Software has moved through eight technological waves over 75 years against a $650B-a-year tool-buyer wallet. The ninth, AI, is being priced against a $10 trillion services-and-labor wallet, an order of magnitude larger.
Software has moved through eight phases of disruptive technological change. Mainframe (1950s), Minicomputer (1970s), Desktop (1980s), Client-server (1990s), Internet/web (1995), Cloud (2005), SaaS (2005), and Mobile apps (2010). We are now entering the ninth era — AI — and it is being built against a market the prior eight did not have access to.
The first two pieces of this series, the productivity ceiling moving from $4.6M to $17.8M per employee, and the four-column reorganization of the existing $26T, are stories about the same fixed-size industry being reorganized. Eight waves of disruptive change, 2,406 enterprise software companies, $26 trillion of accumulated value, 75 years of the Register. All of that built against one denominator: the global enterprise software budget line.
The ninth wave is not building against that line. It is building against the one underneath it — the services and labor line, the operations line, every place a human used to sit and a workflow used to be billed by the hour. That denominator is, depending on which services count, an order of magnitude larger than the one the previous eight waves shared. The implication is that the next decade of the Register will be larger than the first 75 years.
That framing is wrong. The industry is not the same size.
Software, historically, has been a roughly $650 billion-a-year revenue category. Add IT services on top — implementation, integration, custom development, managed services — and you reach somewhere between $1.5 trillion and $2 trillion of annual revenue depending on how you count. That has been the addressable market every enterprise software vendor has been building against for fifty years. $26 trillion of accumulated enterprise value, sitting on top of about $2 trillion a year of revenue, working out to a roughly 13× multiple of revenue at the aggregate level. That is the math the SaaS-era venture industry, the cloud-era public market, and the entire PE software book underwrote against.
What changes in the AI era is not the multiple. It is the denominator.
The first generation of AI-native enterprise software is not selling code; it is selling completed work. Claude Code does not assist a developer; it ships pull requests. Glean does not surface the search result; it produces the briefing. Clay does not enrich the lead list; it runs the outbound. Harvey does not retrieve the precedent; it drafts the brief. The customer is no longer buying a tool the employee uses to do the job. The customer is buying the job done. And the budget line that funds it is no longer the IT software line. It is the services line, the labor line, the operations line — every line on the income statement where a human used to sit and a workflow used to be billed by the hour.
Sequoia Capital's Konstantine Buhler frames the AI transformation as a $10 trillion revolution. We think that's the right anchor. The global services-and-labor wallet AI is now competing for is at least $10 trillion a year, with a wider speculative band reaching toward $50 trillion depending on how much of the labor budget the cohort eventually subsumes.
The verticals visible today, where the market sizing is tight and the AI substitute is already in market, sum to roughly $5 trillion on their own. The named breakdown is below. The $10 trillion Sequoia anchor sits above the visible-today floor and below the speculative ceiling, in the band of services and labor the cohort is actively expanding into.
If the AI cohort captures even a single-digit percentage of the $10 trillion, it produces an industry whose terminal size is an order of magnitude larger than the enterprise software industry that preceded it.
Customer support is the cleanest single illustration. The unit economics of a 500-agent contact center, walked through in Piece 1, show a $2M seat-priced SaaS account becoming a $5–15M work-priced AI account against a $25–50M labor line. Same customer. Three to seven times the software revenue, against a labor budget an order of magnitude larger. That mechanic repeats across every category named above.
Eight waves of enterprise software built $26 trillion of value across 75 years against a single denominator. The ninth wave is being built against a different one. Sequoia anchors that denominator at $10 trillion a year, the AI revolution as a services-and-labor opportunity an order of magnitude larger than the prior software wallet. Today's tight verticals visible in the data sum to roughly $5 trillion on their own. The wider speculative band reaches $50 trillion. The cohort being founded right now is competing across that range.
The next decade of the Register will be larger than the first 75 years. If you are an investor, owner, or operator of a software company, the question worth asking this quarter is not which side of the AI transition your business is on. It is which side of the AI denominator it is on. A company priced against seats is competing against a productivity ceiling that has moved out from under it. A company priced against work is competing for budget lines that did not exist as software line items a decade ago. The Register is the structured place where that story gets written.
— Paul