o Amazon, Alphabet, Meta, Microsoft, etc., the capital expenditure scale of 4 big techs in 2026 is expected to be about $700 billion, and will approach $1 trillion in 2027.
- Considering that their capital expenditures were about $125 billion in 2021, it means an increase of more than five times in just five years.
o However, it is difficult to say that the AI bubble will end with this alone. The core reason for this is the historical '25% law'*.
* In US history, large innovation technology capital expenditure cycles such as the railroad construction boom in the 1870s, the power grid expansion in the 1920s, and the dot-com bubble in the 1990s absorbed funds and continued to expand until they reached about 25% of GDP.
o Currently, if we see the US GDP as about $30 trillion, 25% of GDP is about $7.5 trillion. According to historical standards, it is still in the early to middle stage.
- Based on historical paradigms, the possibility that investment will immediately peak and collapse is low.
o However, the key indicators that determine the end of the bubble are 'funding structure' and 'interest rate environment'.
- Companies are funding capital expenditures with the largest debt/equity issuance in history. If long-term interest rates surge or central bank monetary policy worsens, it could put pressure on the credit market and cracks may appear.
o However, unlike past bubbles such as the dot-com bubble where money flowed into companies with no revenue, AI infrastructure is actually generating revenue.
Ultimately, what matters more than the scale of AI investment itself is whether we can continue to supply funds to AI infrastructure.
Therefore, to judge whether the AI bubble will burst in the future, it is necessary to comprehensively look at factors such as AI revenue growth rate, data center utilization rate, big tech CAPEX, long-term interest rates, and demand for HBM and server memory.
▶ Original source: https://www.barrons.com/articles/ai-capex-bubble-burst-stock-market-history-2f73a9e4?mod=hp_LONGREADER_B_2
