Internet Infrastructure Bubble

The Internet Infrastructure Bubble was the telecom and networking wing of the late-1990s tech boom, driven by aggressive investment in capacity. Its center of was the rapid expansion of long-haul fiber cables, data centers infrastructure, optical transport, and the carriers and suppliers expected to monetize the exploding internet usage. The boom was real in the sense that demand for connectivity was genuinely rising, internet use was accelerating, and new fibre technologies materially increased capacity. But the sector’s valuations and capital commitments were built on the belief that traffic, revenues and margins would rise much faster than they actually did. When those assumptions failed, the result was one of the largest capital-spending reversals and bankruptcy clusters in modern telecom history.

The bubble began with a plausible premise. The Telecommunications Act of 1996 promised more competition, while fiber-optic technology advances suddenly made it possible to increase capacity dramatically. At the same time, internet use was escalating rapidly, and the industry came to believe that one converged IP-based network would carry all services (instead of separate networks: one for phone calls, one for internet data, one for TV/video). The combination of regulatory change, technological advance, and fast-rising internet use created the investment surge and high valuations that defined the telecom boom.

The problem was that the sector started to plan from slogans rather than from disciplined measurement. The widely repeated claim that internet traffic was doubling every three or four months was badly overstated. The traffic was roughly doubling annually, not at the 700 to 1,500 percent yearly rates implied by the myth. But, at that time, aggregate data on actual internet use was difficult to obtain, and the market’s optimists drove valuations anyway.

Cheap capital then did the rest. OECD found that the new regulatory environment and technological innovation excited investors enough to generate large flows of equity issuance, debt flotation, and bank credit. With abundant funds, carriers financed vast fiber backbones and paid spectacular sums for 3G coverage. The plans were huge: 360networks projected a 56,300-route-mile network across North America, Europe, and South America, including major submarine cables; Level 3 by the end of 1999 had completed 9,334 North American route miles, had 6,200 more under construction, and was racing to expand gateway and colocation capacity. [

What failed was not traffic growth, but the price/revenue equation. OECD reported that during 1996 to 2001 telecom revenues in the OECD grew only 7.2% annually on average, slowing to 1.6% in 2001, while infrastructure investment reached nearly $230 billion in 2000. Meanwhile, liberalization and competition pushed bandwidth prices down rapidly, especially in Europe. So carriers were building into a world where throughput might rise, but unit prices and margins were falling.

By 2001 the dislocation had become visible in hard capacity data. The FCC reported that eight new cables were added in 2001 and that reported activated circuits accounted for only 7.9% of total available cable capacity. In the trans-Atlantic region, activated plus idle circuits accounted for only 21.6% of total available capacity. That did not mean the cables were useless; it meant the industry had built far ahead of its monetizable demand.

The endgame was characterized by a combination of structural overcapacity, high leverage, and increasingly opaque financial reporting. As network buildouts far exceeded actual demand, several major firms relied on aggressive accounting practices to sustain the appearance of growth and profitability. Global Crossing, for example, extensively used reciprocal IRU (Indefeasible Rights of Use) transactions to inflate revenues and earnings; according to the SEC, without these arrangements, the company would have fallen significantly short of analyst expectations, with its reported adjusted EBITDA for Q2 2001 shifting from a positive $472 million to a loss of $43 million. Similarly, WorldCom engaged in improper accounting by reclassifying billions of dollars in routine line costs as capital expenditures, thereby deferring expenses and artificially boosting short-term profits. Once investors and creditors figured out the numbers, insolvency followed quickly.

References:

Optica – Optical Fiber Communications and the Internet Boom (PDF)

FRED – NASDAQ Composite Index Data

Federal Reserve Bank of Richmond Economic Quarterly (April 2003 PDF)

OECD – After the Telecommunications Bubble (PDF)

FCC – In the Matter of Review of the Section 251 Unbundling Obligations of Incumbent Local Exchange Carriers (Order PDF)

SEC – Order Instituting Cease-and-Desist Proceedings, Exchange Act Rel. No. 34-51517

Richmond Fed – Boom and Bust in Telecommunications (Fall 2003 PDF)

Understanding the Digital Divide – OECD Report (PDF)

The market impact was severe. The Richmond Fed concluded that telecom accounted for a larger share of market capitalization gained and lost than the dot-coms. Using the NASDAQ Composite as a public-market proxy, the benchmark rose from 1,499.25 on August 31, 1998 to 5,048.62 on March 10, 2000, then fell to 1,114.11 by October 9, 2002.

Credit markets were hit almost as hard as equities. OECD reports that global corporate bond defaults reached $163 billion in 2002 and that 56.4% of that total came from telecommunications. It also describes the episode as the largest cycle of telecom bond defaults since the 1930s. The sector’s bust was therefore not only a stock-market event; it was a debt-market event with significant implications for lenders, bondholders, and refinancing conditions.

The macro impact was paradoxical. OECD found that the telecom sector’s direct weight in national economies was only about 2 to 4% of GDP, so the immediate hit was limited. But the upstream effects on equipment makers and technology suppliers were severe, and the equity-market and credit-market wealth destruction was enormous. This is one reason the bubble remains such a powerful case study: it combined a modest direct GDP weight with very large financial and industrial repercussions.

The network-industry trap: more usage does not automatically mean more profit.

Technological advancements can still be poor investments when early capital is mispriced and overly leveraged.

Infrastructure-heavy, debt-financed business models become fragile when demand timing and price assumptions prove wrong.