The value of software companies is collapsing, and investors still don't think we've hit the bottom. Shares that were hot a year ago have sunk as much as 80% in 12 months - but why?
You might have read about the "SaaSpocalypse" recently and wondered what all the fuss was about.
The SaaSpocalypse refers to a recent massive devaluation of public software companies, where household names like Adobe, HubSpot, and Salesforce are among a large group of companies that have lost over 50% of their value in the last 12 months.
What has caused this massive value reduction?
The short answer is AI, but to nod your head and say 'ok' and move on without understanding it properly is a bad idea, because this is going to reshape the entire industry a lot more before halftime.
There are 4 very important things happening under the hood, which everyone needs to understand.
1. The cost of coding is decreasing (Coding Copilots) 2. The bar for creating code has dropped (Vibe Coding) 3. The users of software tomorrow won't be human (Agentic Revolution) 4. Hollowing out of value (Bring your own brain)
The Decreasing Cost of Code Creation
AI coding tools like Claude, Cursor and a multitude of others can now write code better, and faster than most humans - and it's only going to get better and faster.
Software historically commanded a high value because it took large groups of highly skilled humans months or years to create a piece of software - the bar was high.
10 units of work = 1 unit of software.
AI has turned this upside down, now…
1 unit of work = 10 units of software.
This has an immediate effect on value, because the age-old driver to buy software (a lack of time and resources) doesn't hold in the way that it did. It immediately makes everything you built pre-AI easier to copy and replicate, and therefore depresses its value.
But there is a silver lining.
Although what you built yesterday doesn't have the same moat it once did, you can still achieve more with a team with specialist knowledge than an individual without. The scale of companies is still an advantage, if you can marshal those resources and get rubber on road.
The Bar For Creating Code Has Dropped
Cheap code is one thing. The next bit is bigger.
Tools like Lovable, Cursor, Base44, and Replit now let people who have never written a line of code in their lives build production software. Not toys. Real systems with databases, authentication, payments, the lot.
The pool of people who can build software went from "a few million developers" to "anyone with an internet connection and a problem to solve."
Yesterday: 1 product = 12 months × 10 engineers. Today: 1 product = 1 weekend × 1 person with AI.
Wasn't every SaaS company on earth built on the quiet assumption that their customers couldn't build the software themselves—because building was hard, buying was easy, and so they bought?
That assumption just died.
Why pay $50k a year for a niche workflow tool when an internal hire can spin up a better-fitting version with Lovable in three weeks?
Some buyers still need scale, deep integrations, and battle-tested platforms. Those customers aren't going anywhere. But the long tail of 'we solve this little problem' horizontal SaaS just got cut off at the knees.
The silver lining: knowing what to build is now far more valuable than knowing how to build. Domain experts just got promoted.
The Users Of Software Tomorrow Won't Be Human
This is what really broke the price, and it's also the area where there is the most ignorance in the general public, and it requires 1 MASSIVE change in how you think about AI.
Phase 1 was about Chat > Answer. Phase 2 will be about Objective > Completed - AI will be DOING work via what are called Agents.
For 20 years, SaaS companies priced per human. 100 employees, 100 logins, 100 monthly subscriptions. The more your customer grew, the more they paid. Predictable. Compounding. Beautiful.
Then AI agents arrived.
An AI agent doesn't log in. It doesn't take lunch breaks. One agent can do the work of ten human users — and it never asks for a pay rise.
Yesterday: more humans = more revenue. Today: fewer humans, same output, less revenue.
The signals are already in the data. Atlassian — the company behind Jira — reported its first-ever decline in enterprise seat counts in Q1 2026. And you only need to read the news on literally ANY day right now to see a similar story.
If the unit you charge for is the human, and the human is being deleted, your revenue model is being deleted with them.
Hollowing Out Of Value (Bring Your Own Brain)
If the agents are the ones bringing the intelligence, then the software underneath them risks being reduced to a system of record.
That's not particularly valuable.
Everything clever a piece of software used to do — the workflow logic, the recommendations, the analysis, the routing — gets outsourced to a third-party agent that the customer brought along. The vendor no longer owns the brain. They are just the filing cabinet.
Remember what happened with Car Sat-Navs?!
A decade ago, BMW and Audi were charging around £2,000 for a built-in navigation system. Today, the best sat nav on the planet is sitting in your pocket — Google Maps, Apple Maps — always up to date, traffic-aware, and free.
The car manufacturers didn't win that fight, they surrendered. They now ship a screen, plug in CarPlay or Android Auto, and let your phone do the work. The price they can charge for the nav system has collapsed by around 90%.
That's the trajectory for any software provider whose intelligence gets pushed up into the agent layer. You become the screen. The agent is the brain. And screens don't command premium pricing.
If you want to maintain that premium pricing you'll need to bake logic and intelligence into your API layer, not just data access.
So What??
Cheaper code. Lower build threshold. Pricing model collapse. Intelligence hollowed out.
Each one alone would be a problem. Together, they explain why investors took a hammer to public software valuations, and why investors still aren't sure where the bottom is (it's still going down).
Right now all software businesses are getting lumped together, but it's about how each vendor responds to each one of these 4 forces that will decide whether they end up being worth dollars, or cents...
Make no mistake, AI will continue to re-write the rules of where value sits, and you need to have a plan.