The article highlights a significant disconnect in how depreciation is accounted for in the tech industry, particularly regarding AI and cloud infrastructure. It draws parallels between the depreciation of mining rigs and AI hardware, suggesting that traditional straight-line depreciation doesn’t accurately reflect the remaining value of these assets. The author references insights from financial analysts like Michael Burry and Morgan Stanley, who warn that this mismatch could lead to substantial depreciation charges for major tech companies, potentially totaling upwards of $680 billion over the next few years. Companies like Alphabet and Oracle might see their depreciation expenses skyrocket, which could challenge the prevailing assumptions about improving operating margins. A key point raised is the assumption of useful life for hardware. Hyperscalers like Amazon and Microsoft are extending the useful life of servers, which artificially inflates earnings. For instance, Alphabet adjusted its earnings guidance by $3 billion by extending the longevity of its data center equipment. However, this practice risks masking true financial health, especially as older technologies become less profitable. The article also touches on the challenges facing pure-play AI companies like OpenAI, which faces over $80 billion in deferred commitments this year, raising concerns about their ability to meet financial obligations without the backing of larger tech firms. The discussion stresses the importance of accurate accounting practices in a rapidly evolving technological landscape. As older hardware continues to be utilized, it doesn't necessarily translate into profitability. The potential for increased depreciation charges could reshape the financial outlook for these companies, highlighting a looming crisis in tech accounting that investors need to watch closely.