SUSE AI launched to support open, enterprise-grade AI

Do you believe in open source? Many people claim to support open source, but only a small handful actually walk the talk.

I couldn't make it to susecon 2025 last week, so I jumped at the chance for a face-to-face chat with SUSE CEO Dirk-Peter van Leeuwen and Josep Garcia in Singapore this afternoon.

Here's what I learned about SUSE and open source.

My brush with open source

Quick backstory: I had my first taste of open source in my teens.

Home Internet routers were not available back then, but I needed something to share my Singtel Magix "broadband" with my fledging 10Mbps home network.

I did have a spare Pentium 100 PC though. After loads of fudging, I managed to install Debian Linux, loaded a script to recognise the ATM networking card and got NAT working.

So cool right.

I even ran Counterstrike's Linux server to host games on it. However, since I didn't know about patching, someone soon hacked it and replaced my login prompt with a "Hacked by" note.

So embarrassing.

I guess you can say I have a long and complicated relationship with open source.

How open is open?

Back to my meetup with DP and Josep. I think they truly believe in open source, and not, you know, rolling back access to source code for non-customers.

I imagine that's why SUSE doubled down on open source by offering "multi-Linux" support where it'll even support certain Linux distributions belonging to its competitors.

According to DP, open source is about:

• Sharing ideas.
• Innovating faster.
• Building communities.

What struck me in particular was DP's remarks that open source has no borders.

Going open source for AI

AI is the talk of the town these days. What is SUSE doing about it?

Where some see AI as the next opportunity to lock in customers, DP is clear that he wants customers to be able to innovate with an open, enterprise-grade AI platform.

SUSE AI launched last week:

• Completely private AI.
• Focus on observability.
• Support for agentic workflows.
• Comes with guardrails technology.

I haven't had the chance to dig into it yet, but I hear that it supports GPUs from Nvidia and AMD for inference.

Will keep my eyes open about how it develops.