intel red hat intel innovation

Red Hat and Intel roundtable at Intel Developer Conference 2023

SMBtech sat in on a roundtable lunch (please excuse the clattering cutlery at the beginning of the video). In it Intel and Red Hat executives discussed how RedHat started out, the birth of Fedora and the launch of CentOS Stream. There are many topics covered including how OpenVino has been embraced in Red Hat OpenShift AI to deliver training and inference from the datacenter to the Edge plus working with Intel’s integrated/discrete GPU and Gaudi Accelerator.

The three speakers are:

Chris Tobias GM, Americas Technology Leadership/Global Platform ISV Account Team at Intel Corporation

Gunnar Hellekson GM, Linux, Red Hat

Mark Skarpness Vice President and General Manager of System Software Engineering at Intel Corporation

Intel and Red Hat at Intel Innovation, 2023, San Jose.


00:00:00 Chris Tobias

All right. So welcome everyone. Thanks for coming spending time with us today. So we got a great panel here. We got Gunner Hellickson, who’s the vice president and general manager of the Red Hat Enterprise business unit. And we’ve got Mark Scarpas, who runs all the engineering and.

00:00:20 Chris Tobias

And SAG around open source ecosystem.

00:00:25 Chris Tobias

All right. Just just to do a quick soft opener and then we’re going to have a going to walk through a couple of slides. Do you want to talk about what, what what was announced today on stage with Greg Real Quick and then we can go through these?

00:00:41 Gunnar Hellekson

Actually, that’s what I’m doing. That’s all going to end with.


So should I just? Should I?

00:00:45 Speaker 4

Just go there, all right, no.

00:00:48 Gunnar Hellekson

Up to you. Is there any of you?

00:00:50 Gunnar Hellekson

Don’t want to wind it up anyway. You’re good.


No. All right.

00:00:54 Speaker 5

Not winding up anymore.

00:00:57 Gunnar Hellekson


00:00:58 Gunnar Hellekson

I’m going to.

00:00:58 Gunnar Hellekson

Stand up. Is that right? OK.

00:01:02 Gunnar Hellekson

Morning, everybody. I’m going to subject you to about 15 minutes of exposition on a little, some more detail on what Greg and I talked about on stage this morning.

00:01:15 Gunnar Hellekson

And I’m really excited about it. I’m excited about this. We have been partners for.

00:01:21 Speaker 6

Forever. Forever. Forever.

00:01:25 Gunnar Hellekson

20 / 2025 years, probably pretty close.

00:01:27 Mark Skarpness

Since the beginning of.

00:01:28 Gunnar Hellekson

Yeah, that’s right.

00:01:30 Gunnar Hellekson

But this is a we were talking about this at the reception last night. This feels qualitatively different. Something has changed in the relationship, and I think it’s largely due to like the some intention on Intel’s part.

00:01:44 Gunnar Hellekson

And it just so.

00:01:45 Gunnar Hellekson

Happens that it coincides.

00:01:46 Gunnar Hellekson

With some changes in our own collaboration strategies.

00:01:49 Gunnar Hellekson

And kind of.

00:01:49 Gunnar Hellekson

The two of these things are coming together.

00:01:51 Chris Tobias

In what I.

00:01:51 Gunnar Hellekson

Think is a really fun and really interesting way. So first I’m going to this is intimidating as a slide. Bear with me. I will make it consumable, I promise.

00:02:05 Gunnar Hellekson

I’m going to give you a little bit of history about how we used to work when Red Hat first started building Linux. We you raise your hand if you remember us being sold for $30.00 at Best Buy.

00:02:17 Gunnar Hellekson

At the retailer right. OK, that’s. That was where we started and our number one revenue stream from that. Was anybody anybody know how we made?

00:02:24 Gunnar Hellekson

Our money.

00:02:26 Gunnar Hellekson

Hats and T shirts. We sold hats and T shirts with the logo on it and that was our primary revenue stream. But we were very, very popular. And what folks we discovered actually we had two kinds of customers, one which was folks like me who were developers in their proverbial basements who needed a ready to use Linux so they could do build a web server and things that.

00:02:47 Gunnar Hellekson

They wanted to do.

00:02:49 Gunnar Hellekson

And enormous multinational banks running Oracle databases on Intel.

00:02:56 Gunnar Hellekson

Turns out it’s.

00:02:57 Gunnar Hellekson

Very hard to make both of these audiences happy.

