GitHub Wants Your Code. The SaaS Industry Wants Your Money. It's Time to Leave.

Yesterday, GitHub announced a quiet change to its Copilot data policy. Starting April 24, the company will use interaction data — your inputs, your outputs, your code snippets, your comments, your file names, your repository structure — to train its AI models.

Unless you opt out.

Let’s be clear about what’s happening here. GitHub isn’t asking. They’re telling. The default is in. You have to find the setting, navigate to it, and toggle it off. If you don’t? Your work belongs to them.

The Double Squeeze

Here’s what makes this particularly gross: this isn’t a company struggling to survive. This is a subsidiary of Microsoft, a $3 trillion corporation, deciding that the free and pro-tier developers who built their platform are now raw material for their AI pipeline.

Enterprise customers are exempt, of course. If you’re paying enough, your data stays yours. But if you’re an individual developer, a freelancer, or a small team on Copilot Free or Pro? You’re the training data now.

This is the same playbook SaaS companies have been running for a decade, but the stakes just got higher. Your code isn’t just code anymore — it’s intellectual property that could end up in a model competing with you.

Meanwhile, the SaaS House Is Burning

While GitHub was drafting its privacy update, something much bigger was happening in the markets.

For the first time in history, software stocks are now trading at a discount to the broader S&P 500. Not at parity. Below. The iShares software ETF (IGV) has dropped over 21% year-to-date, and roughly 30% since its September 2025 peak. That’s approximately $2 trillion in market cap — gone.

The reason is simple: AI is breaking the per-seat pricing model.

The entire B2B SaaS industry was built on one assumption — that headcount would keep growing, and every new hire needed a seat. Salesforce, Workday, Atlassian, Monday.com — they all rode that wave for two decades. When headcount grows, revenue grows. Automatic. Durable. Predictable.

AI agents break that link entirely. If one agent can do the work of five employees, enterprises stop buying 500 seats and start buying 100. Or renegotiating entirely. The revenue model doesn’t slow down — it reverses.

The Human Cost

Last week, Atlassian laid off 1,600 people — 10% of its workforce — to “fund its AI push.” One-time severance costs: $225-236 million. Their AI agent, Rovo, has 5 million monthly users. The company is literally replacing humans with the product it’s building, and the humans they’re replacing are the same ones who paid for their subscriptions.

The Guardian interviewed one of the laid-off workers who described losing his “dream job” at Atlassian. His tasks were already being handled by the AI tools the company was building. There’s a brutal irony in a company firing its own customers to build the product that replaces them.

What This Means for Software Buyers

Here’s the part that should keep you up at night: when SaaS companies feel the squeeze, you feel the squeeze too. They’ll do exactly what GitHub just did — find new revenue streams by harvesting your data, changing terms of service, and pushing you toward enterprise tiers.

The pattern is already clear:

• Prices go up while per-seat value goes down
• Data policies shift — your work becomes their training material
• Free tiers get hollowed out while enterprise gets exempted
• Layoffs fund AI investments that replace the users who funded the layoffs

This isn’t one bad actor. This is an industry in structural decline, and the people paying for it are the ones who can least afford to push back.

The Exit Strategy

The answer isn’t to negotiate better terms with your SaaS vendors. It’s to stop needing them.

Open-source alternatives are better than they’ve ever been. Self-hosting is easier than it’s ever been. And for the specific cases where you need something custom? That’s literally what we built Bountymon for.

The SaaS rout of 2026 isn’t just a market event. It’s a signal. The per-seat era is over. The question is whether you’ll be the one paying for the transition, or the one getting paid for it.

GitHub’s Copilot policy change is a reminder: when you don’t own your tools, your tools own you. Opt out of the data collection, sure. But also start thinking about what it looks like to opt out of the whole model entirely.

Because the next policy change won’t be announced in a blog post. It’ll just show up in your invoice.