History of SaaS

Software as a Service is often described as a product of the 2000s internet era, but its roots go back to the 1960s. The core idea — that software should be delivered as a service over a shared infrastructure rather than installed on individual machines — predates the internet. What changed across six decades was the infrastructure, the economics, and the scale at which the model could operate.

This is the complete history of SaaS: from mainframe timesharing to Salesforce, from AWS to the AI era.

🖥️ The Pre-History: Timesharing (1960s–1980s)

The first commercial software-as-a-service model was mainframe timesharing. In the 1960s, mainframe computers cost millions of dollars and only large corporations or universities could afford them. Computing bureaus emerged to sell computing time to companies that could not justify owning a mainframe outright — customers paid for processing time on a shared machine accessed via remote terminals.

Companies like Tymshare (founded 1966) and National CSS built businesses on this model. Customers ran payroll, accounting, and scientific calculations on shared infrastructure they accessed over telephone lines. The pricing model was pay-per-use — minutes of CPU time rather than software licenses. The delivery was remote — software running on the bureau's hardware, not the customer's.

This model declined in the 1980s as minicomputers and then personal computers made local computing affordable. The economic case for timesharing disappeared when a PC could do the same work for a fraction of the cost of remote access. The infrastructure shifted from shared to individual — and the software industry shifted to packaged software: buy a disk, install locally, pay a license fee.

🌐 The Internet Changes the Economics (1990s)

The commercial internet of the 1990s created the infrastructure for the timesharing model to return at vastly greater scale. Browsers, HTTP, and TCP/IP meant that software could be delivered to any machine with an internet connection — without requiring a modem, a proprietary network, or a specialized terminal.

Early web-based software in the 1990s was not called SaaS. It was called Application Service Providers (ASPs). Companies like Employease (HR software), Salesnet (CRM), and NetLedger (accounting) built web-accessible software sold on a subscription basis. These were genuine SaaS businesses by modern definition — but the term did not exist yet, and the infrastructure was limited.

The 1990s ASP market had fundamental constraints: bandwidth was slow, browsers were inconsistent, and cloud infrastructure did not exist. Companies ran their own servers or rented dedicated hosting. The economics of multi-tenancy — multiple customers sharing one infrastructure — were not yet achievable at the scale that would make the model dominant.

🚀 The Founding of Modern SaaS: Salesforce (1999–2005)

Salesforce, founded by Marc Benioff in 1999, is the company that defined modern SaaS. Not because it was the first web-based software business — it was not — but because it named the category, challenged the incumbent model explicitly, and proved at scale that enterprise software buyers would adopt subscription-based cloud delivery.

Benioff's "No Software" campaign was a direct attack on Siebel Systems, the dominant CRM vendor of the 1990s, which required expensive on-premise installations, lengthy deployments, and large IT teams to maintain. Salesforce offered a browser-based alternative: no installation, subscription pricing, and an interface any salesperson could use.

The early Salesforce pitch had to overcome buyer skepticism about security, reliability, and data control — the same objections that cloud vendors still address today. Salesforce's public IPO in 2004, shortly after the dot-com bust, validated the model at scale. A SaaS company could grow to profitability and list publicly. The category was real.

Other SaaS companies followed in Salesforce's wake: NetSuite (ERP, 1998), WebEx (web conferencing, cloud pivot ~2001), and Concur (expense management, cloud pivot ~2000). By 2005, "Software as a Service" had entered common usage as the category name, and analysts had begun tracking it as a distinct market segment.

☁️ Cloud Infrastructure Enables SaaS at Scale (2006–2012)

The second major inflection in SaaS history was not a software product — it was Amazon Web Services, launched in 2006. AWS S3 (object storage) and EC2 (compute) for the first time made it possible for any developer to build and operate software infrastructure at data-center scale without owning a data center. The capital cost of starting a SaaS company dropped from millions of dollars to thousands.

This infrastructure shift enabled a wave of new SaaS companies that would have been economically impossible before it:

This era also saw the emergence of the freemium model as a mainstream SaaS go-to-market strategy. Dropbox's referral program — give 500MB of free storage for every friend you refer — demonstrated that SaaS products could grow virally without a sales team, in the same way that consumer products did.

