Webvan

Louis Borders, who had co-founded the Borders bookstore chain, began raising capital in December 1996 for a company initially incorporated as Intelligent Systems for Retail, which would become Webvan. The pitch was ambitious even by dot-com standards: instead of building a website that let people order groceries for pickup or next-day delivery, Webvan would build its own automated, highly engineered distribution centers and deliver orders to customers' doors within a 30-minute window they chose themselves — a level of service few even brick-and-mortar grocers offered.

In September 1999, Borders persuaded George Shaheen, then the CEO of Andersen Consulting (later renamed Accenture), to leave a secure, high-profile position to become Webvan's president and CEO — a hire the press treated as a signal that Webvan was being run by serious, professional management rather than typical dot-com founders.

Webvan raised its earliest capital from Benchmark Capital and Sequoia Capital, each investing $3.5 million alongside Borders himself in a 1997 seed round. Sequoia later put in another $50 million, and by 1999 the company had pulled in a further $335 million from investors including SoftBank ($160 million) and Goldman Sachs' venture arm ($50 million). In November 1999, Webvan went public, raising roughly $375 million in its IPO and reaching a market valuation of $7.9 billion on its first day of trading — despite the company itself projecting a $65 million loss for the year.

All told, the company raised more than $770 million in private and public capital in just a few years — an extraordinary sum for a company that had not yet proven it could deliver groceries profitably in a single city.

Webvan's central strategic bet was that automation and scale would eventually make grocery delivery profitable, so the company invested heavily in building its own highly automated distribution centers — each costing an estimated $30 to $50 million — rather than starting with a smaller, cheaper operation and iterating. The technology inside these warehouses had, in several cases, not been fully tested before construction began, and the company found itself abandoning or reworking some automated features after operations were already underway.

Rather than perfecting the model in one or two markets first, Webvan expanded rapidly into 26 U.S. cities, signing a $1 billion, ten-year contract with construction firm Bechtel to build out its national warehouse network before any single market had demonstrated it could operate profitably. Warehouses frequently ran at less than 30% of their designed capacity, meaning the fixed cost of each facility was being spread across far fewer orders than the model required.

The core financial problem was simple to state and impossible for the company to solve in time: it cost Webvan more than $100 to acquire each new customer, while the average customer's lifetime value came in under $80. Every new customer the company signed up was, on average, a loss before a single grocery order was even delivered — and Webvan was spending heavily on marketing to acquire customers at that rate across two dozen cities simultaneously. The company never achieved positive margins in even its best-performing market, let alone the network as a whole.

Webvan's stock, worth roughly $30 a share around its IPO, fell below $1 within about a year as losses mounted and it became clear the path to profitability the company had promised investors did not exist at its current scale or cost structure. In July 2001 — just 20 months after going public at a $7.9 billion valuation — Webvan filed for Chapter 11 bankruptcy and laid off more than 2,000 employees, shutting down operations entirely rather than attempting a smaller-scale restructuring.

Webvan remains one of the dot-com era's most-cited cautionary tales precisely because its underlying thesis — that people would eventually want groceries delivered to their door on a schedule they control — turned out to be correct; companies including Amazon (which later built out its own grocery delivery infrastructure, reportedly drawing on lessons from and even some assets connected to Webvan's collapse) and Instacart proved the model could work, just years later and with dramatically different unit economics made possible by smartphones, gig-economy labor, and a public far more comfortable buying groceries online. Webvan's failure wasn't a failure of vision; it was a failure of sequencing — building enormously expensive, purpose-built infrastructure at national scale before validating that the underlying unit economics could ever work, and then treating aggressive geographic expansion as a fix for a profitability problem that expansion actually made worse.