We Always Overshoot: From Fiber to Cannabis to AI

Every generation believes it has discovered the “next big thing.” Railroads in the 1870s, fiber optics and the dotcom’s irrational exuberance in the 1990s, cannabis in the 2010s, and today’s artificial intelligence boom all shared the same pattern: investors raced ahead of real demand. The technology was never fake. The timing was.

I watched two of these booms from inside the industry: the dot-com fiber glut and the Canadian cannabis overbuild. In both cases, exuberance pushed supply far beyond what the market could handle, destroying capital but laying foundations for future growth. I spotted the cannabis overshoot a year before the crash because I lived through the fibre glut. I see the same signals now in AI.

A Familiar Pattern

In the late 1990s, the “information superhighway” was the rallying cry. Companies raced to trench streets and string fiber, convinced the internet would consume bandwidth faster than anyone could build it. I remember telling my staff that the telcos were throwing money out the window of their ivory towers and our job was to catch as much as we could.

The scale was breathtaking. We knew internet traffic was growing fast, but we also knew the physics of laying fiber: once you’ve dug the trench, adding more strands costs very little. The result was a massive overshoot of capacity.

By 2002, after the bubble burst, only about 3% of U.S. fiber capacity was actually lit. By 2005, it was still under 5%, and most of those gleaming strands sat “dark” for years. Lighting them required investment in costly electronics, and demand wasn’t anywhere near the levels promised in the pitch decks of 1999. By 2012 only half of that laid fibre was being used to carry traffic.

The consequences for telecom carriers were brutal. Giants like Global Crossing and WorldCom went bankrupt. Investors lost billions.

Those once-dark fibers became the backbone of the modern internet. It took over a decade, but the glut eventually turned into the infrastructure we rely on today.

The Cannabis Overbuild in Canada

Fast-forward to 2017. Canada was on the brink of legalizing recreational cannabis, and the capital markets responded with a wave of exuberance. Billions of dollars poured into greenhouses, indoor cultivation facilities, and grandiose promises of global domination.

I was close to that market too, and the mismatch was obvious. Demand was real, but projections were wildly inflated. Consumer adoption was steady, not exponential, and regulations slowed distribution. Yet companies raced to build capacity as if every Canadian were going to double their consumption overnight.

That summer, I was warning publicly that there was far more production capacity than the market could absorb. Within a year, the crash arrived. Wholesale cannabis prices collapsed. Many facilities were mothballed before they ever planted a single crop. Companies wrote off billions, and stock prices cratered.

The lesson was the same as with fiber: the product was real, the enthusiasm understandable, but the buildout raced far ahead of demand.