Software as a service

In 1999, enterprise software was a physical artifact delivered on CD-ROMs, installed on local servers, and requiring six to twelve months to deploy. A low-priced CRM system for 200 employees cost $1.8 million in the first year alone—hardware, network infrastructure, licensing, consultants. When Marc Benioff left his thirteen-year career at Oracle to start Salesforce in a one-bedroom San Francisco apartment, he was not inventing new technology. He was recognizing that existing technology had crossed a threshold.

The web browser had become ubiquitous. Internet connections, while still slow, were reliable enough for business applications. Payment processing infrastructure could handle recurring subscription billing. Server costs had dropped to the point where hosting applications centrally made economic sense. The question was not whether software could be delivered over the internet—Application Service Providers had been attempting this since the mid-1990s—but whether a company built from the ground up around this model could outcompete the incumbents.

Salesforce launched its first prototype in November 1999, deliberately designed to look like Amazon with tabs across the top. The comparison was strategic: Amazon had proven that complex transactions could happen through web interfaces. If people could buy books online, they could manage customer relationships online. Benioff's genius was marketing as much as technology. His "End of Software" campaign, including a staged protest outside a Siebel Systems conference in February 2000, positioned SaaS not as an incremental improvement but as an existential threat to the old model.

The economics created unstoppable momentum. Traditional software required massive upfront investment and created lock-in through switching costs. SaaS inverted both dynamics: low initial cost made adoption frictionless while continuous updates made the service increasingly sticky through accumulated customization and data. The subscription model aligned vendor and customer incentives—if the software stopped delivering value, customers could leave. Path dependence that had protected incumbents now worked against them.

Salesforce's IPO in 2004 proved the model. The AppExchange launch in 2005 added ecosystem dynamics—third-party applications built on the platform created network effects that traditional software could not match. By 2024, Salesforce held 24% of the global CRM market, more than the next four competitors combined. The apartment prototype had become the dominant architecture for enterprise software, and the CD-ROM had become a historical curiosity.