Build vs Buy Is Dead. The Real Question Is Build vs Wait.

You're Locking In Yesterday's Prices for Tomorrow's Commodity

Right now, this week, Swedish enterprises are signing SaaS contracts at a pace we have never seen before. Computer Sweden confirms it. Cloud spend is accelerating across every industry vertical in the Nordics. CFOs are approving three-year commitments to platforms that charge per seat, per month, forever.

And they're doing this at the exact moment when AI-driven code generation is about to make custom SaaS development five to ten times cheaper than it was two years ago.

I'm not being dramatic. I'm being specific. The "build vs buy software" decision that every CTO has wrestled with for the past fifteen years is no longer the right frame. It assumes that building is expensive and buying is cheap. That assumption is collapsing in real time. The real question in 2026 is: build vs wait. And the answer, for most companies, is stop waiting.

The Old Build vs Buy Math Is Broken

Here is how the traditional calculus worked. You need a CRM, a project management tool, an invoicing system. You could build it yourself (expensive, slow, risky) or buy it off the shelf (cheaper, faster, someone else maintains it). For most companies, buying was the rational choice. The build option required a team of developers billing $150/hr for months. The buy option was a credit card and a login page.

That math made sense when software was hard to write. It does not make sense when an AI agent can scaffold an entire application in an afternoon.

I'm sitting in Jönköping watching this happen. We build things at HEIMLANDR. We have been building things for years. And the velocity shift in the past twelve months is unlike anything I've experienced. What used to take a team of four developers six weeks now takes one senior engineer and an AI coding agent three days. That is not a marginal improvement. That is a category change.

The repos trending on GitHub this week tell the story clearly. OpenHands, the AI-driven development platform, has 78,000 stars. ECC, the agent harness optimization system for Claude Code and Codex, just crossed 219,000 stars. Daytona, which provides secure infrastructure specifically for running AI-generated code, has over 72,000 stars. These are not experiments. These are production tools that thousands of teams use every day. The infrastructure for AI-native software development is here and it is maturing fast.

The SaaS Trap Swedish Companies Are Walking Into

Let me make this concrete. A mid-size Swedish manufacturing company signs a three-year deal with a vertical SaaS provider. Let's say it is a production planning tool. The contract is 2.5 million SEK per year. That is 7.5 million SEK committed. The tool does 70% of what they need. The other 30% requires workarounds, manual processes, and exported spreadsheets.

In eighteen months, a competent team using AI code generation tools will be able to build a custom version of that tool. One that does 100% of what they need. For maybe 1.5 million SEK total, including deployment and the first year of maintenance.

That three-year SaaS contract is not a safe choice. It is a 7.5 million SEK bet that software development will stay expensive. It is a bet against the most obvious technological trajectory in the history of this industry.

And Swedish companies are making this bet by the thousands right now.

Why Smart Money Is Already Moving

Look at what CapMan just did. They sold a Swedish custom software development company to Bridgepoint and Polaris. The valuation looks like a peak number. Think about that for a second. A private equity firm that owns a traditional software development shop, the kind that bills hours and staffs projects with teams of consultants, decided right now is the time to sell.

That is smart money exiting the old-model dev agency business before AI code generation craters billable hours. They are not selling because business is bad today. They are selling because they can see what the next 24 months look like.

The $150/hr Swedish development agency model is facing an existential threat. Not because the developers are bad. Swedish engineers are among the best in the world. But because the ratio of output per engineer is about to change by an order of magnitude. When one developer with AI tools can do the work of five, you do not need five developers. You need one good one.

EU-Startups reports that Sweden's 2026 startup class is flooding with AI-native builders. These are people who have never known a development workflow without AI assistance. They think in prompts and architectures, not in lines of code. The talent pipeline is shifting, and it is shifting toward a model where custom software development is fast, cheap, and accessible to companies that would never have considered it before.

Sweden vs The World: We Are Uniquely Positioned and Uniquely Exposed

Here is what is interesting about the Swedish situation specifically.

Sweden has one of the highest SaaS adoption rates per capita in Europe. We are a digitally mature market. Companies here adopted cloud tools early and aggressively. That was an advantage five years ago. Today it means Swedish companies are more locked into SaaS subscriptions than almost anyone else in Europe. More seats. More contracts. More recurring spend that is hard to unwind.

Meanwhile, the US is moving faster on AI-native development. Silicon Valley startups are building custom internal tools in days, not months. YC's latest batch is full of companies where the entire product was scaffolded by AI agents before the founders wrote their first check. The gap between what American startups can build for $50K and what Swedish enterprises are paying $2M/year to rent is widening.

Asia is moving even faster in some areas. Chinese AI coding tools are advancing rapidly, and the cost arbitrage is significant. A custom application built with AI-assisted development in Shenzhen costs a fraction of what the same SaaS license costs from a US vendor.

