Sweden's Digital Transformation Is Done TO Its Companies, Not BY Them

The 330-Year-Old Company That Needs Someone Else to Reinvent It

Here is the picture. It is May 2026. HCLTech just renewed and expanded its digital transformation deal with one of Sweden's largest commercial vehicle manufacturers. The core AI and software capability for that deal sits in Noida, not Gothenburg. BCG is publishing urgent warnings that Nordic companies need to treat transformation as a survival imperative, not a nice-to-have. And Microsoft, in a move that Computer Sweden reported on and almost nobody in the C-suite seems to be connecting to the rest of this puzzle, is killing volume discounts for Swedish enterprises on Azure. The cloud infrastructure that every one of these transformations depends on just got more expensive. Nobody is talking about that.

Meanwhile, TCS, an Indian IT services firm, is investing in Swedish engineering talent pipelines through AI hackathons. Not KTH. Not Chalmers on their own initiative. TCS. And Vinnova is funding academic papers on climate data while Karolinska workshopps what "trustworthy AI validation" means.

I sit in Jönköping. I build AI automation business tools and custom AI solutions for companies across Europe. And what I see from here is not a country leading. It is a country being led. By foreign consultancies, foreign cloud vendors, and foreign system integrators. Sweden's digital transformation is happening. It is just being done TO Swedish companies, not BY them.

The Outsourcing of Strategic Intelligence

Let me be clear about what I mean. I am not against outsourcing commodity IT. Staff augmentation, standard infrastructure, basic support. Fine. That is not what is happening here.

What is happening is that the most strategically important capability a company can build in 2026, the ability to design, deploy, and iterate on AI-driven systems that understand your specific business, is being handed over to external parties who have no long-term stake in your business survival.

When HCLTech runs AI transformation for Scania or Volvo, here is what actually transfers: domain knowledge. Every time an external team builds an AI model on your manufacturing data, your logistics flows, your customer behavior patterns, that team walks away with an understanding of your business that you yourself don't fully own. The algorithms might sit in your cloud tenant. The institutional understanding of how to make them work, how to retrain them, how to evolve them. That walks out the door.

This is not a services problem. This is a strategic sovereignty problem.

And it gets worse. BCG's latest Nordic report frames transformation as something companies "must do." Read it closely. The framing is catch-up, not leadership. The implicit message: you are behind, and you need to hire us to fix it. This is not analysis. It is sales. And Swedish boards are buying it because they do not have internal teams who can push back with better ideas.

The Nordic Mid-Market: Not Even Outsourcing

The big companies at least have a plan, even if I think it is the wrong one. The real crisis is in the Nordic mid-market. Manufacturers with 200 to 2,000 employees. Logistics companies. Industrial suppliers. The companies that make up the actual backbone of the Swedish economy.

These companies are not outsourcing their AI strategy to HCLTech. They are doing nothing. Or worse, they are doing "pilot projects" with no path to production. A proof of concept with a chatbot. A dashboard that nobody checks. A machine learning model that was impressive in the demo and has been collecting dust for eight months.

I talk to these companies every week. The pattern is always the same. They know something is happening. They read Dagens Industri. They hear "AI" at every conference. But they do not have a CTO who understands AI agent development. They do not have a team that can evaluate whether a vendor's claims are real. They are paralyzed between the fear of missing out and the fear of wasting money.

And the Swedish ecosystem is not helping them. Vinnova funds research. Business Sweden promotes exports. Nobody is funding or building the practical bridge between "AI exists" and "here is how a 500-person manufacturing company in Värnamo deploys it in production in 90 days."

How This Looks from Jönköping vs San Francisco

In the US, the conversation is different. Not better, necessarily. But different. American mid-market companies have access to a dense ecosystem of AI-native firms, open-source tooling, and a venture-funded startup layer that competes aggressively to serve them. There are problems with that model. Hype cycles. Vaporware. Companies that raise $50M and ship nothing. But the infrastructure of capability exists.

In China, the government has made AI deployment in manufacturing a policy priority with real funding behind it. Not research grants. Deployment subsidies. Tax breaks for automation. Direct state investment in capability building at the company level.

In India, firms like HCLTech and TCS have built a global business precisely because other countries, including Sweden, created a vacuum. They are good at what they do. I respect the execution. But every contract they win in Sweden is a contract that a Swedish or Nordic firm could have won if the ecosystem were being built properly.

The EU's approach, and by extension Sweden's, is regulatory-first. The AI Act is real and it matters. But regulation without corresponding investment in capability is just a tax on people who are already behind. You cannot regulate your way to competitiveness. You can only regulate your way to compliance. Those are not the same thing.

From Jönköping, which sits in the middle of Småland's manufacturing heartland, I see this gap every day. World-class physical engineering. Factories that would impress anyone. And digital capabilities that are a full generation behind what the products deserve.

The Microsoft Squeeze Nobody Talks About

Now layer in the Microsoft pricing move. Killing volume discounts for Swedish enterprises means the operational cost of cloud infrastructure is going up. For companies already spending millions on Azure, this is a real hit. For companies thinking about starting their digital transformation, it just got more expensive before they even begin.

