# This Is How The Stock Market Collapses. Source: https://www.youtube.com/watch?v=te3W5UHierw Recap page: https://rapidrecap.app/video/te3W5UHierw Generated: 2026-01-31T21:03:35.508+00:00 --- ## Quick Overview The stock market collapse scenario discussed is not automatically guaranteed, but it hinges on several critical risks related to AI investment, energy demands, and geopolitical tensions, especially concerning Taiwan's semiconductor dominance, where a disruption could cause a global economic contraction nearly double the 2008 financial crisis loss, highlighting the fragility of the current AI-driven economy. **Key Points:** - JPMorgan's paper, "Smothering Heights," suggests that 8% of the S&P 500's total returns since ChatGPT's launch are accounted for by just 42 AI-associated stocks. - Capital expenditure by hyperscalers is projected to hit $315.0 billion in 2025, with Meta, Amazon, Google, and Microsoft spending heavily on AI infrastructure. - Energy demands are a major risk, as training frontier AI models currently requires as much power as the entire US grid is adding annually through new data center builds (projected 25 GW additions by 2025 vs. 30.2 GW cumulative power requirement by 2030). - The geopolitical risk centers on Taiwan, which manufactures 92% of the world's 5-nanometer chips, making it a single point of failure for the global tech stack. - A Chinese blockade of Taiwan could cause a staggering estimated loss of USD 2.7 trillion in the first year, almost double the loss from the 2008 Global Financial Crisis, potentially shrinking China's economy by 7% and Taiwan's by almost 40%. - Despite massive investment, an MIT report notes that 95% of organizations currently see zero return on their GenAI investment, and CEO confidence in AI strategy has dropped from 82% in 2024 to 49% in 2025. - China's domestic AI chips (like Huawei's 910C) are significantly less power-efficient than leading US chips (like NVIDIA's B300), though Chinese firms are rapidly closing the gap in overall AI model performance. ![Screenshot at 0:06: The chart illustrates the massive divergence in performance between the S&P 500 including AI stocks \(up 13.14%\) and the S&P 500 excluding the top 17 AI-associated stocks \(up only 4.76%\) since Q4 2022, establishing the concentration of market gains in AI.](https://ss.rapidrecap.app/screens/te3W5UHierw/00-00-06.jpg) **Context:** The video analyzes a JPMorgan report titled "Smothering Heights" that examines the financial and geopolitical risks associated with the rapid growth and concentration of the Artificial Intelligence (AI) industry, particularly focusing on the market dominance of a few hyperscalers and the critical reliance on Taiwan for advanced semiconductor manufacturing. The analysis highlights that this growth is straining energy grids and creating geopolitical choke points that could trigger a severe global economic downturn if disrupted. ## Detailed Analysis The video details the risks underpinning the AI stock market boom, drawing heavily from a JPMorgan report, "Smothering Heights." The core finding is that AI stocks are disproportionately driving market returns; 8% of the S&P 500's total returns since late 2022 come from just 42 AI-related stocks, meaning the broader market is significantly underperforming. This concentration is fueled by massive capital expenditure from hyperscalers (Meta, Amazon, Google, Microsoft, AWS), projected to reach $315 billion in 2025, with Meta spending nearly 70% of its revenue on capex and R&D by 2025. This spending is creating an energy bottleneck, as US data center additions are projected to peak around 2027, demanding up to 30.2 GW of new power capacity by 2030—far exceeding current US grid buildout projections. The third major risk is geopolitical: Taiwan manufactures 92% of the world's most advanced 5-nanometer chips, making it a critical single point of failure. A Chinese blockade could cost the global economy an estimated $2.7 trillion in the first year alone, nearly double the 2008 financial crisis loss. Furthermore, an MIT report indicates that despite the investment, 95% of organizations see zero return from GenAI, and CEO confidence is falling. While Chinese firms like Huawei are rapidly closing the performance gap with NVIDIA's specialized chips, they remain significantly less power-efficient, and the US grid cannot currently support the projected energy demand for AI growth. ### JPM AI Concentration - AI stocks account for 80% of the S&P 500's returns since Q4 2022 - S&P 500 ex-AI stocks were only up 4.76% in the same period, while the full index was up 13.14% ### Hyperscaler Capex - Projected total capex of $315.0B in 2025, up from $23.8B in 2015 - Meta's capex/R&D as a share of revenue is projected near 70% by 2025, far above the S&P 500 median of around 15% ### Energy Risk - US data center additions project a peak of nearly 15 GW in 2027 - AI training power demand requires 30.2 GW cumulative capacity by 2030, which exceeds current US grid buildout projections ### China Risk & Taiwan Dependency - 92% of the world's 5nm chips are made in Taiwan (TSMC) - China's direct subsidies for listed firms are highest in Software & Services and Tech Hardware - A blockade could cause a 2.8% decline in global output, almost double the 2008 crisis loss ### AI Performance & Efficiency Gap - Huawei's 910C chip is ~3x less power efficient than NVIDIA's B300 - China's AI model performance is currently 1-2 generations behind the US, but closing the gap rapidly ### Careerist Ad - Promotes an AI Automation Online Bootcamp offering up to $140K/yr salaries after 4 months of part-time study (6-7 hours/week). ![Screenshot at 0:06: Chart showing the impact of excluding 17 AI-associated stocks on S&P 500 performance since Q4 2022, where the full S&P 500 gained 13.14% while the ex-AI index gained only 4.76%.](https://ss.rapidrecap.app/screens/te3W5UHierw/00-00-06.jpg) ![Screenshot at 0:11: A table from the JPM report detailing the massive growth in returns, earnings, and capex/R&D for the 42 AI-related stocks compared to the broader S&P 500 ex-AI index since November 2022.](https://ss.rapidrecap.app/screens/te3W5UHierw/00-00-11.jpg) ![Screenshot at 0:43: A graphic illustrating the massive projected capital expenditure by hyperscalers, reaching an estimated $315.0 billion in 2025, with AWS leading the spending.](https://ss.rapidrecap.app/screens/te3W5UHierw/00-00-43.jpg) ![Screenshot at 1:51: A map highlighting Taiwan as a critical geopolitical area, labeling four major risks: Capital, Energy, China, and 'What's Next'.](https://ss.rapidrecap.app/screens/te3W5UHierw/00-01-51.jpg) ![Screenshot at 7:47: A bar chart comparing US data center power load additions \(GW\) through 2030, showing a massive projected spike in demand for 100-500 MW and 500-1,000 MW facilities starting around 2024.](https://ss.rapidrecap.app/screens/te3W5UHierw/00-07-47.jpg)