We’re Drowning in Software
A year ago I wrote about vibe coding. Now anyone can build your product in a weekend — so where does the edge go?
Hey, this is Arttu. 👋 Welcome to a new edition of the Euro SaaS Edge.
A year ago, almost to the week, I wrote about Lovable — Sweden’s vibe-coding poster child, which back then added around 2m in ARR every single week. I admired its onboarding: a few minutes from the landing page to a working app. A year on, I want to return to that moment, because what felt novel then has quietly become the condition everyone now builds in.
Software was supposed to eat the world
In 2011, Marc Andreessen published an essay with a title that became the industry’s working thesis: “Why Software Is Eating the World.” For more than a decade he looked right. Software ate retail, media, banking, logistics, and taxis. If you could describe a process, someone wrote software to run it.
He was right. But “eating the world” had a side effect nobody put on the invoice. Once software is everywhere—and cheap to make— the world fills up with it. We didn’t end up with software eating the world. We ended up drowning in software.
What a year of vibe coding changed
When I wrote about Lovable, turning a prompt into a working application still felt like a magic trick. Twelve months later it is simply how software gets made. Lovable, Cursor, Bolt, v0, Replit — pick your tool. The cost of turning an idea into shipping software has not fallen. It has collapsed.
What used to take a funded team a quarter now takes one person an afternoon and a credit card. That is genuinely wonderful if you are a builder. It is rather less wonderful if you are trying to build a business.
The flood
The result was predictable, and it is already here. Every category is filling with near-identical, AI-powered tools. Another AI CRM. Another AI HR assistant. Another martech platform with SEO/GEO optimisation. Open Product Hunt on any given morning and scroll — you will lose count.
This is not a fringe impression; it is in my own data. Europe added around 1,750 net new SaaS companies in the year to March 2026, pushing the continent past 11,000 — ahead of the United States. And the new arrivals are overwhelmingly AI-native: depending on the market, 20–35% of the SaaS ecosystem is now AI-focused, with roughly one in four companies built around AI.
I made the point in that review that AI has stopped being a product category and become infrastructure, much like databases or cloud hosting before. That is the cause of the flood. When AI is simply infrastructure, building an AI-powered anything is no longer remarkable. It is the default. The barrier to building has effectively been removed. The barrier to being noticed has never been higher.
But are those real companies?
It is the right pushback and worth facing directly. Pure SaaS company counts measure formation activity, not revenue-bearing businesses. When AI tooling collapses the cost of starting, the numerator inflates faster than the underlying market. So, the honest question is whether companies with real customers and real revenue, aka quality-adjusted growth, are also up, or if the platforms must now filter out the majority of this noise.
The data I triangulate say: yes, but it reads “maturing,” not “booming.” The revenue-segmented cohorts from GP Bullhound’s European survey — the €5m to €25m ARR band — showed a median 2-year CAGR of 35%, with ARR per head climbing from €95k to €143k. Funding and unicorn data tell a similar story: deal counts are down, cheque sizes are up, and there are around 400 European unicorns, with the scaling gap to $100M ARR still unclosed.
So new companies are forming fast. Real, revenue-bearing businesses are growing more slowly. Both can be true. And the gap between the two is not a flaw in the data — it is the story. When the cost of building collapses, count and quality diverge. That divergence is the new market.
When building is free, the feature is not the moat
Here is the part founders are slowest to accept. If you can vibe-code your favourite tool in a weekend, so can your competitor. And — this is the uncomfortable bit — so can your customer.
A feature you can build in an afternoon is not a moat. It is a screenshot. The whole logic of “we’ll win because our product does X” assumes that X is hard. For a growing share of X, that is no longer true.
So where does the edge go?
It does not disappear. It moves. When building is cheap, differentiation shifts to the things a prompt cannot generate:
Proprietary data and the workflow built tightly around it. Anyone can build the interface, but not everyone has the data that makes it useful.
Distribution. Being where the customer already is, with a brand they already trust, beats a better feature they never see.
Trust and compliance. Slow to build, impossible to fake, and increasingly the first question a B2B buyer asks.
Depth. The unglamorous last 20% of a real workflow — the edge cases, the exceptions — that nobody prompts their way through.
Switching costs are earned honestly, through genuine integration into how a customer actually works.
Notice what these have in common. None of them are produced by writing code more quickly. The moat has moved from building the product to being the one the customer chooses.
Why this is good news for Europe
I’ll be honest: I find the news genuinely optimistic for European SaaS, and I do not say that often.
For a decade, the game rewarded speed and capital — deploy money, hire fast, and out-ship everyone. That was never Europe’s strongest hand, and our struggle to reach $100M ARR partly reflects it.
However, trust, compliance, depth, and long-term customer relationships — the things that matter when a feature is commoditised — are precisely where European SaaS has always been (too quietly competitive). The flood acts as a leveller. When everyone holds the same building tools, the advantage swings to whoever the customer believes and whoever does the boring, deep, specific work. That is a more European game than the last decade ever was.
So what do you do?
If you are building in a crowded AI category, the instinct is to ship faster. Resist it. Speed is now table stakes, not an edge.
Software did eat the world. The mistake was assuming that it made software valuable. It made software cheap. What is valuable now is everything around it — the data, the trust, the distribution, the depth. So please stop asking whether you can build it. Everyone can. Ask why anyone would choose yours.
Thanks for reading Euro SaaS Edge. If this was useful, subscribe for free to get new posts.