIA and Palantir, which has once again drawn market attention to his bearish stance on the AI industry.

Previously mentioned by Wall Street Insights, investment banks such as Morgan Stanley and Bank of America have also issued warnings, believing that the market severely underestimates the true scale of current AI investments and is unprepared for the impact of future depreciation costs, which could lead to the actual profitability of tech giants being far below market expectations.

Accounting "Tricks," Extending Hardware Lifespan to Smooth Costs

Burry believes that tech giants are playing "tricks" in their accounting practices.

In his post on the X platform, he pointed out that tech giants are underestimating depreciation by extending the "useful life" of assets, thereby artificially inflating earnings, which he describes as "one of the most common frauds of modern times."

He argues that against the backdrop of tech companies massively purchasing NVIDIA chips and servers to expand computing power, these computing devices, which typically have a product cycle of only 2 to 3 years, should not have their depreciation periods extended.

However, "hyperscale" players, including Meta, Alphabet, Microsoft, Oracle, and Amazon, are doing just that, with some companies even extending the depreciation cycle to 6 years.

(Expected lifespan of servers and network equipment for tech giants each year)

However, Amazon has already shortened the expected lifespan of some servers and network equipment from six years to five years in the first quarter of 2025, citing the acceleration of AI technology development.

If this trend reverses and expands, it will lead to accelerated recognition of depreciation expenses, impacting short-term profits.

Depreciation "Time Bomb," Market Has Yet to Digest Future Profit Impact

Previously, Bank of America also believed that Wall Street was "slow to react" to the growth rate of future depreciation expenses.

Bank of America analyst Justin Post pointed out in a report that as capital expenditures for Google, Meta, and Amazon significantly increase in 2024 and 2025, their depreciation expenses will inevitably accelerate after 2026. According to their calculations, by 2027, the depreciation expenses of these three companies alone are generally predicted to be nearly $16.4 billion lower than the actual situation, which means their future actual profitability may be far below the current market consensus.

In addition, the "short lifespan" issue of AI assets exacerbates this risk. Hardware technologies such as GPUs used for AI computing have rapid iterations and high loads, with an effective lifespan of only three to five years. This trend is contrary to the practice of tech giants extending the lifespan of their equipment.

Wall Street's Alarm: Underestimated Capital Expenditure and "Off-Balance Sheet" Expansion

Morgan Stanley's research emphasizes that the capital expenditure intensity of tech giants on AI is approaching peak levels seen during the internet bubble, but public data does not fully reflect the investment landscape.

Morgan Stanley highlights two factors that have led to the underestimation of actual investment scale.

First is the rise of financing leases. Companies like Microsoft and Oracle are increasingly using this "off-balance sheet" tool to build data centers, with the initial investment not counted in traditional capital expenditures.

According to their estimates, when financing leases are included, Microsoft's capital expenditure to sales ratio for the fiscal year 2026 will jump from 28% to 38%.

Second is the delayed effect of "construction in progress (CIP)," where a large amount of capital already spent is sitting on the balance sheet and has not yet begun to be depreciated, with its impact on profits just beginning.

Bank of America warns that the AI infrastructure market may repeat historical patterns of overcapacity and price wars caused by aggressive investments.

BofA believes that if supply growth continues to exceed demand, more aggressive pricing strategies may emerge in the industry as early as 2027.

At that time, to maintain the utilization rate of data centers, hyperscale vendors may be forced to lower prices, thereby eroding their profitability. This prediction also provides another layer of evidence for Burry's view that the profitability of tech giants is overstated.