So let me tell you a story about a company that wanted to launch an EU-native application and AI stack. At the beginning of 2025, we started actively thinking about a truly EU-native proposition at Greater. Political and public opinions about big tech companies were starting to shift, and compliance and regulation in the EU were increasing rapidly.
Since most of us are engineers at heart, we began by setting out some clear parameters:
- Everything we use needs to be owned/licensed by an EU company or open source
- Open source must be under Apache, MIT, or GPLv3 (or compatible licenses)
- Extra scrutiny will be applied when open source software is partially or fully owned by companies outside the EU
Humble Beginnings
We had our parameters, we had our vision and now we were off to the races… or were we? We kicked off with an initial meeting to select a number of technologies and vendors we wanted to work with. And guess what? A lot of it was owned and maintained by companies outside of the EU. But even more surprisingly, a lot wasn’t. The market has grown. With everyone focusing on the big tech companies, many have overlooked a vibrant ecosystem of companies that have long embraced principles like privacy and digital sovereignty. Even something “simple” like selecting a cloud provider proved to be a challenge. The space of European hosters and cloud providers is much larger than we ever imagined.
Now, I could go into all the details of why one is better than another but to be honest, most of them have their own strengths and weaknesses.
Exoscale
We eventually chose Exoscale as our infrastructure partner, simply because we had a good feeling and some great experiences with their development teams. Let’s be very clear: Greater is a sizable consultancy partner in the DevOps/Cloud Native space, but Kubiqo is just getting started. Still, Exoscale saw us as a serious partner and treated us accordingly from day one. This wasn’t always the case with other cloud providers. Our experiences ranged from “Sure, let’s be partners, please pay us a few thousand euros a month” to no response at all.
Now, the initiated among you might say: Wait a minute Exoscale is Swiss, not an EU company. That’s somewhat correct. However, A1 Digital has completely acquired Exoscale and A1 Digital is an Austrian company. So, we’re back in the EU.
Back to SUSE
With infrastructure taken care of, the next question was: what are we going to run? This part was a no-brainer. We wanted to host modern applications, so Kubernetes was the natural choice for its unparalleled scalability and automation. Greater already had a partnership in the Cloud Native space with SUSE. Still, we initiated a fresh selection process to ensure the technology truly matched our service offering. In the end, products like RKE2, K3s, Rancher, and SUSE Security (NeuVector) were too good to ignore. So after some discussion, we decided to partner up with SUSE again this time not just as a consulting partner, but as a Managed Service Provider (MSP).
The support we received was outstanding. SUSE actively helped us during development and even squashed a few bugs along the way.
Getting into the AI Hype
And then we had to address the commercial pink elephant in the room: AI. The AI hype is real. As engineers, we can’t just close our eyes and call it marketing fluff. Even our daily work has changed AI now helps us find information faster and acts as a “rubber duck” for testing our own ideas and code. Still, we approached AI with the same parameters as the rest of our platform. This created an interesting challenge, and once again SUSE came through with their SUSE AI proposition.
Together, we started forming an idea: integrate local AI capabilities into our application stack. We worked with the SUSE AI development team on a proof of concept that ran AI workloads within applications using a fully local AI stack based on Kubernetes and backed by Exoscale GPU infrastructure. We were excited. Running open source LLMs like Llama or Gemma locally fully integrated into our apps, without ever needing to leave EU borders was a breakthrough. We then expanded the platform with RAG (Retrieval-Augmented Generation) training capabilities using a vector DB. Now you can run LLMs that are enhanced with company- or application-specific knowledge, all within the EU.
Closing Statement
Of course, we use many other tools like CI/CD, automation, observability, and more but we’ll cover those in a later article. That’s another story altogether. As a closing thought, I want to sincerely thank the three companies that helped form Kubiqo: Exoscale, SUSE, and Greater thank you for your trust and support. And to everyone interested in working with us whether as a customer, partner, or colleague don’t hesitate to reach out through LinkedIn or send us an email at info@kubiqo.io