Your SaaS Vendor Is Now Your Competitor's Cheapest Build

Here is the number that should ruin your Monday. If your company pays 400,000 SEK a year for a SaaS tool, you are not buying convenience anymore. You are subsidizing the R&D budget of a company whose AI models are, right now, cheap enough to let your competitor build a better version of that exact tool in six weeks. That is not a hypothetical. That is the actual state of build vs buy software decisions in August 2026, and most Swedish leadership teams have not updated their mental model since before ChatGPT could write a working function.

I run HEIMLANDR from Jönköping. Small city, not Stockholm, definitely not San Francisco. And from here the picture is clearer than it is from a glass tower in either place, because we do not have the luxury of pretending SaaS subscriptions are a rounding error. Every krona matters. So when I say the build vs buy question has been flipped upside down, I am not repeating a LinkedIn hot take. I am telling you what we see in client budgets every single week.

Build vs Buy Was Never About Cost. Now It Actually Is.

The old logic went like this: build is expensive and slow, buy is fast and predictable. Procurement teams built entire careers on that assumption. Gartner quadrants, RFP processes, three-year contracts with escalation clauses baked in. It made sense in 2019 because building software meant hiring five to ten engineers, running a project for a year, and hoping it worked.

That assumption died. Not gradually. It died the day AI coding agents got good enough to ship production code without a human writing every line. Spiceworks and itbrew.com are both writing about this right now, this week, because the math finally broke through to the people who control budgets. In-house software built with AI agents is genuinely competitive with vendor solutions for the first time in the history of enterprise software. Not "good enough for a prototype." Competitive. Shipping. Running in production.

What changed is not that engineers got smarter. What changed is that the cost of iteration collapsed. A senior developer with Claude Code, Cursor, or OpenHands running agentic workflows can now do in a day what took a team a sprint. Look at what is trending on GitHub this week if you want proof it is not just marketing. ECC has 239,000 stars because it optimizes agent harness performance, memory, and security for exactly this workflow. Graphify turns your entire codebase, docs, and database schemas into a queryable knowledge graph so an AI agent understands your system before it touches it. This is not toy tooling. This is infrastructure for building fast, in-house, with fewer people, at quality levels that make "just buy the SaaS" a genuinely uncomfortable question to answer in a board meeting.

The Vendor Lock-In Model Is Cracking, and Vendors Know It

You can see the panic if you know where to look. IBM and Red Hat are racing to build what they call "trust infrastructure" for AI-era open source. Read between the lines: that is an admission that the old model, where you lock a customer into a platform and extract rent for a decade, no longer holds when the customer's own AI can replicate 70% of your platform's value in a fraction of the time. BMC pushing Control-M as a "trusted AI orchestration" layer is the same story from a different enterprise vendor. They are not selling the whole stack anymore. They are trying to become the plumbing underneath a stack their customers now build themselves.

That is the tell. When the vendor repositions from "we are your software" to "we orchestrate the software you build," they have already conceded the argument. They know build is winning. They are just trying to stay relevant inside the build.

The Swedish Procurement Problem

Here is where I get uncomfortable, because this is where I live. Swedish companies, especially outside the Stockholm bubble, are still running procurement logic from before the AI coding agent era. I sit in meetings where a CFO proudly tells me they negotiated a good multi-year rate on a SaaS platform, as if that is a win. It is not a win. It is a three-year commitment to paying for software that a competitor with two good engineers and an AI agent subscription can now build for less than the first year's license fee.

Compare that to what I see in reporting from Dagens Industri and Breakit this year. The Nordic startup scene understands this shift faster than Nordic enterprise does. Founders building lean, AI-native companies are treating custom software as the default, not the exception. Meanwhile mid-size Swedish industrial companies, the backbone of this country's economy, are still running annual SaaS renewal cycles designed for a world where writing software required a team of ten.

The US has the opposite problem in a way. Silicon Valley moves fast on the build side but the SaaS incumbents there are massive, entrenched, and have the capital to buy or crush anything that threatens them. In Sweden we do not have that scale of incumbent gravity. That should be an advantage. It should mean Swedish CEOs move faster into build-first thinking because there is less institutional weight pulling them back. Instead I see the opposite: a kind of cautious, risk-averse procurement culture that treats "we built it ourselves" as a liability rather than a competitive edge.

