Franz Seiser

You still have people on duty that watch very carefully if it's going right.

If you see things are not going as expected, you can stop it.

Of course, we also make sure we can roll back quickly.

So we see an upgrade is going wrong by the click of a button.

So to say you roll back to the original situation to also be as much safe as you can.

You can easily compare that with the production side, which is highly automated.

Still, you have a few people in the command center making sure the system is running smoothly.

I don't know if my glass ball is so much better than yours.

But what we see, of course, this route going down, everything is fully softwareized, everything is automated, operations, of course, coming into play much, much more.

All these AI topics we see is going to take a bigger and bigger role.

The good thing is, again, our architecture is helping here tremendously because one of the big benefits is these whole systems are generating data in real time or near real time, which, again, is a big change.

In the old days, you had one data point roughly every five minutes.

With one data point every five minutes, an AI system cannot do anything because in the five minutes, a lot can happen.

So the more data they have and the more accurate the data is, the better they can act upon this data.

And that's clearly what we're going to see happen.

They can also see if things start to behave strangely, that the network itself can start to heal itself.

At the very minimum, you need connectivity.

And this is where we are coming from, right?

So there is clearly a role.

How much we need to put AI distributed and what's the right edge?