Michael Burry compares AI capex boom to dot-com bubble, warning on circular-financing risks and valuation extremes.

Michael Burry believes one of the most important investing lessons of his career didn't come from predicting the housing crash, but from an observation years earlier during the dot-com era. After watching an interview with venture capitalists Marc Andreessen and Ben Horowitz, Burry revisited his early years in Silicon Valley and the framework that has informed his view of technology bubbles ever since—one he argues remains directly relevant to today's AI infrastructure boom.

Burry's journey began in the late 1980s and early 1990s, a period he describes as "an exhilarating and wondrous time" to be in Silicon Valley. After winning an IBM Thomas J. Watson Scholarship in high school, he attended UCLA and secured summer internships at IBM's research facilities in Silicon Valley, where he stress-tested hard disk drives and learned to analyze data using APL. Beyond his formal work, Burry spent summers hunting for discounted computer parts, building custom PCs, and participating in the pre-internet bulletin board system (BBS) community. He was witnessing firsthand the emergence of personal computing and networking—and observing the inner workings of one of the world's largest technology companies.

The turning point came in 1998, when Burry returned to Silicon Valley and found IBM's once-bustling Cottle Road campus largely deserted. "That something so huge, vibrant and important could disappear so fast greatly informed my interpretation of the 1998-2000 bubble as I lived it from my workplace on Sand Hill Road," he wrote. The rapid decline of a company that had once anchored the technology industry reinforced his skepticism toward market euphoria.

This observation crystallized into a principle. When Burry later researched WorldCom in 2000, he questioned whether the telecom company's earnings justified its acquisition-driven growth after accounting for its cost of capital. From that analysis emerged what Burry calls a "timeless theme": investors should ask a simple but critical question of any company undertaking massive spending campaigns: "What will the ROIC be?" Return on invested capital, he suggests, is the discipline that separates genuine value creation from bubble-era excess.

Today, as companies pour billions into AI infrastructure, Burry's framework remains unchanged. The same investing discipline that allowed him to spot the hollowness of the dot-com and WorldCom booms applies directly to evaluating whether today's AI spending will generate returns or merely reflect market euphoria.