Asia, particularly the fast-moving parts of the Chinese and Southeast Asian tech ecosystem, has fully internalized build-first thinking already. They were never as SaaS-dependent as the West to begin with. Sweden risks being caught in the middle: too expensive to compete on cost, too cautious to compete on speed.

Is Swedish and EU Policy Ready for This? No.

This is the part nobody wants to say out loud. The EU AI Act is built around a world where AI is a product feature you regulate at the point of deployment. It was not designed for a world where AI is the engineering department itself, quietly replacing SaaS vendors from the inside. There is no regulatory framework right now that deals seriously with the fact that companies are training and fine-tuning models on data flowing through the SaaS tools they subscribe to, and then that same underlying AI capability gets sold back to the market as a competing product.

Swedish regulators are focused on data protection and AI transparency, which matters, but nobody in government is asking the actual economic question: what happens to the Swedish software services sector when the build-vs-buy calculation flips this hard, this fast? Sweden has an excellent developer talent base and a genuinely good position to lead in AI-native software development. But leading requires policy that supports fast-moving small teams building serious software, not procurement frameworks written for an era of five-year vendor contracts.

Where This Goes: The Next Two to Five Years

The trajectory is not subtle. As AI capability keeps compounding toward something closer to general-purpose reasoning, the cost of building custom software keeps falling toward zero marginal cost for a huge share of what SaaS vendors currently sell. Not all of it. Deep infrastructure, security-critical systems, and genuinely hard distributed systems problems still need real engineering judgment. But the mid-tier SaaS market, the CRM add-ons, the internal tools, the workflow automation platforms, the reporting dashboards, that entire category is walking into an existential crisis.

Here is what I expect by 2028. SaaS vendors that survive will have repositioned as orchestration and trust layers exactly like IBM, Red Hat, and BMC are doing right now, or they will have become AI companies themselves, selling access to models rather than access to features. Everyone else gets replaced by internal builds that took a team of three and a subscription to a coding agent.

As we move closer to AGI-level capability, and I think we are closer than most CEOs are comfortable admitting, the build side of this equation gets stronger every quarter, not weaker. The gap between "what a small team can build with AI agents" and "what a SaaS company charges for the same capability" will keep widening in favor of build. Companies that wait for certainty before shifting strategy will wait themselves out of the market.

What To Actually Do About It

Stop treating build vs buy as a procurement exercise and start treating it as a strategic audit. Pull your top ten SaaS contracts by spend. For each one, ask honestly: could two senior engineers with modern AI tooling replace 80% of this in under two months? If the answer is yes, you have found where your competitors are about to eat your margin.

What To Look At

If you want to see what serious build-first teams are actually using right now, these are worth your time. OpenHands is an open AI-driven development environment that shows what agent-driven engineering looks like at a production level, not a demo level. Ponytail is worth studying for a different reason, it optimizes for the code you never have to write at all, which is the real lesson here: the winning move is often not building more, it is building less and smarter. And if your team is drowning in API testing overhead while trying to move fast, look at Bruno or Hoppscotch as open alternatives to the Postman subscription you are probably overpaying for right now, which is itself a small, satisfying example of the exact pattern this whole article is about.

This is exactly the work we do at HEIMLANDR. We are not a SaaS development company that tries to sell you a twelve-month roadmap. We build fast, with AI agents doing the heavy lifting where it makes sense and senior engineers doing the judgment work where it does not. If you want to test the theory before you bet the company on it, start with an MVP development sprint on the SaaS tool costing you the most and see what a real build timeline looks like in 2026. For teams ready to go further, our AI agents practice and custom SaaS development work exist for exactly this shift, replacing rented software with owned, AI-accelerated systems that get cheaper to run every year instead of more expensive.

Sweden has the engineering talent, the trust culture, and the technical education to lead this shift instead of getting run over by it. What we lack right now is enough CEOs willing to admit their procurement playbook is five years out of date. Software development Sweden can win this decade if leadership treats build vs buy as a live strategic question instead of a solved problem from 2019.

The SaaS vendor invoice sitting in your inbox right now is not a fixed cost. It is a live decision you are re-making every renewal cycle, whether you realize it or not. Look at it again before you sign.

Fredrik Brunnberg is the CEO of HEIMLANDR.IO, building AI and software solutions from Jönköping, Sweden. This is the daily HEIMLANDR briefing. If you found this valuable, share it with someone who builds things.