OpenCode Hits 120K Stars — And Atlassian Is Firing 1,600 People to Build What You Can Get for Free
While enterprise SaaS giants lay off thousands to fund proprietary AI, open-source coding agents are hitting critical mass. The math is getting ugly for vendors.
This week told you everything you need to know about where enterprise software is heading — and it’s not in the direction SaaS vendors want.
The Numbers Don’t Lie
OpenCode, the open-source AI coding agent, just hit 120,000 GitHub stars. Five million developers use it monthly. It supports 75+ LLM providers, runs locally, stores zero code on external servers, and costs exactly $0. It went viral on Hacker News this week with 293 points and 137 comments — the kind of engagement that usually signals a project crossing from “promising” to “inevitable.”
Meanwhile, Atlassian laid off 1,600 people this week — 10% of their workforce — and is spending $225-236 million on restructuring. Their stated reason? “Pivoting to AI.”
Read that again. You’re paying Atlassian $20-50 per seat per month for Jira and Confluence. They’re taking your money, firing the humans who built those products, and redirecting the savings into AI features you didn’t ask for. The Arc browser team — acquired by Atlassian — was reportedly hit particularly hard.
This is the SaaS model in its late-stage decomposition: charge more, deliver fewer humans, wrap the difference in “AI.”
What OpenCode Actually Means
OpenCode isn’t just another coding tool. It’s proof that the build-vs-buy equation has fundamentally shifted:
- Terminal-first, privacy-first: No code leaves your machine. Try getting that guarantee from a proprietary vendor.
- Model-agnostic: Use Claude, GPT, Gemini, or run Ollama locally. You’re not locked into anyone’s pricing treadmill.
- Multi-session, parallel agents: Start multiple coding agents on the same project. No per-seat pricing. No “contact sales.”
- 800 contributors, 10,000+ commits: This isn’t a weekend project. It’s an ecosystem.
The message is clear: if you can get enterprise-grade AI coding for free, why are you paying $19-39/month per developer for Copilot? Why are you locked into Claude Code when OpenCode lets you use Claude and GPT and local models in the same interface?
The Atlassian Pattern
Atlassian isn’t alone. Block did the same thing. So did dozens of smaller SaaS companies over the past year. The pattern is always the same:
- Raise prices (often via “AI add-ons” or seat-based increases)
- Lay off engineering teams
- Announce “AI-powered” features that replace the humans you were paying to support
- Ship those features six months late and half-baked
- Raise prices again
You’re not a customer in this model. You’re a funding mechanism for their AI R&D.
The Bounty Alternative
Here’s where it gets interesting for the software-buying conversation. OpenCode shows that when a problem is compelling enough, the open-source community builds a better solution — for free — faster than a billion-dollar company can.
The same dynamic applies to Jira (see: Linear, Plane), Confluence (see: Notion alternatives, GitBook), and every other tool in the enterprise stack. But the adoption gap is what keeps people paying. Teams don’t switch because switching is painful.
That’s exactly what bounties fix. Post a bounty to migrate your team from Jira to Plane. Post one to set up OpenCode with your preferred models. Post one to self-host the whole stack. You pay once for the transition, then you own the result forever.
No seat licenses. No annual renewals. No layoffs funding someone else’s AI pivot.
The Bottom Line
Five million developers are already using free, open-source AI coding tools. Enterprise vendors are cannibalizing their own teams to build competing products you’ll pay extra for. The gap between what’s free and what’s “enterprise” has never been narrower — or more absurd.
The question isn’t whether open-source AI tools are good enough. It’s why you’re still paying for the proprietary ones.
Found this useful?
Share it with your team to start the conversation about SaaS savings.