Is the AI Boom About to COLLAPSE?
Ed Zitron: The Rot Economy and the End of Hypergrowth
The tech industry has been operating under a 'rot economy' or 'growth at all costs' mindset for years, driven by subscription models and app monetization. However, the era of hypergrowth is ending, and AI is merely a symptom of this larger problem. Past hype cycles like Clubhouse, the metaverse, and NFTs failed to deliver on their grand promises, indicating a potential systemic issue with how the industry pursues innovation and investment.
Chris Hayes: AI's Tangible Use Cases vs. Metaverse Hype
Unlike the metaverse, which lacked clear use cases and user adoption, AI, particularly large language models (LLMs), demonstrates tangible applications such as document review and synthesis. While the effectiveness and reliability of these applications are still debated, the ability to articulate specific functions for AI distinguishes it from previous speculative tech trends.
Chris Hayes: Useful Tech vs. Justified Investment
The usefulness of a technology, like railroads or the internet, does not guarantee that the financial investment in it during a boom period is justified. Both railroads and the internet were transformative but experienced significant crashes due to overbuilding and speculation. The current AI investment frenzy may follow a similar pattern, where the technology's eventual utility doesn't prevent a market correction.
Ed Zitron: The Astronomical Cost of AI Compute
The AI industry, particularly companies like Anthropic, faces immense computational costs, spending billions on renting GPUs and custom silicon for training and inference. For instance, Anthropic reportedly spent $10 billion on training and inference through March 2026, while only projecting $5 billion in revenue. This massive expenditure, often subsidized by hyperscalers, makes the current business model highly questionable.
The Unsustainable AI Compute Subsidy
The current AI boom is characterized by massive subsidies, with companies like OpenAI and Anthropic spending billions more than they earn. This is driven by the immense cost of compute, which is significantly higher than previous technological infrastructure investments, creating a precarious financial situation.
Nvidia's Paradoxical Role
Nvidia's position as the dominant chip manufacturer is complicated by its deals to support data center construction and its role as a major customer for companies like Coreweave. This creates a self-serving dynamic where Nvidia profits from selling chips and also from the infrastructure that buys them, blurring the lines of a free market.
The Perilous Data Center Debt Bubble
A significant amount of debt ($178.5 billion in data center deals) is being raised by new companies to build AI data centers, often with unproven demand and questionable business models. This debt is largely 'crap' due to poor due diligence, creating a potential financial crisis if demand doesn't materialize.
Venture Capital and Private Equity Woes
The broader venture capital and private equity markets are struggling, with low returns and difficulty selling companies, particularly in the software sector. This lack of liquidity and profitability in investment spheres exacerbates the financial strain on the capital-intensive AI industry.
The Illusion of AI Replacing High-Skill Labor
The promise of AI replacing high-paying jobs, like law firm associates, is largely based on future capabilities ('coulds' and 'wills') rather than current reality. AI models currently struggle with nuanced judgment, risk management, and legal precision, often leading to errors like hallucinated citations, making them unreliable for critical tasks.
The Broken Scaling Law and Diminishing Returns
The core assumption that more compute directly leads to better AI performance (scaling laws) appears to be breaking down. Companies are hitting diminishing returns in pre-training, and while inference costs might decrease for some models, overall usage (tokens) is increasing, keeping companies unprofitable.
The Inevitable Cataclysm: Profitability or Mass Unemployment
The AI boom faces a dilemma: either it remains unprofitable and collapses, or it becomes profitable by replacing millions of jobs, leading to a societal and economic cataclysm. The math for profitability requires revolutionary advancements and trillions in new revenue, which seems implausible in the short term.
Ed Zitron: The AI Bubble's Debt Foundation
The current AI boom is not driven by sustainable growth but by an unsustainable amount of debt and speculative investment, creating a financial bubble poised for collapse. Companies are raising billions based on future potential rather than current profitability, leading to a precarious market situation.
OpenAI's Unprecedented Bet: Oracle's Risky Alliance
The financial survival of OpenAI, which is burning billions, is critically dependent on Oracle's massive investment in data centers. Oracle's bet on OpenAI achieving unprecedented revenue growth by 2030 is mathematically precarious, as Oracle itself carries substantial debt and negative cash flow, making its future tied directly to OpenAI's success.