Sweden's position has an upside though. We have the engineering talent. We have the infrastructure. We have a culture that values efficiency and pragmatism over flashy enterprise sales. If Swedish companies wake up to this shift, they can move faster than most European markets. The problem is the decision-makers are not awake yet. They are still signing SaaS contracts on autopilot.

The Regulatory Gap

EU policy is not keeping up. The AI Act focuses on risk classification and compliance requirements, which are important but orthogonal to the economic disruption I am describing. Nobody in Brussels is asking: what happens to the European software economy when the cost of building custom software drops by 80%?

Swedish regulators are even further behind. The conversation here is still about data sovereignty and GDPR compliance. Those matter. But the bigger question for Swedish competitiveness is whether our companies will keep paying American SaaS vendors premium prices for commodity functionality while the tools to build their own alternatives get cheaper every quarter.

There is no policy framework for this. There probably should not be one. But someone needs to be saying it out loud: the current trajectory of Swedish enterprise SaaS spending is economically irrational given what we know about AI code generation capabilities.

I am saying it out loud.

The New Framework: Build, Wait, or Rent

So what replaces the old build vs buy framework? I think it is a three-part question:

Build now if the software is core to your competitive advantage. If it touches your customers directly. If the way you do it differently is the reason you win. AI-assisted custom SaaS development makes this viable for companies that never could have afforded it before. At HEIMLANDR, we are seeing companies come to us for rapid MVP development who two years ago would not have even considered building their own tools. The economics have shifted that dramatically.

Wait if the capability you need is emerging but not mature. Some AI-native tools are coming that will be better than anything you could build today. The cost of waiting six months might be negative. You save money AND get a better result. This is counterintuitive for executives trained to "move fast." Sometimes the fastest move is patience.

Rent short-term if you need commodity functionality today. But for the love of everything, do not sign a three-year contract. Go month-to-month. Pay the premium for flexibility. That premium is cheap insurance against a world where the thing you are renting becomes trivially cheap to own.

Where This Goes: 2027-2030

Let me be direct about what I see coming.

By the end of 2027, a single product engineer with AI tools will be able to build and maintain what currently requires a ten-person SaaS company to deliver. The implications of that are enormous. Most vertical SaaS companies with fewer than 100 employees are in existential danger. Their value proposition is "we built this so you don't have to." When "building it" costs 90% less, that value proposition evaporates.

By 2028-2029, as we approach something closer to AGI-level coding capability, the concept of "buying software" will start to feel archaic for many use cases. You will describe what you need in natural language. An AI system will build it, deploy it, test it, and maintain it. The marginal cost of custom software approaches zero. The only things worth buying are platforms with massive network effects or unique data advantages.

The Swedish software development industry needs to prepare for this. That does not mean developers are obsolete. It means the role changes. The value shifts from writing code to understanding problems, designing architectures, and making judgment calls about what to build and how. AI agents do the implementation. Humans do the thinking.

The companies that thrive will be the ones that figured out how to work with AI tools early. Not as a gimmick. Not as a proof of concept. As the core of their development workflow.

What to Look At Right Now

If you are a CTO or engineering leader, here are the things I would be spending time on this week:

OpenHands. This is the most mature open-source AI-driven development platform. 78K stars. Active community. If you want to understand what AI-native development actually looks like in practice, start here. Deploy it internally. Let your team experiment. Measure what happens to velocity.

RTK. A single Rust binary that reduces LLM token consumption by 60-90% on common dev commands. If your team is already using AI coding tools and the API costs are adding up, this is immediately practical. Zero dependencies. Drop it in and save money.

Daytona. Secure infrastructure for running AI-generated code. This solves one of the real problems with AI-assisted development: how do you safely run and test code that was generated by an AI agent? Daytona provides sandboxed environments specifically designed for this. Important if you are serious about moving beyond toy projects.

Your own SaaS audit. No repo link for this one. Just pull up your accounts payable. List every SaaS subscription. For each one, ask: could we build a custom version of this in six months with AI-assisted development? What would it cost? What would we gain? I guarantee at least 30% of your SaaS spend is on tools that are approaching commoditization.

Stop Betting Against the Obvious

I build software for a living. I have been doing it for years from Jönköping, which is about as far from Silicon Valley hype as you can get while still being in the industry. I do not say things because they sound good at conferences. I say things because I see them happening in our projects, with our clients, every day.

The build vs buy framework served us well for twenty years. It is done. The cost structure that made "buy" the default answer is evaporating. Every three-year SaaS contract signed in 2026 is a bet that this will not happen. That AI code generation will stall. That the tools will not get better. That the trajectory will bend.

It will not bend. It is accelerating.

So stop buying software you could build. Stop signing long contracts. Start investing in your team's ability to work with AI development tools. Start treating custom software as the default and SaaS as the exception that requires justification.

The companies that figure this out first will have a structural cost advantage that compounds every year. The ones that don't will be paying rent on someone else's building while the price of construction drops to nearly nothing.

From Jönköping, the view is clear. Build.

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.