Massive debt-fueled capital expenditures on AI data centers by hyperscalers like Microsoft, Google, Amazon, Meta, and Oracle—projected to exceed $1 trillion globally in 2026—combined with lagging revenue growth and falling prices for large language model tokens are the main drivers behind trader sentiment on an AI bubble burst. Recent analyst reports from Capital Economics and others highlight stretched valuations relative to expected productivity gains, while Michael Burry has increased short positions and shifted his timeline for a reckoning to 2027. Delayed IPO plans for OpenAI and Anthropic, alongside incidents like Oracle's payment issues on major projects, underscore risks from unproven business models and a potential "compute commencement wall" of higher costs next year. Key upcoming catalysts include earnings reports, any IPO filings, and regulatory scrutiny on energy demands and financing structures that could shift market-implied odds.