But here is the part that frustrates me. Almost nobody is connecting this to the broader picture. If your AI transformation runs on Azure, and Microsoft controls pricing, and HCLTech controls the implementation, what exactly do you control? Your data? Maybe. Your models? Depends on the contract. Your ability to switch providers? Probably not, because you don't have the internal team to evaluate alternatives.

This is vendor lock-in on multiple levels. Infrastructure. Services. Knowledge. And every level of lock-in reduces your strategic freedom.

Swedish companies need to think about this differently. Open-source infrastructure is not a hobby project for startups. It is a strategic hedge against exactly this kind of dependency. Tools like Daytona for secure AI code execution, Agno for production-grade AI agents, and frameworks like OpenHands for AI-driven development are maturing fast. They are not toys. They are alternatives to writing blank checks to Redmond and Noida.

Where This Goes: 2027-2030

Let me project forward, because this is where it gets genuinely uncomfortable.

AI agents are getting more capable every quarter. By 2027, a well-designed AI agent system will be able to handle 60-70% of the cognitive work in a standard back-office function. Customer service, procurement analysis, quality control pattern recognition, logistics optimization. This is not speculative. The trajectory from what is shipping today makes this conservative.

By 2028-2029, as we move closer to what some labs are calling early AGI capabilities, the gap between companies that built internal AI competence and those that outsourced it will become a chasm. Companies with internal capability will iterate weekly. Companies dependent on external vendors will iterate quarterly, maybe. That speed difference compounds. In five years it is the difference between market leadership and irrelevance.

The regulatory picture will not save laggards. The EU AI Act creates compliance requirements but does not create capability. Swedish companies that spent 2024-2026 focused on compliance instead of capability will find themselves compliant, dependent, and losing market share to Asian and American competitors who built first and complied second.

I think Sweden has maybe 18-24 months to change the trajectory. After that, the dependency patterns harden. The institutional knowledge sits permanently offshore. The internal talent pipelines, the ones TCS is currently building because Swedish institutions are not, serve foreign interests by design.

What to Actually Do About It

If you run a Swedish company, or any Nordic company, here is what I would do right now.

First: Build an internal AI core team. Even if it is three people. Even if it is one person and two contractors. You need humans inside your organization who understand how AI agents work, how models are trained and deployed, and who can evaluate vendor claims with real technical judgment. This team is not a cost center. It is your strategic immune system. If you don't know where to start, we help companies build exactly this kind of capability at HEIMLANDR.

Second: Get something into production in 90 days. Not a pilot. Not a proof of concept. Something real, in production, touching actual business processes. A rapid MVP that solves one real problem for one real department. The goal is not perfection. The goal is organizational learning. You cannot learn about AI from PowerPoint decks. You learn by shipping.

Third: Audit your vendor dependency. Map every critical system. Who built it? Who maintains it? Where does the knowledge sit? If your primary AI vendor disappeared tomorrow, how long would it take you to recover? If the answer is "months" or "we don't know," you have a problem that needs solving this quarter.

Fourth: Invest in open-source infrastructure. Not because it is cheap, though it often is. Because it gives you optionality. When Microsoft raises prices, when your vendor changes terms, you need alternatives that you can actually execute on. Open-source AI tooling has matured to the point where this is practical, not idealistic.

What to Look At

Agno (39,800+ GitHub stars): A production-ready framework for running AI agents. If you are exploring AI agent development for business process automation, this is one of the most serious open-source options available. Python-based, well-documented, actively maintained.

Daytona (72,400+ stars): Secure, elastic infrastructure specifically designed for running AI-generated code. As AI agents write more of your software, the execution environment matters enormously. Daytona addresses this directly.

OpenHands (72,400+ stars): AI-driven software development. Not a gimmick. Real, functional tooling that is changing how small teams ship production code. Relevant for every company trying to do more with less.

Dagens Industri's coverage of BCG's Nordic transformation report: Read between the lines. Understand that "urgency" framing from a consultancy is also a business development tool. Then decide what urgency means for your specific situation.

The Uncomfortable Conversation Sweden Needs to Have

I write this on May Day, International Workers' Day. There is an irony. The Swedish labor movement was built on the idea that workers and companies should control their own destiny. That the means of production matter. That sovereignty over your economic future is not a luxury.

AI is the means of production now. And Sweden is handing it to foreign firms, foreign cloud providers, and foreign consultancies. Not because Swedish engineers are less capable. They are world-class. But because Swedish institutions, from Vinnova to the boardrooms of our largest companies, have defaulted to a procurement mindset when they need a building mindset.

You do not buy your way to AI competence. You build it. One team, one deployment, one production system at a time. We do this every day at HEIMLANDR.IO, from Jönköping. Small team. Real systems. No position papers.

The question is not whether Swedish companies will use AI. They will. The question is whether they will understand it, control it, and evolve it themselves. Or whether they will be passengers in their own transformation.

I know which side I am building for.

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.