El Podcast

E168: AI - Biggest Bubble in Human History? Tech Economist Says YES

Episode Summary

Tech economist Dr. Jeffrey Funk explains why the current AI surge is the largest bubble in history—driven by massive losses, cheap subsidized pricing, and trillions in infrastructure spending with almost no real returns. He warns that when the narrative finally collapses, the fallout could dwarf dot-com and the 2008 crisis, reshaping tech, energy, and investment for years.

Episode Notes

Tech economist Dr. Jeffrey Funk argues that today’s AI boom is the biggest bubble in history—far larger than dot-com or housing—because colossal infrastructure spending is chasing tiny, unprofitable revenues.

Guest bio:

Jeffrey Funk is a technology economist and author of Unicorns, Hype and Bubbles: A Guide to Spotting, Avoiding and Exploiting Investment Bubbles in Tech. A longtime researcher and professor of innovation and high-tech industries, he now writes widely on startup hype, AI economics, and investment manias, including a popular newsletter and presence on LinkedIn.

Topics discussed:

• Why Funk thinks the AI boom is the “biggest bubble ever”
• OpenAI’s revenues, mounting losses, and opaque accounting vs. Microsoft’s audited numbers
• Nvidia, cloud providers, and “circular finance” in AI infrastructure
• Sora, video generation, and the economics of ultra-expensive AI features
• Comparisons with the 1929 crash, the dot-com bubble, and the 2008 housing crisis
• How much of AI is real utility vs. hype, scams, and accounting tricks
• Hallucinations as an inherent limitation of large language models
• World-model approaches, quantum computing, and why breakthroughs are harder than advertised
• Energy use, exploding electricity demand, and Bill Gates’ shifting climate rhetoric
• Possible winners after the bubble: why it’s still “wide open”
• Labor markets, layoffs, and why “AI took their jobs” is mostly a PR story
• College and career advice for young people in an AI-saturated economy
• China, regulation, and small language models
• What the pop might look like: shuttered data centers, broken pensions, and a long VC winter
• Final advice: how to think more clearly about tech futures and bubbles

Main points:

• Investment vs. returns: A bubble is simply when more money goes into companies than comes out; by that standard, AI is extreme—OpenAI’s losses and projected $115B cash burn dwarf its revenues.
• Subsidized demand: OpenAI’s ultra-low prices and free tiers artificially inflate usage and pump up Nvidia and cloud revenues; if prices reflected true cost, demand (and infra spending) would fall sharply.
• Accounting red flags: Discrepancies between OpenAI’s figures and Microsoft’s audited statements, plus aggressive depreciation assumptions for AI chips, echo Enron-style financial engineering.
• Bigger than past bubbles: Unlike dot-com, where consumers paid for internet access, PCs, and e-commerce (≈$1.5T in 2024 dollars), AI currently generates tiny, niche revenues relative to the trillions being poured into infrastructure.
• Tech limits: LLM hallucinations are a built-in feature of statistical generative models, not a temporary bug; GPT-5 and similar systems haven’t solved this, and world-model or quantum fixes would be extremely costly and distant.
• Real but narrow use-cases: AI can help with things like drafting emails, simple ads, and some coding assistance, but broad productivity gains across manufacturing, construction, healthcare, etc., remain largely unrealized.
• Jobs & layoffs: Headlines about AI-driven mass unemployment are mostly hype; unemployment overall is low, many “AI layoffs” are reversals of pandemic over-hiring, and outsourcing plus H-1B dynamics matter more than LLMs.
• Crash mechanics: When the narrative finally flips and big investors (like Michael Burry) exit or short AI, overbuilt data centers, utility expansions, and VC portfolios will be left stranded, hurting pensions and index investors.
• Careers & education: Young people should be skeptical of hype, but still learn math, coding, and predictive AI; trades and biotech remain attractive, and the key skill is learning to reason about trends instead of chasing bandwagons.

Top 3 quotes:

On what a bubble really is:

“When people are putting more money into companies than they’re getting out, it becomes a bubble. It’s just exaggeration.”

On Nvidia, cloud, and OpenAI’s losses:

“Who cares if Nvidia and the cloud providers are making so much money if OpenAI is losing billions to subsidize them? The car might be selling, but if you’re selling it for half price, it’s not a good business.”

On how young people should respond:

“If you’re young, don’t worry too much about the bubble. Be open-minded, be curious, learn to think for yourself instead of believing what the tech bros say, and things will work out.”