Jason Fried published a post this week arguing that a bespoke software revolution isn’t coming. He’s right. Most people don’t want to build software. They want their problem solved.

The three-person accounting firm doesn’t want a custom invoicing system — they want the invoices out the door. The logistics company doesn’t want Joe’s new system — they want the routes optimised. The excavator analogy is good: giving everyone access to a powerful tool doesn’t turn everyone into a contractor.

Fried is correct. And SaaS vendors who are nodding along are drawing the wrong conclusion.

What SaaS vendors are hearing

They’re hearing: our customers aren’t going to build their own alternatives, so we’re safe.

That’s not what’s happening. The threat was never that your customers would fire you and roll their own. The threat is that someone else builds a better, cheaper, tighter version of your product — and for the first time, they can do it for a fraction of what it used to cost.

Building a credible SaaS competitor used to cost roughly $500K and 18 months. The emerging estimate is closer to $5K and a few weeks. That’s not a trend. That’s a structural shift that has already occurred.

The moat was never love

SaaS vendors have long known their customers don’t love them. The moat was switching costs — data lock-in, workflow entrenchment, the pain of retraining 200 staff on a new system. High net revenue retention wasn’t a measure of delight; it was a measure of friction.

That friction is the thing eroding. Not because customers are building their own tools. Because:

AI is eating migration cost. An AI agent can extract, transform, and migrate data between systems in hours. What used to require a six-month IT project and a $200K consulting engagement is becoming a weekend task. The EU Data Act compounds this — as of September 2025, European SaaS vendors are legally prohibited from charging exit fees and must provide data portability on demand.

The cost to build a challenger has collapsed. AI-native startups now reach $10M ARR 2.4x faster than traditional SaaS companies. Y Combinator reports a 60% decrease in MVP development time since 2022. A well-capitalised team with a focused vertical thesis can now build a credible alternative to your product before your next board meeting.

The $285 billion signal

In February 2026, enterprise software stocks lost $285 billion in market cap in a single session after Anthropic released Claude Cowork plugins. The press framed it as “AI will replace SaaS.” That’s not what the analysts said.

Morgan Stanley called it “clear competitive pressure” — not from customers self-building, but from Anthropic shipping a competing product. Thomson Reuters and RELX didn’t fall because their customers were going to build their own legal research tools. They fell because a well-funded competitor with better AI infrastructure just entered their market.

The lesson isn’t that SaaS is over. It’s that the horizontal, per-seat bundle with commodity features and high switching costs is the specific thing the market stopped believing in. Vertical SaaS in regulated industries barely moved.

The hostage scenario

Here’s the thing Fried didn’t say, because it wasn’t his point: some SaaS vendors aren’t just banking on switching costs — they’re relying on them while actively declining to improve.

They know renewal rates stay high not because customers are happy, but because the migration cost is prohibitive. So they raise prices. Let the product stagnate. Deprioritise support. They’ve done the math, and the math says customers are stuck.

That calculus is breaking. Not all at once, and not evenly — but the companies Chargebee calls it “business model debt” — the ones running 100% NRR by expanding trapped customers rather than winning happy ones — are the most exposed.

When a vertical challenger shows up with better UX, a lower price, and an AI agent that handles the migration, the customers who were “too locked in to leave” discover they were never locked in. They were just waiting for the door to open.

What this means if you’re building

If you’re an early-stage team choosing tools: bet on vendors who are earning renewal, not just collecting it. Look for products where the moat is proprietary data, genuine workflow depth, or network effects — not just the pain of leaving.

If you’re building a product: the window to go after an entrenched SaaS player in a specific vertical is as open as it has ever been. The switching cost moat that protected them is eroding. What you need is a focused wedge, a clear migration story, and the nerve to move fast.

Fried is right that most people just want the hole dug. The opportunity is being the company that digs it better — and now, for the first time, you don’t need a JCB to compete.

If this is the kind of bet you’re making, we help early-stage teams figure out where to point the shovel — jfsi.io.

Sources: Jason Fried — A bespoke software revolution? I don’t buy it. · The SaaS CFO — The SaaSpocalypse: AI Agents, Vibe Coding, and the Changing Economics of SaaS · Tech Startups — Anthropic’s Claude Plugins Spark $285B Selloff · Bloom VP — The New Software Moats · Chargebee — 2026’s Real SaaS Threat Isn’t AI, It’s Business Model Debt · Clairfield — How the EU Data Act Rewrites the Rules for SaaS Providers · The Register — Rise of AI Means Companies Could Pass on SaaS · Thinking Tech Stocks — The Great Software Correction of 2026