The "SaaSpocalypse": what the market got right and what it overreacted to

The "SaaSpocalypse" is more nuanced than the headlines suggest. As someone who sees real value in AI tooling, and also spent 20 years building SaaS applications, I think some context is useful.

Anthropic launched plugins for Claude Cowork and the market wiped $285 billion off SaaS stocks. ServiceNow, Salesforce, HubSpot, Atlassian, Asana, Adobe: all down double digits. The narrative was that "AI agents replace software, sell everything".

The nuance is that there are two distinct categories of software value.

The first is single-user workflow tools: one person reviewing a contract, analysing a dataset, drafting a document. Cowork is genuinely capable here. If your product's core value is a UI on a task one person does at a desk, that value is being compressed. And if you're a customer paying for multiple single-function SaaS tools, you should be evaluating whether one Claude Max subscription at £80-160 per month now replaces several of them. For some combinations, the maths already works.

The second category is multi-user coordination systems, where teams work across shared state such as project management, CRM, sales pipelines and more. Architecting this requires concurrent database access with transactional integrity. Role-based permissions, audit trails, integration APIs. This is what enterprise SaaS actually is, and is NOT what Cowork is (today).

Cowork can credibly replace the first category but not the second, because it lacks the entire application stack that makes multi-user software work.

Maybe that is the next step - Cowork becomes a SaaS platform?

But even if Anthropic added all those layers, they wouldn't be disrupting SaaS, they would be building SaaS. And at £80-160 per user for what is still fundamentally a single-user agent, that's not cheaper than £20-50 per seat for tools that already handle all of this plus years of embedded organisational configuration.

Back to the SaaSpocalypse: the market repriced two different things at once and it's worth separating them.

• Single-function workflow tools face a genuine structural threat, and their customers should be reviewing value for money.
• Enterprise coordination platforms face budget pressure from AI spend reallocation, which is real but very different from obsolescence.

What we saw last week was an overreaction layered on top of a legitimate correction, tipping investors who were already nervous about SaaS growth into full capitulation.

That said, Cowork is genuinely exciting: the quality of Opus 4.6 is significantly better than 4.5, and combining that with specialist tools is a real step towards unlocking the ROI that AI has been promising but not quite delivering.