AI Bubble Signals: Michael Burry Joins the Debate as Adoption Rates Flatten

Going Short on Generative AI

TLDR: Michael Burry, the legendary investor known for predicting the 2008 housing crisis, has joined Substack and is drawing attention to concerning signals in the generative AI market. New data from the Census Bureau and Ramp shows AI adoption rates flattening across all firm sizes, raising questions about whether LLMs are delivering real economic value.

Summary:

There's a fascinating intersection happening in the tech world right now between financial contrarianism and the AI hype cycle. Michael Burry, who famously shorted the housing market before the 2008 crash, has joined Substack and is already making waves in the AI bubble debate. His presence adds considerable weight to the growing chorus of skeptics questioning whether generative AI will deliver on its transformative promises.

The most compelling evidence comes from hard data rather than speculation. According to figures from the Census Bureau and Ramp, AI adoption rates have started to flatten across companies of all sizes. This is a significant development that doesn't fit the narrative of AI as an unstoppable general-purpose technology. If LLMs were genuinely boosting productivity and creating value, we'd expect to see accelerating adoption curves, not plateaus.

What's particularly interesting is the employment picture. General-purpose technologies historically create more jobs than they displace, at least in the medium term. Think about how the internet, despite initial fears, created entirely new categories of employment. The current data suggests generative AI isn't following this pattern. Companies aren't rushing to adopt it, and when they do, they're not necessarily expanding their workforce as a result.

For architects and technical leaders, this presents a strategic dilemma. There's genuine pressure to integrate AI capabilities into products and workflows, but the ROI picture remains murky. The smart approach might be to focus on specific, measurable use cases rather than broad AI initiatives. Where exactly in your stack can AI demonstrably reduce costs or improve outcomes? That's a different question from "how can we use AI?" and it deserves different treatment in roadmap discussions.

The broader lesson here is about technology evaluation in general. Burry's track record reminds us that markets can stay irrational for extended periods, but fundamentals eventually matter. The same principle applies to technology adoption. If the productivity gains don't materialize, the hype cycle will correct. Teams should maintain optionality, invest carefully in AI capabilities that solve real problems, and avoid betting the architecture on assumptions about future AI capabilities that haven't been proven yet.

Key takeaways:

• Michael Burry, known for predicting the 2008 crisis, is now actively discussing AI bubble concerns on Substack
• Census Bureau and Ramp data shows AI adoption rates flattening across all company sizes
• The absence of job creation around AI challenges the "general-purpose technology" narrative
• Technical leaders should focus on specific, measurable AI use cases rather than broad initiatives

Tradeoffs:

• Early AI adoption provides competitive positioning but risks investment in unproven technology
• Waiting for clearer ROI data reduces risk but may cause you to fall behind if AI does deliver
• Building AI-native architectures enables future capabilities but creates technical debt if assumptions prove wrong

Link: Going Short on Generative AI

The above content is AI-generated based on newsletter sources. Opinions expressed are interpretations of the original material.