Why I'm Short the AI Buildout Even Though I'm Long Computers (unverifiablerewards.substack.com)

A recent analysis sheds light on the troubling financial landscape of AI infrastructure amidst a booming AI market. Despite the perceived success of large language models (LLMs) in broadening coding and knowledge retrieval access, the mounting economic pressures—particularly rising interest rates—have created a significant disconnect between AI lab revenues and the hefty debts incurred for infrastructure buildout, which has surged to nearly $500 billion. Prominent players like Anthropic, OpenAI, and Google DeepMind are major consumers of GPU data centers, but their revenue growth has plateaued, undermining the financial projections that justified such massive investments. Companies are increasingly relying on borrowed capital at escalating costs, a strategy that risks shareholder value and long-term sustainability. The implications are significant for the AI/ML community, as the expected growth in AI lab revenues is not materializing at the rates needed to support the infrastructure costs. Cheaper alternatives from competitors are eating into profit margins, raising concerns over the viability of ongoing investments in AI infrastructure. With diminishing returns predicted from increased compute usage and declining pricing power, the cycle could result in a major market downturn by late 2026 or early 2027. This scenario highlights the urgent need for a reassessment of business models and funding strategies within the AI sector as it grapples with the realities of the broader economy.