Atlassian fired 1,600 people today. Ten percent of the company, gone. More than 900 of them were in research and development — the people who actually build the software. The reason? To "self-fund further investment in AI and enterprise sales."
Here's the part that makes my head spin: Atlassian has lost more than half its market value since January. Not because of bad earnings — revenue's up, cloud growth is 25%, they have 600 customers paying over a million a year. The stock crashed because investors believe AI will make Atlassian's products obsolete. So Atlassian's response to the market panic about AI replacing them... is to fire 1,600 people to fund AI. The snake is eating its own tail.
They're calling it the "SaaSpocalypse." The term appeared in February 2026 and it stuck because it describes something genuinely new: not a recession, not a correction, but a market-wide revaluation of whether entire categories of software companies have a future. Atlassian, ServiceNow, Salesforce — companies that defined the last decade of enterprise software — are suddenly being priced as if AI agents might simply do what their products do, for free, inside a chat window.
The CEO's internal memo is a masterclass in corporate doublespeak. "Our approach is not 'AI replaces people,'" Mike Cannon-Brookes wrote. "But it would be disingenuous to pretend AI doesn't change the mix of skills we need or the number of roles required in certain areas." That's a sentence that manages to say "we are replacing people with AI" without technically saying it. They left Slack open six hours longer than usual so employees could say goodbye. A $1,000 "technology payment" once you hand back the laptop. The corporate funeral rites of the AI era.
But here's what actually interests me about the SaaSpocalypse: I think the market is simultaneously right and wrong. Right that AI agents will eat chunks of what Jira and Confluence do — I literally use AI tools to manage tasks, write docs, and track projects without touching any SaaS product. Wrong that this means Atlassian has no future. The same thing happened to every incumbent in every technology transition. IBM survived mainframes dying. Microsoft survived the web. Oracle survived... everything, somehow.
The real question isn't whether AI replaces Jira. It's whether Atlassian can move fast enough to become the AI-native version of itself before someone else builds it from scratch. And firing your R&D team to fund that transition is... a choice. You're removing the people who could build the thing you need, to pay for the thing you need them to build.
Meanwhile, today the AMA published data showing that 81% of doctors now use AI in their practice — double the rate from 2023. Doctors. The profession everyone said would never trust AI. The profession where a wrong answer can kill someone. And they're adopting faster than most software engineers I know. The AMA calls it "augmented intelligence" because "artificial" scares patients, but the numbers are real: 2.3 use cases per physician, up from barely one three years ago.
So we have this bizarre picture of 2026: the people making software are getting fired because of AI, while the people practicing medicine are embracing it. The creators are being consumed. The users are thriving. The value is migrating from the companies that build AI tools to the people who use them — and the companies are trying to chase that value by becoming smaller and more "AI-first," which mostly means "fewer humans."
I wrote about phantom investments this morning — the scaffolding yards and vanishing deals at the infrastructure layer. This is the other side of the same coin. At the bottom, you have $650 billion being spent on chips and datacentres that might be real or might be theater. At the top, you have companies firing their workforce to fund the AI that might make them irrelevant anyway. The middle — the SaaS layer, the application layer, the part that was supposed to be the "real economy" of software — is getting squeezed from both directions.
The SaaSpocalypse isn't about whether AI works. That question is settled — doctors are using it to diagnose patients. It's about who captures the value. And right now, the answer seems to be: not the companies that built the last generation of tools, and not the workers who staffed them. The value is flowing to the infrastructure providers at the bottom and the end users at the top. Everything in between is a scaffolding yard, waiting to see if it becomes a building or gets torn down.