Breaking Down Michael Burry’s Big Nvidia Short Thesis and Open AI’s Massive Loss Projections | Jack & Max
Timestamps are as accurate as they can be but may be slightly off. We encourage you to listen to the full context.
Podcast Summary
In this episode of Monetary Matters, Jack Farley and Max Wiethe dive deep into Michael Burry's controversial short thesis against Nvidia, examining his claims about artificially extended depreciation schedules and circular financing in the AI sector. (00:55) The hosts also analyze HSBC's shocking projections that OpenAI will lose nearly half a trillion dollars between now and 2030, despite massive revenue growth. (18:17) The discussion explores whether AI represents a transformative technology or a dangerous bubble, with particular focus on the sustainability of venture capital funding for money-losing AI companies.
- Main themes: AI bubble analysis, Michael Burry's Nvidia short thesis, OpenAI's projected losses, circular financing in tech, and Federal Reserve rate cut expectations
Speakers
Jack Farley
Jack Farley is the host of Monetary Matters and a financial analyst who focuses on macro-economic trends and market analysis. He has established credibility through interviews with Federal Reserve officials and deep dives into banking and financial sector analysis.
Max Wiethe
Max Wiethe is the host of "Other People's Money" podcast and serves as co-host on this Monetary Matters episode. He brings expertise in investment analysis and market commentary, particularly focused on technology and venture capital trends.
Key Takeaways
AI Companies Are Artificially Inflating Earnings Through Depreciation Games
Michael Burry's core argument centers on AI companies extending the weighted average lives of their depreciation schedules from 2-3 years to 5-6 years. (00:55) This accounting practice allows companies to spread out massive capital expenditure costs over longer periods, making their current earnings appear stronger than they actually are. For example, if a company spends $100 billion on AI infrastructure, they're taking a $20 billion annual depreciation charge instead of the more realistic $33-50 billion charge. This strategy masks the true cost of AI investments and creates an unsustainable earnings picture that could collapse when these assets lose value faster than expected.
The Real Bubble Center Is OpenAI, Not Nvidia
While Nvidia faces criticism for being overvalued, the analysis reveals that OpenAI represents the true epicenter of potential AI bubble risk. (18:17) HSBC projections show OpenAI losing cumulative losses approaching half a trillion dollars through 2030, despite revenue growth from $13 billion to $214 billion. This creates a scenario where venture capitalists must continue funding massive losses indefinitely, making OpenAI far more vulnerable than profitable companies like Nvidia that are actually selling products to these money-losing entities.
Circular Financing Creates Unsustainable Market Dynamics
The AI ecosystem relies heavily on venture capital funding unprofitable companies like OpenAI, which then purchase services from profitable tech giants like Microsoft and Google, who in turn buy hardware from Nvidia. (12:32) This circular flow of money is unsustainable because it depends on continued VC liquidity injections into companies that may never achieve profitability. When venture funding dries up, the entire chain could collapse, making companies dependent on this financing particularly vulnerable.
Semiconductor Cyclicality Will Eventually Return
Despite current secular growth narratives, semiconductors historically operate as cyclical businesses, and Nvidia will likely return to this pattern. (04:55) The current environment treats Nvidia as having steady, predictable earnings growth, but when the AI arms race slows and companies become more focused on return on investment rather than just spending, Nvidia's revenue and profit growth could become much more volatile. This cyclical nature would justify lower valuation multiples than the market currently assigns to the stock.
AI Revenue Growth Is Real But Profitability Remains Questionable
Recent data shows impressive revenue growth in AI coding agents, with Microsoft Copilot doubling to over $1 billion ARR and new competitors like Claude and Cursor reaching significant scale. (24:10) However, the critical question remains whether these companies can achieve gross margin profitability - meaning they make money on each additional sale rather than losing more money as they scale. If AI companies are spending $3 billion to generate $1 billion in revenue, even rapid growth becomes meaningless from an investment perspective.
Statistics & Facts
- Since the beginning of 2018, Nvidia earned approximately $205 billion of net income and $188 billion of free cash flow, but despite spending $112 billion on stock buybacks, there are 47 million more shares outstanding. (08:37)
- HSBC projects OpenAI's revenue will grow from $13 billion in 2025 to $214 billion in 2030, but the company will lose cumulative operating losses approaching half a trillion dollars, including $108 billion annually in both 2028 and 2029. (18:17)
- Microsoft Copilot doubled its Annual Recurring Revenue (ARR) from $500 million to over $1 billion since the beginning of the year, while Claude reached $1 billion ARR and Cursor achieved $800 million ARR, all since summer 2024. (24:10)