💥 The SaaS Explosion (2012–2018)

By 2012, SaaS was no longer a challenger model — it was the default for new software businesses. The decade between 2012 and 2018 produced the highest concentration of landmark SaaS companies in history: Slack, Zoom, Stripe, Shopify, Intercom, Atlassian (IPO 2015), ServiceNow, and dozens of category leaders in every vertical.

Several structural trends drove this expansion:

• → Venture capital availability: SaaS metrics (ARR, NRR, CAC/LTV ratios) gave investors a consistent framework for evaluating growth-stage businesses. Capital flowed to SaaS at unprecedented rates from 2012 onward.
• → Product-led growth: Slack, Zoom, and Dropbox demonstrated that SaaS products could grow primarily through product virality — individual users adopting tools that then spread through organizations from the bottom up, without a traditional sales-led motion.
• → API economy: Stripe, Twilio, and SendGrid demonstrated that infrastructure components could be productized as APIs. This created an entirely new SaaS subcategory — developer-first products where the buyer was an engineer, not a business decision-maker.
• → Mobile ubiquity: Smartphones created new surface areas for SaaS products and new use cases (field service, mobile CRM, communication tools) that desktop software could not address.

🏭 Vertical SaaS and Platform SaaS (2018–2023)

The post-2018 era produced two distinct SaaS evolutions. The first was vertical SaaS: software built specifically for a single industry rather than a horizontal function. Companies like Toast (restaurants), Veeva Systems (pharma), Procore (construction), and Mindbody (fitness) proved that a product designed for a specific industry's workflows — even at the cost of a smaller total addressable market — could command higher prices, lower churn, and faster adoption than horizontal alternatives.

The second evolution was platform SaaS: products that stopped being pure software and became platforms on which other software is built. Shopify's app store, Salesforce's AppExchange, and Stripe's financial infrastructure layer transformed these companies from SaaS vendors into ecosystem operators. The platform model created defensibility that pure SaaS products lack — customers who have built on a platform's APIs and app ecosystem face switching costs that make churn nearly impossible.

This era also saw the emergence of compound SaaS: companies like Rippling and HubSpot that built multiple products on a shared data spine, enabling them to expand into adjacent markets without starting from scratch. The compound SaaS model is the most ambitious architecture in modern SaaS — it requires sustained investment to build but creates category-defining moats when successful.

🤖 The AI SaaS Era (2023–Present)

The commercial release of GPT-4 in March 2023 began the most significant structural shift in SaaS since the cloud infrastructure era of 2006. AI capabilities became accessible via API — any SaaS team could integrate LLM-powered features without building or training models. This created a new product category (AI SaaS) and disrupted existing categories simultaneously.

The AI era has three distinct dynamics playing out simultaneously:

• → AI-native products: New companies built entirely around AI capabilities, with no pre-AI version. These include coding assistants, AI writing tools, AI customer support agents, and AI-powered data analysis products. The fastest-growing SaaS products of 2023–2024 were predominantly AI-native.
• → AI augmentation of existing SaaS: Incumbent SaaS companies (Salesforce, HubSpot, Zendesk, Notion) adding AI features to existing products. This is the dominant volume of AI SaaS product releases — features added to established platforms rather than new companies.
• → AI infrastructure and tooling: A new layer of the SaaS stack — vector databases, LLM observability platforms, prompt management tools, model evaluation frameworks — that AI-native products require to operate reliably at scale.

Whether the AI era represents a fundamental restructuring of the SaaS industry — as cloud infrastructure did — or a technology cycle that will be absorbed into the existing SaaS stack is the open question of 2024 and 2025.

What to Do Next

The history of SaaS is the history of recurring transitions: from local to cloud, from enterprise to self-serve, from horizontal to vertical, from software to platform. Founders who understand the pattern recognize that AI is the current transition — and that the companies who will define the next era are building now, often in categories where the AI-native approach looks marginal today. The timesharing companies of the 1960s could not see that their model would become the default. The SaaS companies of the 2000s could not see that the category they were pioneering would become the default. Positioning for the next transition before it becomes obvious is how category-defining companies are built.