In August 2011, venture capitalist Marc Andreessen wrote an essay in the Wall Street Journal that became one of the most quoted pieces in the history of Silicon Valley. The title was simple: Why Software Is Eating the World. His argument was that software companies were going to take over enormous chunks of the global economy, and that every industry had better wake up or get eaten alive.
He was right. Amazon ate retail. Netflix ate video rental. Spotify ate the music store. Airbnb ate hotels. Uber ate the taxi industry. Software didn't just show up to the table, it ate everybody else's food, ordered dessert, and asked for the check.
For the next decade, the smartest money in the world bet heavily on software. Venture capital poured in. Private equity firms built entire strategies around it. Valuations went through the roof. A company that could show recurring subscription revenue, a growing customer base, and some semblance of a moat could raise money on a napkin and go public at 20 times revenue without breaking a sweat.
That world is over.
So What Happened?
Fifteen years after Andreessen's prophecy, something started eating software itself. And that something, of course, was artificial intelligence.
The term people started throwing around was the SaaSpocalypse. Between January 15 and February 14 of this year, approximately $2 TRILLION in market cap evaporated from the software sector. This wasn't a normal tech correction. It wasn't rising interest rates or a macro scare or a bad earnings season. It was a structural shift in how businesses think about software. And it happened fast.
Here's the kicker…
The SaaS business model, which is the subscription software model that every company from Salesforce to your project management app to your HR platform runs on, was built on one beautiful, simple equation: more employees at a company equals more software licenses equals more revenue. You hire 50 new people, you need 50 more seats. The software company prints money. Beautiful business, right?
Now what happens when AI makes one person as productive as five? The company doesn't need five seats anymore. Revenue per customer drops. Growth stalls. Valuations compress.
I'm gonna say that again so it hits right. When AI makes one person as productive as five, companies stop buying five seats. They buy one. That's not good for the software companies that built their entire business around selling five.
In February 2026, approximately $285 billion in market value vanished from software stocks IN A SINGLE TRADING SESSION. ServiceNow dropped 7%. Salesforce fell 7%. Intuit plummeted 11%. Thomson Reuters collapsed nearly 16%. The catalyst wasn't a recession or a regulatory crackdown. It was AI agents doing the work that human employees used to need software subscriptions to do.
Companies are moving from "one tool per task" to "one agent per outcome." Instead of buying separate subscriptions for project management, CRM, email marketing, customer support, and analytics, businesses are building AI agents that handle entire workflows on their own. And when you don't need all those separate tools? You cancel all those separate subscriptions.
That's the game right now.
The Numbers Don't Lie
Let me give y'all the valuation data because this is where it gets REAL.
In Q3 2021, private SaaS revenue multiples hit a staggering 41.48x. That means investors were paying $41 for every $1 of annual revenue a software company generated. That's like paying $41 for a dollar bill because the dollar bill had a good year.
At the end of 2022, on the eve of ChatGPT's launch, software companies traded at an average EBITDA multiple of 30. By the end of 2025 that fell to 22. Today the median multiple on forward profitability is only 16. Revenue multiples dropped as well, from 10 to 12 times expected revenue in 2022, down to about 4 times under current valuations.
To put that in plain English: the premium investors used to pay for software companies because of their predictable recurring revenue and high growth rates? GONE. The market is essentially saying that a lot of these companies don't have a moat anymore, because AI is climbing over the wall, knocking it down, and setting up shop on the other side.
The SaaS index, which represents companies that sell subscription-based software to enterprises, fell 6.5% in 2025, compared with a 17.6% RISE in the S&P 500. So while the rest of the market was popping bottles, software was getting its face beat in.
And 2026? Even worse.
OpenAI and Anthropic Are Coming For Everything.
Now I need to talk about something that a lot of people haven't connected yet, which is the fact that the companies building the AI that's disrupting software? They're also building the software directly. They're not just selling shovels to the gold miners. They're out here mining too.
Let's start with OpenAI.
In January 2025, OpenAI launched something called Operator, an AI agent that literally browses the web for you, clicks things, fills out forms, and completes tasks from start to finish. On its own. No human needed. And by July 2025 they folded it directly into ChatGPT as Agent Mode, so every ChatGPT user now has access to an AI that can operate software on their behalf.
Think about what that means for a second. Every company that built a beautiful SaaS product with a gorgeous dashboard and a smooth user interface? Congratulations. Your users might not need to open your product anymore. They'll just tell ChatGPT what they want done, and ChatGPT will go do it inside your product without the user ever touching it. You're not even getting the engagement credit. Your product is becoming the back end that an AI agent operates invisibly.
But wait. It gets worse.
In June 2025, OpenAI revealed it was building collaborative document features that would put it in DIRECT competition with Google Docs and Microsoft Word. Not "inspired by." DIRECT competition. And ChatGPT already has a deep research feature that goes out, reads dozens of sources, synthesizes everything, and gives you a full research report. That's a product that a whole category of research software companies built their entire businesses around.
900 million people use ChatGPT every week. 193 million use it every single day.
Now let's talk about Anthropic, the company that makes Claude.
On January 12, 2026, Anthropic launched Claude Cowork. This is Anthropic shipping a full enterprise productivity platform that connects to Slack, Google Drive, Gmail, Notion, Asana, Jira, Monday.com, ClickUp, HubSpot, Figma, Canva, Box, and 50 other tools, and then handles your workflows across ALL of them simultaneously.
The day Cowork launched, Monday.com's market cap dropped $300 million before the session closed. BEFORE THE SESSION CLOSED. That's how fast the market processed what had just happened.
