We remember the internet bubble. This mania looks and feels the same (crazystupidtech.com)

Authors who covered the 1999 internet bubble argue today’s AI mania looks — and may be — even bigger: three years into the AI revolution, global AI capital expenditures plus VC already top ~$600 billion this year, with Gartner projecting AI-related spending could exceed $1.5 trillion in 2025. Markets are wobbling: despite strong NVIDIA results, major AI-exposed names have slumped in the past month (Microsoft -12%, Amazon -14%, Meta -22%, Oracle -24%, CoreWeave -47%), and big tech’s concentrated spending (Microsoft, Alphabet, Meta, Amazon, NVIDIA, Oracle, Apple ≈ one-third of S&P 500) leaves the market vulnerable to rapid re‑rating. The authors stress AI is transformative but caution that the timeline for profitable returns is likely much longer than current hype suggests. Key technical and financial implications: hyperscale capex to build data centers and train LLMs has accelerated investment risk — several firms now spend roughly $70–100B/year on AI infrastructure (Oracle ≈ $20B). If demand softens or write‑downs occur, losses would be enormous and contagious across suppliers (NVIDIA, AMD, CoreWeave, OpenAI). Valuation signals are mixed (NVIDIA P/E ~45; S&P P/E ~24 near dot‑com peak levels), and four systemic vulnerabilities stand out: excess spending, leverage, speculative deals, and geopolitical competition with China — a factor that adds national‑security and policy uncertainty to valuation risk.