Overview
- The IMF cautions that AI-fueled market euphoria has pushed risk-asset prices beyond fundamentals and says a shock could endanger equity at roughly 10% of U.S. banks and 30% of European banks, with shadow‑bank exposures amplifying systemic risk.
- Venture funding into AI startups in 2025 is reported at about $192.7 billion and OpenAI’s valuation hit $500 billion, as prominent investors including Paul Tudor Jones and David Einhorn draw parallels to late‑1990s bubble dynamics.
- Goldman Sachs counters that the cycle is still early, arguing current infrastructure spending is justified by prospective multi‑trillion productivity gains, with U.S. AI investment estimated at under 1% of GDP compared with higher shares in past tech build‑outs.
- Power demands from planned projects are enormous—reports cite data‑center plans requiring around 10 gigawatts and a Meta site targeting 2 GW—while Oracle outlines a supercomputer effort linking 800,000 GPUs beginning in 2026, underscoring continued build‑out.
- Analyses highlight mounting real‑economy pressures, with data centers consuming up to 39% of electricity in some U.S. states and local prices doubling since 2020 in heavy build regions, even as an MIT‑cited survey found 95% of organizations report no measurable AI ROI.