Episode Summary
Executive Summary: Arthur Hayes argues the AI boom is a late-stage bubble fueled by circular financing, subsidized usage, and cheap capital that will eventually break under depreciation math, energy costs, and politics. He says the eventual AI credit event and bailout will redirect liquidity into Bitcoin and crypto, while he’s currently rotating into T-bills, ETH, and selective energy bets.
Main Topics: AI bubble, circular financing, and the coming credit break (Priority: 5/5): Hayes says AI capex is being financed through circular deals where hyperscalers, chipmakers, and labs reinforce one another. He believes GPU depreciation is mispriced, returns are overstated, and the eventual mismatch between cost of capital and actual economics will force a reset. Crypto rotation and capital preservation (Priority: 4/5): He explains selling Hype, Near, and Zcash because the asymmetry was gone or risk had risen, preferring Bitcoin, T-bills, and selective setups with better risk/reward rather than chasing every rally. Energy, oil, and macro fragility (Priority: 4/5): Hayes views current oil weakness as a fake-out. He expects restocking, strategic inventory rebuilding, and renewed geopolitical risk to push hydrocarbons higher, supporting energy equities and making oil a key macro variable. Ethereum as a large-cap crypto trade (Priority: 3/5): Compared with Bitcoin and smaller tokens, Hayes sees ETH as a relatively clean, large-cap asset that is meaningfully below prior highs and below its long-term trend, making it attractive on a chart basis. Perpetual swaps and market structure (Priority: 4/5): Hayes revisits the origin and mechanics of perps, arguing they are a superior retail product because they offer 24/7 high leverage with socialized loss, making them structurally likely to dominate trading. US vs China AI economics and commoditization (Priority: 4/5): He argues Chinese models and infrastructure can commoditize AI tokens and models because they are far cheaper while good enough, eroding the premium of US AI brands over time. Political backlash against AI and regulatory risk (Priority: 4/5): Hayes expects AI to become a major political issue as ordinary people feel excluded from gains while bearing local costs like pollution, data centers, and higher living expenses. He sees a future anti-AI policy wave as a bubble catalyst.
Key Arguments: AI capex is being allocated through circular revenue and financing loops, so reported growth may not translate into durable economics. GPU and chip depreciation is being underwritten on multi-year schedules even though useful life is closer to two years, creating a future credit problem. If AI funding cracks, the response will likely be money printing and bailout support, and that liquidity will then flow toward Bitcoin and broader crypto rather than back into AI. Energy prices matter because AI depends on power, but oil could become less relevant only if the market fully believes AI can survive even materially higher hydrocarbon costs. ETH has a comparatively strong large-cap setup versus most altcoins because it has a durable brand, size, and lower zero-risk than smaller tokens. Perpetual swaps win because retail wants 24/7 leveraged exposure, and socialized loss is the only practical way for exchanges to offer that product safely. China will pressure AI margins by offering cheaper, open or good-enough models, reducing the value of US AI branding and forcing price competition. AI may become a political liability as voters who do not share in the gains mobilize against data centers, pollution, and job displacement.
Data Points: AI chip improvement cadence: every two years - Hayes says Moore’s-law-like improvement cannot be changed by simply printing more money. Capital allegedly poured into economy: $10 trillion - Used as an example of how monetary stimulus cannot alter chip improvement cycles. Timeframe for depreciation mismatch: 2025–2026 capex boom; 2027–2028 stress window - He says loans and GPU financing done now will hit trouble a few years later as depreciation catches up. Oil stress level: $120 WTI - Hayes suggests oil could revisit around $120 in 6–12 months under restocking and geopolitical pressure. Oil stress upper range: $150 oil - He says $150 oil would likely test the AI narrative, but may still not stop it if the bubble remains euphoric. Ether relative trend: ~30% below 200-week moving average - Used to argue ETH looks attractive versus Bitcoin on a chart basis. Bitcoin relative trend: at its 200-week moving average - He contrasts Bitcoin’s position with ETH to make a relative-value point. Emerging-market yield: over $115 billion annually - Shown in sponsor copy about yield opportunities in emerging markets. Emerging-market yields: 10% to 40% - Sponsor copy referencing high yields accessible through tokenization/DeFi products. Tokenized stock competition prize: $100,000 - Sponsor copy for a trading competition on MetaMask and Ondo. Trading competition duration: five weeks - Sponsor copy for the MetaMask/Ondo competition. Perp leverage: up to 100x - Hayes repeatedly cites high leverage as central to perp demand. U.S. launch constraint: lower leverage, around 10x (implied) - Discussion of how regulation may reduce the highest-leverage features onshore. ETH prior all-time high: $5,000 - Hayes notes ETH still trades below its 2022 peak. Bitcoin all-time high range cited: $69k to $125k - Speaker references Bitcoin’s rally above prior highs during the cycle.
Pivotal Quotes: "I do not want to put any new capital, whether it's free or not, into AI because it does not meet its cost of capital. So, therefore, this capital goes straight to crypto." — Arthur Hayes: Explaining why he thinks an AI bubble pop would redirect liquidity into Bitcoin and crypto. "The Fed can't print Moore's Law." — Arthur Hayes: Core thesis that money printing cannot override technological depreciation and chip improvement cycles. "Perpetual swaps ... are a product for [retail] combined with socialized loss mechanisms." — Arthur Hayes: Describing why perps are structurally attractive and likely to dominate trading.
Implications: Listeners should treat AI as a crowded, capital-intensive trade with political and credit risk, while watching ETH, Bitcoin, and energy as hedges/alternatives. Hayes’ view implies the next major macro move could be an AI-led risk unwind followed by a large liquidity wave into crypto.