In early February close to a trillion dollars of market value dropped out of software stocks in about a week. A trader at Jefferies called it the ‘SaaSpocalypse’, which is a terrible word, and Forrester ran a headline saying SaaS as we know it is dead (the post under it was calmer). Eight months later the headlines have moved on, but I keep coming back to what that week said, because it wasn’t really about software.
I spent years inside big organisations, a big consultancy among them, where systems moved slowly and we called it governance. When the selloff hit I wasn’t shocked. It felt like the spreadsheet had finally caught up with something a lot of us already knew.
What happened
Investors started pricing in fewer seats at companies like Atlassian, which Atlassian denied. Workday, which had already cut 8.5% of its people in 2025, cut another 2% that same week. Retool found that 35% of the teams it surveyed had already replaced at least one SaaS tool with something they built themselves. Nobody’s rebuilding Salesforce over a weekend, though, and shipping a first version is maybe 2% of the work; the maintenance and the edge cases are the rest.
So SaaS is being squeezed, and the squeeze shows where the money went.
The pilots that go nowhere
MIT’s Project NANDA found in 2025 that 95% of enterprise AI pilots show no measurable effect on the bottom line. Meanwhile most people in those same companies use AI tools every day, on their own, because the tools help them.
Think about who sells those pilots. The big consultancies run the strategy decks, the transformation programmes and the change offices, and bill by the head for all of it. Their model needs the work to be big, slow and staffed, and AI is the first technology in a long time that makes good work small. A team of ten with good tools now does what a programme of a hundred used to promise.
I don’t think those firms can fix that from the inside (I’ve sat in the meetings where they try). Their whole price list is built on the premium for bigness, and that premium is what the market just marked down. I’d be fine if the big ones didn’t make it, and something smaller and more useful grew in the gaps.
Who gets to build
For decades the deal was that serious output needed serious infrastructure: teams, budgets, vendor stacks, a governance model and a consultancy to run it. That was true because the tools needed it, and the tools don’t need it anymore. Pieter Levels runs a portfolio of products on his own. Maor Shlomo built Base44 with a team of eight and sold it to Wix for $80 million six months in. 37signals has stayed under about 80 people for more than twenty years.
You don’t have to be an enterprise anymore to make something an enterprise would pay for. That moves the question from who has the budget to who’s willing to sit down and make the thing they need.
I’m one of those people now, with CDiT, a few clients and a lot of my own tools. Whether that’s a wave or a good year for a few of us, I don’t know yet. Time will tell though…
If this poked at something you're building: talk to me.
MAKE YOUR CASE.