The Telecom & Fiber Bubble
Less famous than the dot-coms, and more important for the AI question. Overbuilt infrastructure plus suppliers funding their own customers: this is the research corpus's closest structural match to today.
Updated
What happened
While the dot-coms grabbed headlines, telecom carriers were doing the physical work of wiring the internet, and overdoing it on a heroic scale. Between 1995 and 2001, global telecom and fiber investment ran to an estimated $2 trillion, laying 80-90 million miles of fiber-optic cable (Internet History (opens in a new window)). US telecom capital spendingMoney a company spends on long-lived physical things: buildings, machines, chips, data centers. Spent now, paid back (hopefully) over years.Full definition in the glossary peaked around $213 billion in 2000, roughly 1.0-1.2% of US GDP (Richmond Fed (opens in a new window)). At the 2001-02 trough, industry estimates put as much as ~95% of long-haul fiber "dark": installed but carrying no traffic (LA Times (opens in a new window)).
What inflated it
The distinctive fuel was vendor financingA supplier lending customers the money to buy its own products. Inflates the supplier's sales until the customers can't pay.Full definition in the glossary: equipment makers lending customers the money to buy their own gear. By end-1999, nine major vendors (including Lucent, Nortel, and Cisco) had at least $25.6 billion of such loans outstanding, roughly double the 1996 level, and McKinsey judged 30-40% of it "at risk" because the borrowers, mostly upstart carriers with no profits, could not realistically pay (CNET/McKinsey (opens in a new window)). For the five North American vendors, outstanding vendor loans equaled 123% of their pretax earnings. Vendors booked the sales as revenue immediately; the risk sat quietly in receivablesSpreading the cost of an asset over its useful life in the accounts. Stretch the assumed life and reported profits go up, with no change in the business.Full definition in the glossary. Post-mortems argue that without this loop, sector growth would have stalled around 1998-99 instead of peaking in 2000: the loop bought the boom an extra one to two years (American Affairs (opens in a new window)). The full mechanism is unpacked on Circular Financing, Explained.
What popped it
The dot-com downturn cut demand forecasts, and then the loop ran backward. Financed customers (Winstar, One.Tel, various startups) began failing in 2000-2001, forcing the vendors to write down their loans: Lucent took a $501 million vendor-finance charge in fiscal 2000, about 41% of that year's earnings, and repossessed gear fetched as little as 10 cents on the dollar (CNET (opens in a new window)). Credit spreadsThe extra interest a risky borrower pays compared to the safest borrower (the US government). Widening spreads mean lenders are getting nervous.Full definition in the glossary on telecom debt blew out, funding closed, and the bankruptcies of Global Crossing and WorldCom in 2002, the latter wrapped in accounting fraud, finished the job (Richmond Fed (opens in a new window)).
How far it fell, and how long recovery took
The Nasdaq fell 78%; telecom equipment names fell 90-99%, with the heaviest users of vendor financing hit hardest. Lucent and Nortel never recovered as independent companies. US telecom capex roughly halved within two years. Sector indices took a decade or more to recover, and only after consolidation had transferred the assets to new owners.
Who captured the value
The fiber itself proved immensely valuable, on someone else's balance sheet. Bankruptcy buyers picked up entire multi-billion-dollar networks for cents on the dollar, and as traffic caught up over the following 5-10 years, the "worthless" capacity became the physical foundation of streaming, cloud computing, and the modern web (Internet History (opens in a new window)). First-wave investors funded it; second-wave owners and everyone who uses the internet collected.
Echoes in today's AI boom
- The financing rhyme is direct. Nvidia investing billions in CoreWeave and OpenAI while they commit billions back for chips is the equity-flavored descendant of Lucent lending to carriers, with important structural differences argued both ways. Compare The AI Money Loop with the defenders' rebuttals.
- The capex scale is bigger this time. Telecom peaked at 1.0-1.2% of US GDP; AI-related investment was estimated near 5% by 2026. Side-by-side numbers on Building Ahead of Demand.
- One difference cuts against the optimists. Dark fiber could wait a decade for demand because glass lasts decades. GPUs lose most of their market value in 3-5 years, so AI's overcapacity, if that is what it is, cannot wait nearly as long. See the GPU depreciation controversy.
- Watch the write-down. Telecom's turning point was Lucent admitting its financed customers couldn't pay. The equivalent today would be a chip maker or cloud writing down an equity stake in a customer or tearing up an offtake agreementA long-term contract to buy a project's future output, signed before it's built. Lenders rely on these promises, so the buyer's health becomes the project's foundation.Full definition in the glossary, one of the confirmation signals listed on Conclusions.