Every era of enterprise software has layered on top of the last, and in every era the vendor still got replaced. The ceiling on what software can deliver per person just moved from roughly $4.7M to roughly $22.9M, set by a cohort of 263 AI-natives founded since 2019. The record below is what the last fifty-seven years built. The next three years will decide which of these 1,425 vendors are current line items on a budget, and which are previous ones.
Enterprise software produced $25.5 trillion of value across 1,425 vendors over fifty-seven years, and the record that history left behind is the one you are now buying from. The productivity ceiling of the cohort writing the next decade of checks is $22.9M of value per employee, roughly five times the cloud-native generation it is replacing. The names below are every vendor the market has priced to that logic. The next three years will decide which of them are current line items on your budget, and which are previous ones.
Enterprise software has moved through twelve phases since IBM unbundled in 1969: mainframe, minicomputer, personal computer, client-server, web, open source, SaaS, cloud, mobile, API, developer tools, and now AI. The instinct is to read these as a sequence of replacements. The more honest read is that the substrate has layered. Mainframe software still runs most of the world's banks. The COBOL code that clears your paycheck was likely written before the engineer running it was born. What gets layered is the substrate. What gets replaced, reliably, is the vendor on the line item, and that is the record below.
The shape of the curve, not any single name. The productivity ceiling moved more in the seven years since 2019 than in the forty-three before it. The 263 AI-native companies in that cohort have set a value-per-employee bar roughly five times higher than the cloud-native generation they are repricing on Monday-morning earnings calls. The three frontier labs at the top of the cohort are now jointly worth more than the twelve SaaS pure-plays everyone learned to value across two decades, on about one-tenth the headcount. Whichever vendor a board is reviewing this week is being held against that math whether the deck says so or not.
Every enterprise-software vendor founded by the selected year, valued above $500M today, and not yet acquired by that year. Drag the timeline to watch the record fill in across fifty-seven years. Search for a name, a domain, or an industry. Click any row for the full entry.
| Company | Value | Employees | Value/Emp | Department | Sub-industry |
|---|
The metric that travels cleanly across eras is market value divided by full-time headcount. It captures what each generation of software was actually able to deliver per unit of human labor, and the curve it traces is steep. The AI-era gap is larger than the gap between any two prior generations combined.
The productivity ceiling matters because every cohort below it is now being repriced against it. Block's February 2026 announcement is the waypoint. A workforce reduction from roughly 10,000 to just under 6,000, paired with a raised profit outlook and a 25% single-day gain, moved the company's per-employee market cap from roughly $3.4M to about $6.4M. One SaaS-era company pricing itself against the AI cohort underneath. Anthropic, at $800 billion of enterprise value across 2,500 employees, is already worth nearly four times IBM at about 1% of the headcount. That is the shape of the leading companies a CFO is about to see in a software review.
Every vendor on the register above is either a line item on a budget today or the residue of one. The next three years will decide most of the contested cases.
If your company is on the register, or if someone you worked with is, send them the entry. Most of the value on the register above is owned by people who underwrote it against assumptions the agent era is now repricing in real time.
Airframe is the system of record for enterprise AI.