Software is Eating Labor
Alex Rampell of a16z explains how software is evolving from digitizing records to performing labor, shifting the industry's focus from the $300 billion SaaS market to the $13 trillion labor market. This transition, accelerated by AI, is forcing a change in business models from seat-based pricing to outcome-based pricing, creating new opportunities and expanding the total addressable market.
The central thesis is that software is transitioning from augmenting human labor to replacing it, setting its sights on the massive labor market. While the worldwide SaaS market is approximately $300 billion, the U.S. labor market alone stands at $13 trillion. This shift represents the next significant evolution for the software industry, where capital is transformed into software that directly performs jobs.
The “Filing Cabinet” Era of Software
Historically, the software industry’s growth, accumulating a market cap of around $2.2 trillion, has been built on a simple premise: turning physical filing cabinets into digital databases. This process, while valuable, didn’t fundamentally change the nature of the work. A human who once read a paper file now reads a digital record on a screen.
This pattern is visible across numerous industries:
- Airlines: American Airlines and IBM created Saber to replace massive filing cabinets of flight records with a mainframe system accessed by travel agents.
- Sales: The paper leads from “Glengary, Glenn Ross” were digitized by companies like Act!, Siebel, and ultimately, Salesforce, but a salesperson still managed the records.
- Manufacturing & Inventory: Companies like SAP and JD Edwards digitized inventory management, which was previously done on paper.
- Healthcare: Epic Systems, founded in 1979, became a giant by digitizing the vast paper-based health records in hospitals and doctors’ offices.
- HR & Payroll: ADP and Workday replaced physical time cards and attendance slips with digital systems, but the core process of a human reviewing that data remained.
In every case, the software served as a more efficient system of record, but the human operator was still essential to the process.
The Business Model Shift: From Seats to Outcomes
This historical context is crucial because the traditional SaaS business model is built on it. The “tall, grande, venty” model of pricing—selling seats on a per-month basis—is predicated on the idea of software as a tool for a human worker.
This model is now facing an existential threat and a massive opportunity. Consider a customer support center with 1,000 agents. The company might spend $75 million annually on labor but only $1.4 million on Zendesk software seats.
If AI can answer all the questions, how many seats do you need? Zero. … On the other hand, … maybe Zendesk could charge $5 million a year. “Hey, don’t spend $75 million a year on support anymore. Spend $5 million a year on support. Just pay it all to us.”
This illustrates the precipice on which companies like Zendesk stand: their revenue could either collapse to zero or multiply by shifting to an outcome-based pricing model. They are no longer selling a tool; they are selling the completed job.
AI: From System of Record to System of Action
AI enables software to move from being a passive filing cabinet to an active operator on that filing cabinet. Instead of just storing information, the software can now perform the tasks associated with it.
- Accounting: QuickBooks can move from being a record of accounts receivable to actively calling clients to collect overdue payments.
- Healthcare: An EHR system can deploy an AI nurse to call a post-surgery patient, ask about their pain level, and advise them on next steps.
- HR: Workday can go from storing resumes to actively calling previous employers to conduct reference checks.
- Sales: Salesforce can evolve from a CRM to an automated system that actively prospects and qualifies customers.
This is happening now. A real-world example is a company that finds a job listing for a receptionist on Craigslist that has been open for six months. The software company “applies for the job,” offering to perform 8 of the 10 listed responsibilities for less than half the cost of the human salary. This approach massively expands the market, as the software is selling into the multi-trillion-dollar labor budget, not the comparatively small software budget.
Why AI is Better for Certain Jobs
It’s not just about cost. AI excels in roles where humans are not ideal:
- Intermittent Demand: AI can scale instantly to handle unpredictable surges, like customer support during a Black Friday sale or airline re-bookings during a storm, without the need to hire and fire temporary staff.
- Demoralizing Jobs: Tasks like debt collection involve frequent rejection and hostility. An AI doesn’t get tired or demoralized by negative interactions.
- Regulatory Certainty: AI can be programmed to adhere strictly to legal scripts (e.g., in financial services), eliminating the risk of human error that could lead to regulatory fines.
- Languages: An AI agent can communicate in dozens of languages instantly, providing service to non-English speaking customers that would be impossible to staff with humans.
This technological shift also makes previously unviable business models feasible. A niche marketplace like an “Airbnb for bicycles” might have failed due to high customer acquisition and support costs. Now, with AI sales reps and AI support centers, the core economics can work, expanding the market for entirely new categories of businesses. The opportunity is global, aiming to capture a piece of the worldwide labor market and make the current software market look small in comparison.