5,048.62

The Nasdaq Composite crossed 1,000 for the first time in July 1995, the summer Netscape's IPO persuaded Wall Street that the internet was a money machine rather than a curiosity. What followed was five years of compression: every assumption about revenue, profitability, and time to maturity was squeezed until it snapped. Price-to-earnings ratios became irrelevant because earnings were absent; analysts substituted page views, unique visitors, and "eyeballs" — a term that entered business vocabulary with genuine seriousness during this period. The index crossed 2,000 in July 1998, 3,000 in November 1999, and 4,000 in December 1999. It added its fifth thousand in ten weeks.

The fuel was institutional and structural. The 1979 Department of Labor clarification of the ERISA prudent man rule had opened pension capital to venture funds in the early 1980s; by the late 1990s that capital had compounded into enormous funds chasing limited deals. The Small Business Investment Act of 1958 had created the infrastructure; Kleiner Perkins and Sequoia Capital had proved the model; two decades of positive returns had made "venture allocation" a line item in every endowment. When capital is abundant and institutional memory is short, prices rise until they cannot.

On 10 March 2000 the Composite closed at 5,048.62, a number so specific that it became, in retrospect, a precise marker of collective delusion. The day itself was unremarkable in terms of news. No single announcement broke the mood. The index simply reached a point at which there were more sellers than buyers, and it did not recover that level for fifteen years.

The decline was neither gradual nor orderly. The Composite lost roughly 60 percent in the first year after the peak, then continued lower as the losses propagated from pure-play internet names into the telecommunications infrastructure companies that had borrowed billions to build the pipes those names supposedly needed. By the trough — the intraday low of approximately 1,114 on 10 October 2002 — the index had shed 78 percent of its peak value, erasing an estimated five trillion dollars in market capitalisation across US equities.

The companies that disappeared did so by multiple mechanisms. Some filed for bankruptcy directly: Webvan, the grocery-delivery operation that raised more than $375 million in venture capital and $375 million more in its November 1999 IPO before shutting in July 2001, is among the cleanest examples of cash incineration at industrial scale. Pets.com, which spent heavily on brand advertising before its revenue model collapsed, liquidated in November 2000, less than ten months after its IPO. Both had been treated as infrastructure plays — essential services whose inevitability justified front-loaded spending — and both discovered that the market would not wait for the future to arrive.

Others were acquired for fractions of their peak valuations. Broadcast.com, which Mark Cuban and Todd Wagner sold to Yahoo in 1999 for $5.7 billion in stock at the absolute height, was shut down by Yahoo in 2002 after the acquiring stock lost most of its own value. The transaction became a textbook example of an acquihire gone wrong at scale, in which the currency of the deal — Yahoo equity — collapsed before integration could produce anything. Kozmo.com, eToys, drkoop.com, and several hundred other names with varying degrees of product viability followed their own paths to the same outcome.

What remained was structurally different from what had gone in. Amazon, which had listed in May 1997 and watched its stock fall from a split-adjusted peak of roughly $113 in December 1999 to under $6 by September 2001, survived because Jeff Bezos had built distribution infrastructure rather than marketing spend. The fall was severe enough that serious analysts questioned whether the company would survive; it did, and its subsequent trajectory made the 2000 peak look like a rounding error. eBay retained its marketplace network effects. Priceline, widely mocked in 2000, remained operational.

The dividing line was not simply sector or product category but the relationship between cash and time. Companies that had assumed continuous capital access — that a Series B would always be followed by a Series C, and that a C would always be followed by a public offering — discovered in 2000 and 2001 that the assembly line had stopped. Burn rate and runway, which had been politely discussed in board meetings during the ascent, became existential figures overnight. A company with eighteen months of cash could wait out the correction; one with four months could not.

The destruction produced several lasting institutional effects. The venture industry contracted sharply: according to the National Venture Capital Association, commitments to US venture funds fell from approximately $105 billion in 2000 to roughly $5 billion in 2003. Funds raised at peak valuations faced the J-curve with no exit market to recover it; limited partners who had over-allocated to the asset class spent years working down that exposure.

The Sarbanes-Oxley Act of 2002 — passed in the aftermath of Enron and WorldCom as well as the broader market failure — imposed new auditing and corporate governance requirements on public companies that made the IPO path more expensive and compliance-intensive. Its most direct effect on the startup ecosystem was to extend the period between founding and public offering: if going public meant Sarbanes-Oxley compliance, staying private longer became rational. The median time from founding to IPO, which had been roughly four years in the late 1990s, began its extension toward the seven-to-ten-year window that characterised the next cycle.

The telecommunications infrastructure that the bubble had over-built — fibre optic cable, router capacity, data centre space — remained in place and priced near zero. This created the physical substrate for the next generation of internet companies, which built on cheap bandwidth and abundant server capacity that they had not paid to install. Google, which listed in August 2004, and the cohort that followed it operated in a world where the marginal cost of delivery had collapsed precisely because the previous cycle had over-invested in that delivery.

The phrase "irrational exuberance" — Alan Greenspan had used it in December 1996 to describe what he saw even then as elevated asset prices, three full years before the peak — acquired its definitive meaning retroactively. The Composite did not close above its March 2000 closing peak again until 23 April 2015, a gap of fifteen years and forty-four days. It crossed its March 2000 peak for the first time in intraday trading on 27 April 2015, also fifteen years after the fact. The number 5,048.62, for anyone who lived through the correction, carries the same mnemonic weight that specific disaster figures always carry: precise enough to be real, high enough to have mattered.