Ed Zitron critiques the financial dynamics of AI infrastructure in his extensive piece, highlighting a troubling trend where companies prioritize stock prices over actual performance. He identifies three key areas of concern: earnings optics, financing structures, and real profitability. The concept of earnings optics revolves around companies extending the useful life of their servers to artificially boost net income, with giants like Microsoft and Alphabet doing so to lower depreciation costs. Zitron frames this as a conspiracy, yet these companies often disclose such practices, suggesting it’s more about accounting strategies than deception. Financing structures are another critical point. Zitron points to the shift of leverage into private credit and special purpose vehicles, evidenced by Meta’s Hyperion data center, which allows the company to keep debt off its balance sheet. This creates a scenario where risks are obscured, a phenomenon he calls "risk laundering." The mismatch between long-term financing and the rapidly evolving AI landscape raises concerns about sustainability and financial health. Profitability remains an open question. Zitron argues that no generative AI company shows a clear path to profit, but this overlooks the evolving nature of the tech sector. The real issue is not whether AI can be profitable, but who will capture that profit amid shifting costs and market dynamics. The discussion around letters of intent (LOIs) reveals that while they signal potential deals, they are also merely options that can lead to stranded capital if not realized. Zitron's analysis suggests that the financialization of AI could clash with the commoditization of AI compute, creating potential instability for over-leveraged firms as the market evolves.