Troubling: AI's self-investment spree sets off bubble alarms on Wall Street (finance.yahoo.com)

Big Tech has begun recycling huge sums inside the AI ecosystem: most notably Nvidia’s announcement it could invest up to $100 billion in OpenAI to power training and inference on Nvidia chips, plus deals such as Nvidia’s $6.3 billion tie-up (and 7% stake) in CoreWeave, a reported $2 billion link to xAI, and multiple cross-investments between OpenAI, Oracle, CoreWeave and AMD. Wall Street sees a growing pattern where infrastructure providers bankroll customers who then buy ever more compute, and customers in turn invest in their suppliers — effectively blurring sales, capital, and demand signals across the AI stack. The significance for the AI/ML community is twofold: technically, these flows accelerate deployment of GPU-heavy model training and hyperscale infrastructure (potentially speeding model iteration and capability gains); economically, they increase systemic risk. Analysts warn circular financing can overstate durable demand for compute, tightly couple valuations, and amplify contagion if a major player (many eyes on unprofitable OpenAI) stumbles — echoing vendor-financing failures from the dot‑com bust. Other red flags include rising debt (Oracle’s recent $18B raise, some AI firms levering up) and weakened customer resilience. Proponents argue such capital expedites essential infrastructure buildout, but critics say the leverage and intertwined stakes could transform isolated setbacks into wide-ranging shocks for the AI industry.