The Market Huddle THE BIGGEST AND MOST DANGEROUS BUBBLE (Guest: Julien Garran)
84 snips
Oct 17, 2025 
Julien Garran, a macro strategist and partner at Macro Strategy, shares his insights on the AI bubble, arguing it could be the biggest threat to economic stability we've seen. He delves into why large language models are unlikely to yield sustainable profits and discusses the role of monetary policy in capital misallocation. Julien also highlights the rising costs associated with AI model upgrades and the implications of recent market dynamics. His take on the fragility of AI applications and comparisons to past bubbles will leave listeners both informed and intrigued.
AI Snips
Chapters
Transcript
Episode notes
Income S-Curve Drives Commodity Booms
- Garran's S-curve '$5k/$10-15k' framework links rising incomes to explosive commodity demand as middle classes adopt modern living.
- He sees India replicating China's early-2000s resource surge as millions enter higher income brackets fast.
GPU Data Centers Likely Unprofitable
- Data-center economics are weak: Blackwell GPUs cost ~$50k plus ~$25k to deploy and degrade fast, making current rental rates loss-making.
- Garran estimates breakeven rent ~$6.31/hr but observed rates are ~$3.79/hr, implying structural losses.
Vendor Financing Echoes Dot-Com Bust
- Garran compares Nvidia's receivables and vendor financing to Cisco in 2000 and flags Nvidia receivables up 626% over 30 months.
- He warns vendor financing and round-tripping compute sales resemble the pre-dot-com vendor-credit bubble.