And it didn't stop there. Anthropic shipped 21 specialized plugins covering productivity, marketing, sales, finance, legal, data analysis, HR, engineering, design, and operations. There's even a plugin for investment banking, one for equity research, and one for private equity. You type one command and Claude reviews your priorities, organizes your task list, and sets up your day. The people building this stuff are calling it "vibe working." The idea is the same as "vibe coding," which is where you just tell an AI what you want built and it builds it. Except now instead of code, you're describing the work outcome you want, and Claude just goes and does it.
Then in February 2026, Anthropic dropped Claude Code Security, which scans entire codebases for security vulnerabilities and suggests patches. The day that dropped, CrowdStrike fell 7.2%, Zscaler fell 7.1%, and Palo Alto Networks dropped 2.6%. Cybersecurity companies. Getting destroyed by an announcement from an AI company. And in March 2026, Anthropic added a feature where you can send a prompt from your phone and Claude accesses programs on your actual computer, including browsers and spreadsheets, to complete the task. Your computer. Remotely. From your phone. Now Claude Mythos is out and I don't even know how crazy that model is going to be.
So let me paint the full picture for y'all. OpenAI is coming for your browser, your documents, your research tools, and your workflows. Anthropic is coming for your productivity software, your project management tools, your security software, your CRM, and your desktop. So basically, that's the unemployment line for like... everybody.
The software companies that built moats around specific workflows for the last 20 years are now watching two of the best-funded companies in the history of technology ship products that make those moats irrelevant. And they're doing it fast. Anthropic's core code for Cowork was AUTONOMOUSLY WRITTEN BY CLAUDE in just a week and a half. The AI is literally building the tools that are replacing the software that humans spent years building.
Enter Private Equity. And Then Exit Private Equity.
Here's where shit gets uglier than a Master P sneaker.
For the past decade, private equity firms that specialized in software were basically printing money. The playbook was simple: find a boring but stable software company, cut costs, raise prices on the captive customer base that can't easily switch, load it up with debt because interest rates are low, grow revenue, sell it to another buyer at a higher multiple in 3 to 5 years. Rinse and repeat.
Nobody ran this playbook harder than Thoma Bravo.
Thoma Bravo is a private equity giant managing about $180 billion in assets, and roughly 80% of their portfolio is enterprise software. They've made over 530 acquisitions in software and technology over the years. They are THE defining firm in software-focused private equity. Like if software private equity was the NBA, Thoma Bravo is the Golden State Warriors at their peak.
In October 2021, right at the peak of that 41x valuation mania, Thoma Bravo bought a customer experience software company called Medallia for $6.4 billion.
Today, that $6.4 billion is... gone.
Thoma Bravo is nearing an agreement to hand over Medallia to its lenders, wiping out $5.1 billion in equity for Thoma Bravo and its co-investors. Medallia has been crushed under $3 billion in debt owed to Blackstone, KKR, Apollo Global, and Antares Capital.
READ THAT AGAIN. They bought it for $6.4 billion. They're handing the keys to creditors. The equity, all $5.1 billion of it, is completely wiped out. That is one of the largest private equity write-offs you will ever see on a single software deal. That's not a loss. That's a catastrophe.
Now to be fair, Blackstone's credit head said Medallia's problems were due to execution issues, not purely AI. But here's the thing, you can't separate the two anymore. The entire software private equity landscape is feeling this. Medallia is just the loudest example today. It won't be the last.
In fact, back in February, Thoma Bravo and another big software PE firm (Vista Equity Partners) were out here rushing to reassure their fund investors that everything was fine, that their portfolios were healthy, that the AI selloff was an overreaction. When firms are rushing to reassure investors, that is not a good sign y'all. That's the finance equivalent of the "Everything Is Fine" meme.
Why This Matters to You and Me
Now look, I know what some of y'all are thinking. "Carl, I don't own Thoma Bravo. I don't care about some software company called Medallia. Why does this matter to me?"
I got you. Here's why.
Private equity firms that bought software companies at inflated prices loaded those companies with debt. That debt has to get paid back through the cash flows of the business. When those cash flows slow down because AI is eating into their revenue, they can't service the debt. And when you can't service the debt, you hand the keys to the creditors, which is exactly what just happened with Medallia. But this isn't just one company's problem.
Gartner, the most establishment research firm in enterprise technology, predicts that 35% of point-product SaaS tools will be replaced by AI agents by 2030.
The companies that survive this shift are going to be the ones that genuinely integrate AI into their products, not just slap a chatbot on top of what they already built and call it an AI feature. We're already seeing how the market is judging this. Public investors are now valuing software companies based on their AI integration, or their death risk from AI disruption, rather than just how big their market is. That's a totally different game than the one these companies thought they were playing.
The Irony Is Thick
Marc Andreessen wrote that software would eat the world. He was right. And now his own venture capital firm, Andreessen Horowitz, which has a core strategy built around enterprise software, SaaS, and cloud companies, is watching AI consume the very thing it spent 15 years betting on. But they're already invested in AI so... LOL.
The next wave is already here, and the scoreboard is updating in real time. Thoma Bravo losing $5.1 billion on Medallia is today's headline. It won't be the last one.
Software ate the world. Now AI is eating software. The only question left is what's next on the menu.
And whoever figures that out first? They eat.
This is not legal advice, financial advice, or attorney advertising. This newsletter is for educational and informational purposes only. For actual guidance on investments or financial decisions, please consult a licensed financial advisor. Don't be out here making moves based on a newsletter. You're smarter than that.
Don't be stingy with the 🏀. Pass it to a friend.
See y'all next week.
CJB