Overview

• Goldman Sachs says AI outlays remain under 1% of US GDP and appear sustainable given expected productivity gains worth roughly $8 trillion in present value.
• Regulators and market voices highlight risks, with the Bank of England warning of a higher chance of a sharp correction and UBS noting most bubble‑worried investors are still heavily invested.
• Citi estimates OpenAI’s announced 26 gigawatts of compute would require about $1.3 trillion by 2030, far outpacing projected revenue and raising financing and execution questions.
• Reporting points to operational and monetization strains at OpenAI, including large losses and low paid conversion for ChatGPT, alongside scrutiny of circular vendor financing and power constraints.
• Even as caution grows, big‑ticket commitments continue, from Google’s $15 billion India data‑center hub to Oracle’s swelling AI contract backlog, while JPMorgan and Wedbush flag still‑rising hyperscaler capex and chip demand.