Vendor Landscape · 1975–2026
Enterprise software has moved through four waves since 1975, each roughly twelve to fifteen years apart, and in every one the incumbents mistook the weight of their installed base for a moat. It wasn't. Three signals recur across all four transitions: the incumbent's product stops moving, the challenger wins net-new projects, and the economics invert. The current wave, the one replacing SaaS with AI-native software, is resolving faster than any before it. The three AI-native labs at the front are already worth more than the twelve best-known SaaS pure-plays built over two decades.
Enterprise software history looks, at first glance, like a record of continuous growth. More vendors, more categories, more spend with each cycle. The pattern underneath is something different. It is succession: one cohort of software being displaced by the next in a transition that is usually invisible to the buyer until it has already finished. Some of the vendors in the record below you still pay out of habit. Some you stopped paying years ago. Some were absorbed into the holding pages of whoever acquired their customer list.
We reconstructed that replacement history from 114,000+ deployment case studies spanning five decades, and the record resolves into four waves, each roughly twelve to fifteen years apart. The explorer below is the primary evidence. Scrub through any year between 1974 and 2026 to see which vendors entered the market in each era, and how the cohort composition shifted as each wave progressed.
Each wave is twelve to fifteen years apart, and the pattern is reliable. The incumbent's product stops moving, the challenger wins net-new projects, and the economics invert. Three of these have already finished. The fourth is happening now, and it is the wave that brought you to this page.
The first wave was driven by a collapse in the price of the underlying hardware: minicomputers, and then commodity PCs, fractured the vertical grip IBM and the minicomputer cohort had held over enterprise computing, and the companies that built for the new substrate inherited the category. What makes it instructive is how quiet it looked from the incumbent's side. DEC, Wang, and Data General were still posting revenue growth even as their customers' procurement plans had already shifted decisively toward open systems, and by the time the financials caught up, the category had already been reassigned.
Marc Benioff founded Salesforce in March 1999 from a rented apartment near Telegraph Hill in San Francisco, having spent the previous thirteen years inside Oracle. The pitch to his first employees was that the next generation of enterprise software would be delivered through a browser, billed monthly, and would never require a CD to be shipped or installed. By 2010, Siebel had been absorbed into Oracle and was running in maintenance mode, and Salesforce was on a run rate of well over a billion dollars in subscription revenue, writing the majority of net-new CRM contracts in the Fortune 500.
The second wave is the one most people currently working in enterprise software lived through as adults, and the mechanics recur in every later wave: license-plus-maintenance pricing could not defend against subscription economics once CFOs started comparing total cost over a five-year horizon, and an eighteen-month release cycle could not defend against a competitor shipping every quarter. Once CIOs noticed the challenger's product was improving faster, the renewal decision had already been quietly made for them.
The third wave was subtler, because it unfolded inside the customer's existing contracts rather than alongside them. Workday did not kill Oracle HRMS so much as siphon every net-new workload that would otherwise have expanded it, and Snowflake did not kill Teradata outright so much as make Teradata's next renewal negotiable in a way it had never been before. In each case the outcome was the same: the incumbent became an account to harvest rather than a platform to build on, and by the time finance noticed that the line had tilted, the floor had already moved under it.
The fourth wave follows the same shape as every prior transition: a new substrate, new economics, a new cohort assembling around the categories most exposed, and an incumbent cohort once again mistaking the weight of its installed base for a durable moat. We track 17,000+ AI tools and the adoption footprint of each one across the Fortune 1000, and the category-level displacement curves are resolving on a timeline comparable to the SaaS transition of the mid-2000s: unmistakably past the point where it could be written off as a passing cycle.
The signals recur in the same order across all four waves, and they lead the incumbent cohort's revenue roll-off by roughly two to three years. Operators who read them correctly reallocated spend quietly, ahead of their competitors. Operators who ignored them paid a rebuild tax a half-decade later, after the replacement had already become the default answer in the category and the old line item was being cut from the next budget.
Five decades of enterprise software bookings and replacement outcomes, reconstructed from SEC filings going back to 1975, archived procurement disclosures, deployment histories, the real-time research of independent analyst firms, and direct interviews with operators who ran the stack through each transition. Working backward from vendor-produced case studies or analyst coverage underwritten by the companies being rated systematically overstates incumbent durability and understates the speed of the handoff, which is why we did not. Airframe takes no vendor money, publishes no sponsored research, and no company in the 17,000 we track has any influence over how it appears in the record.
The first read on this transition, two years in, was that the incumbents would win because they understood AI was coming and they held the data to train on. That read has not aged well. The three AI-native labs at the front of the current cohort are OpenAI at $852 billion, Anthropic at $800 billion, and xAI at $250 billion, for a combined private-market valuation of roughly $1.9 trillion. For comparison, the twelve best-known SaaS pure-plays founded between 1999 and 2009 are collectively worth about $844 billion, built over more than twenty years; that cohort includes Salesforce, Workday, ServiceNow, Shopify, Atlassian, HubSpot, Okta, Twilio, Veeva, Dropbox, Box, and Zendesk.
The twelve best-known SaaS pure-plays produced $844 billion of value over twenty years. Three AI labs, in under a decade, have produced $1.9 trillion. The market has already voted on which transition it thinks is larger. Their combined revenue is not there yet. The growth rate suggests it will be.
The valuation gap is not the only place the cohort is pricing itself in. Through the first quarter of 2026, the largest of the AI labs have moved into direct relationships with the largest software-focused private-equity firms, structured as ventures rather than as customer contracts. OpenAI is in advanced talks with TPG, Bain Capital, Advent, and Brookfield on a roughly ten-billion-dollar enterprise AI venture. Anthropic is in parallel talks with Blackstone, Hellman & Friedman, Permira, and General Atlantic on a roughly one-billion-dollar consulting venture, with its Accenture partnership already operating and roughly thirty thousand consultants trained against its model. The pattern is structural, not transactional. The cohort that is winning the valuation comparison above is, in the same window, pricing itself into the operating-partner relationships that decide where the next decade of software budget gets routed. Diligence on a portfolio that includes one of those labs cannot credibly be routed through a venture co-owned by it, and that constraint is going to shape the next two years of the buying conversation more than any single product release.
The shape of the arc matters more than the absolute numbers. The signal-01 and signal-02 patterns are resolving faster in this cycle than they did in any of the previous three, and the incumbents' installed base, the thing that has historically bought them a decade of graceful decline, is compressing on the same curve. Part 3 · The Trapped Trillions → is about the capital stuck behind that compression. Part 1 · The Register → is the full record of everyone who built the waves before this one.
Of the 1,425 vendors in the record above, 263 entered the market since 2019, roughly one in every six, competing directly for budget lines enterprise incumbents have held for a decade or more. The signals described here are not historical; they are active right now, across the software, infrastructure, and developer tooling layers where most enterprise stacks are concentrated. The question reads two ways depending on where you sit. Inside one company, it is which of your next renewals is the one that sets the direction for the five years after, and which of your vendors is sitting on a curve that has already turned. Above a portfolio, it is which assets in your book are compounding through this transition, which are repairable, and which were underwritten against assumptions that are no longer holding. Both questions resolve against the same record: a structured account of what comparable companies are actually deploying, what is working, and what is being quietly walked away from. Airframe is where that record lives.