Several organizations are experimenting with artificial intelligence, using AI models from major U.S. players. These models are effective, but they also come with significant risks. You have no control over their availability, and what happens to your precious data? Fortunately, there is another way. A groundbreaking collaboration between two companies, Piros and ReBatch, alongside Red Hat, Dell Technologies, and NVIDIA is making sovereign AI possible.

American AI models work well, but do we really want a future in which government bodies and medium—to large companies depend on American AI models that run on infrastructure from American cloud providers like Microsoft, Google, and Amazon?

Probably not . . . or at least not in all cases. But what’s the alternative? At a recent Sovereign AI Event, ReBatch unveiled an exciting new solution it has been working on with its partners Piros, Red Hat, NVIDIA, and Dell. They’re the first to deliver an AI model that allows you to retain full control.

What’s the problem?

What if a company like Microsoft or Google suddenly stops offering the AI model you’re working with? Or they suddenly put out a different model when you’ve already tailored your application to the first one? Will the model continue to deliver the same results with your application?

As a Belgian company or government body, you can’t be certain that the answer to that last, crucial question is yes. And that’s precisely because you don’t manage the model yourself. It’s controlled by an American multinational and ultimately always serves the needs of that company, not those of the customer. That’s just how it is.

So, what’s the alternative?

Many organizations think they cannot develop an AI model because it would be expensive and difficult. But that’s no longer true. Belgian companies and institutions can now choose Sovereign AI: artificial intelligence they set up and control entirely in-house. During the Sovereign AI Event, we demonstrated how this works.

Each partner provides a crucial piece of the puzzle. The large language model you need for AI, and for which you normally rely on American tech companies, is built by ReBatch. They start with open-source AI models, and the result is an AI model that’s equivalent to a model from, say, Microsoft but tailored to your organization and your language, whether it’s Dutch, French, German, and so on.

What are the other pieces of the puzzle?

To make an AI model work, you of course also need infrastructure. NVIDIA is the perfect partner for the graphics processors. The company is the global market leader in GPUs, with which you can train AI models. But NVIDIA also offers the best software to run AI models on their GPUs, which of course makes perfect sense. Who knows the hardware better than the manufacturer itself?

You choose Red Hat as the operating system for our AI models — Red Hat Enterprise Linux (RHEL), RHEL AI, and Red Hat OpenShift AI, on which you develop, train, and use AI models for business applications. Why Red Hat? Because, like Rebatch, we have a very close relationship with the open-source community. For setting up and managing the Red Hat environment, Piros as the largest Belgian Red Hat Partner is the obvious choice.

Finally, all that software and the AI model demand powerful servers. Dell can do that. They’re able to supply the right servers with the right GPUs, on which the AI model runs in a cost-effective way at the customer’s premises.

And the result?

If you put all these pieces together, you get a sovereign AI solution, where the hardware, software, and AI model operate entirely on-premises at the organization itself. You become completely independent of the American big-tech AI models, and you are certain that none of your data is being shared with third parties.

Are you eager to know more about Sovereign AI? Contact us for a no-obligation appointment.

Do you want to know more about our services? Discover them here.