The $285 billion wipeout was a verdict on pricing, not on any single product.
The stat that explains it is the price of a seat, roughly $20 to $150 per user per month in the categories AI agents now target.
Most software companies still charge by the seat, and almost none charge for outcomes yet.
Buyers have leverage right now. Use it before your vendors reprice.
Two days. $285 billion. Every analyst had a theory.
Most of those theories were about the product that set it off. They missed the point.
The real story is about a number so ordinary that nobody questioned it for twenty years. Here’s that number, and why it matters whether you buy software or sell it.
What actually happened?
In roughly 48 hours, the SaaS sector lost about $285 billion in market value, as investors feared AI agents were challenging the recurring per-user license fee. The trigger was a product release. Anthropic shipped 11 Claude Cowork plugins aimed at legal, sales, marketing, finance and HR workflows.
The damage was uneven and sharp. Thomson Reuters had its largest single-day fall on record, down 15.83%. LegalZoom dropped 19.68%. The iShares software ETF had its worst two-day decline since 2008. A Jefferies trader coined the name “SaaSpocalypse.”
It didn’t stop there. Losses reportedly grew to about $1 trillion within a week. One tally put the total at $2 trillion by April.
A caution: sources disagree on the exact start date, anywhere from late January to early February. Treat the dollar figures as market estimates, not audited numbers.
What is the one stat that explains it?
The price of a seat.
The workflows shown in the Cowork demo, including legal review, financial analysis, support triage and project management, sit in categories where vendors charge roughly $20 to $150 per seat per month.
That’s it. That’s the whole story.
For two decades, SaaS ran on a simple loop. More employees means more seats. More seats means more revenue. Investors paid premium multiples for how predictable that loop was.
An agent doesn’t need a seat.
Why does the seat break?
Picture a restaurant that charges per chair instead of per meal. It’s a great business while every diner needs a chair. More guests, more chairs, more revenue.
Then a robot starts delivering meals straight to people’s homes. Nobody needs a chair. The food is just as good. The business model quietly stopped working.
That’s the seat problem. If one agent can do the work of several employees, companies need far fewer seats, and the expansion revenue that headcount growth drove for decades comes under pressure.
Customers feel it before vendors do. ProductLed’s Wes Bush says founders keep telling him the same thing: customers are canceling, not for a competitor, but for ChatGPT or Claude. The product didn’t get worse. The free alternative got better.
The market’s math was already shifting. Several SaaS companies had reported slowing growth in Q4 2025, and the median SaaS revenue multiple reportedly fell from about 7x to below 5x within twelve months. The Cowork launch gave nervous investors a story to attach to numbers they already disliked.
Is the panic justified?
Partly.
SaaStr’s Jason Lemkin argues the crash isn’t really AI killing SaaS. Markets overreact. A two-day panic is not a business forecast.
And incumbents are moving. Salesforce’s Agentforce ARR reportedly grew 240% year on year to $1.5 billion, while forward earnings multiples for enterprise software compressed from 39x to 21x.
Here’s the part most people miss. The pricing shift has barely started. SBI found that about 80% of software companies still use some form of seat-based pricing, and only 0.6% use outcome-based pricing. The first real example is small but telling: Intercom charges $0.99 per resolution for its AI support agent, instead of only charging for seats.
The longer-term risk is real too. Gartner has estimated that $234 billion in enterprise SaaS spending is exposed to “agentic arbitrage” by 2030. That’s a forecast, and forecasts move. But analysts are taking the seat seriously.
So the market repriced in two days what vendors haven’t repriced in twenty years. That gap is where both the risk and the opportunity live.
What should CXOs do this quarter?
1. Find your seat exposure. List your top ten software contracts by spend. Mark every one priced per user. Those are your most negotiable, and your most exposed.
2. Read the agent clause. Many licenses were written before agents existed. Check how yours treats automated or non-human usage, and whether you’re being charged for it.
3. Ask for outcome-based pricing before they offer it. Vendors under pressure will get creative. Get in early and ask what a per-result price would look like.
4. If you sell software, find your new unit. If your revenue counts users, your customers and your investors will soon ask what you count instead. Have an answer ready: usage, workflows completed, or results.
The takeaway
The $285 billion is the headline. The seat is the lesson.
Markets didn’t decide that software has no future. They decided that charging by the person may not survive when the “person” doing the work is increasingly an agent.
The companies that come through this will stop selling access and start selling results.
Everyone else will spend the next few years explaining what a seat is worth.
