WorldCom's traffic myth helped inflate a bubble, and the data always said so
Jim Chanos argued that internet traffic was doubling annually in the late 1990s, not quarterly as MCI WorldCom claimed. Independent research confirms the "doubling every 100 days" figure was a fabrication, one that helped justify hundreds of billions in overinvestment.
The number that helped inflate one of the largest capital misallocations in American corporate history was not an accident. Jim Chanos, the short-seller who tracked MCI WorldCom’s finances through the boom years, put the core deception plainly: “Internet was really growing fast. It was doubling every year, not every quarter.”
That distinction is not a rounding error. A network doubling every quarter would require infrastructure investment orders of magnitude beyond what annual doubling demands. If WorldCom and its UUNet subsidiary were telling analysts, investors, and competitors that traffic was doubling every hundred days, they were describing a market roughly sixteen times larger, per year, than the one that actually existed. Capacity build-out premised on that figure would not be oversupply. It would be a structural error baked into the investment thesis from the start.
The external record supports Chanos. Researchers Andrew Odlyzko and Kerry Coffman examined backbone traffic data in the late 1990s and found growth rates consistent with annual doubling, not quarterly. The Economist, in coverage of the WorldCom collapse, identified the “doubling every 100 days” claim as a myth that WorldCom’s UUNet operation had promoted and that the broader industry had uncritically repeated. Trade publications and network operator communities that tracked actual backbone measurements found no evidence the quarterly-doubling rate was real. The figure circulated because it was useful, not because it was true.
Internet was really growing fast. It was doubling every year, not every quarter. Jim Chanos
What makes the WorldCom case study durable is the mechanism, not just the outcome. A single large player with credible-sounding data can set the terms of an entire industry’s planning assumptions. Competitors who doubted the traffic projections faced a difficult choice: either build at scale to match WorldCom’s stated premise, or risk being left behind if the figure turned out to be correct. Skepticism carried a competitive cost. That asymmetry rewards the party willing to assert an inflated number and punishes the parties that demand evidence before committing capital.
Chanos was, at the time, a named skeptic of WorldCom’s accounting. His broader argument was that the company’s reported revenues and capital expenditure were not coherent, that the model required perpetually accelerating growth to service the debt load the expansion was generating. The traffic claim was one load-bearing pillar of that model. If growth was annual rather than quarterly, the addressable market for WorldCom’s capacity was a fraction of what the company’s valuation implied. The fiber being laid across the country was being priced for a demand curve that did not exist.
The practical lesson for any period of fast-moving infrastructure investment is not that growth figures should be distrusted wholesale. Annual doubling of internet traffic is, by itself, a remarkable rate. The lesson is that the difference between doubling annually and doubling quarterly is not a matter of enthusiasm or framing. It is a specific empirical claim that can be checked against backbone measurement data, and in WorldCom’s case, the data were available to anyone who looked. Odlyzko and Coffman were looking. The problem was that the incentive structure of the boom made careful measurement less profitable than credulity.
WorldCom filed for bankruptcy in 2002 in what was then the largest such filing in American history. The fiber overbuilding it helped catalyze took years to clear. The traffic myth was not the only cause of either outcome, but it was the kind of foundational claim that, once accepted as given, made the rest of the financial architecture look coherent from a distance. Chanos’s point is a simple one, and the Odlyzko research gives it a factual floor: the internet was genuinely growing, and that growth was still being materially overstated by the company with the most to gain from the overstatement.