00:03:01 Gunnar Hellekson

And so we split the work. We said that we’re going to create Red Hat Enterprise Linux and we’re going to create Fedora Linux and lots of differences between the two of them. But the big one from a developer point of view is that Fedora was going to update every six months. It was taking the latest and greatest from the upstream and putting it into a consumable format for the kind of hobbyist and developer user.

00:03:20 Gunnar Hellekson

And then there was going to be.

00:03:22 Gunnar Hellekson

Another version which was going to update every three years and sometimes 5 and sometimes six years. We were going to do a Red Hat Enterprise Linux release and we were going to commit to.

00:03:29 Gunnar Hellekson

Supporting that for a really long time.

00:03:32 Gunnar Hellekson

This model satisfied us for years, and as I talked about in the keynote, there was a big change between hybrid cloud, AI security, the appetite for customers to get early access to innovation and partners getting early access to innovation is enormous and we couldn’t we could give, we could offer folks Fedora.

00:03:53 Gunnar Hellekson

Which was updating every six months, and that’s interesting as kind of a place to see what’s happening, but not interesting as a place of actually building something worthwhile, something enduring.

00:04:03 Gunnar Hellekson

Right, in order to do that, you need something that looks a lot like a product that is going to be supported for a long for a long period of time, which is red.


Hat revised.

00:04:11 Gunnar Hellekson

Linux. So we created this kind of third category internally we’ve been saying calling midstream.

00:04:17 Gunnar Hellekson

I’m not sure.

00:04:17 Gunnar Hellekson

If that’s going to stick.

00:04:18 Gunnar Hellekson

Or not. But we created this thing called CentOS Stream.

00:04:21 Gunnar Hellekson

Which is sits in between.

00:04:23 Gunnar Hellekson

The Fedora and the Red Hat Enterprise Linux releases and a good way of thinking about sent to a stream is that this is Red Hat taking 99% of the work that we did internally, entirely inside red hats walls, and we’re now doing.

00:04:39 Gunnar Hellekson

It in public.

00:04:40 Gunnar Hellekson

So read it sent to a stream is literally. You’re watching day to day. All the code commits of the Red Hat engineers as we’re building the next version of Red Hat Enterprise Linux. Now I like this in kind of a philosophical way because this kind of aligns us with our openness and our transparency principles. But as a practical matter, it turns out this is extremely useful for doing early enablement.

00:05:01 Gunnar Hellekson

Especially in a multi partner environment which is another thing you heard about in the keynote this.

00:05:06 Gunnar Hellekson

Is we for 25 years they have been building delicious hardware and then we have been lighting it up with the operating system and then we ship it to the market and cross our fingers and six to nine months later, somebody ends up figuring out.

00:05:19 Gunnar Hellekson

What to do with?

00:05:19 Gunnar Hellekson

It right, with the advent of this sent to a stream, this gives.

00:05:25 Gunnar Hellekson

Us an opportunity to do this.

00:05:26 Gunnar Hellekson

Enablement earlier do it with multi party do with multiparty enablement of these of these hardware features.

00:05:33 Gunnar Hellekson

So that when.

00:05:34 Gunnar Hellekson

The next version of Red Hat Enterprise Linux releases. Then we’re there with Intel. We’re there with nginx. I mentioned earlier. We’re there with guys AI, which I mentioned earlier, right?

00:05:44 Gunnar Hellekson

So this is a.

00:05:45 Gunnar Hellekson

Way of getting innovations into the ecosystem as a way of accelerating kind of the time to market or really the?

00:05:52 Gunnar Hellekson

Time to benefit?

00:05:53 Gunnar Hellekson

From a lot of these capabilities that Intel.

00:05:55 Gunnar Hellekson

‘S building does all of that.

00:05:57 Gunnar Hellekson

Make sense?

00:05:59 Gunnar Hellekson

Cool. So you have any questions about that?

00:06:03 Gunnar Hellekson

OK, so that was the, that was the fertile ground on which this new collaboration started. We suddenly realised, oh, wait, they had Intel has this interest in expanding their ecosystem and and getting things in the market earlier. We have a way of doing that. So OK, fine.


What is it?

00:06:17 Gunnar Hellekson

Specifically that we’re going to do and that’s the that was. That’s where this this collaboration.

00:06:26 Gunnar Hellekson

The idea here is to allow us to figure out earlier where these capabilities that we’re jointly building together can be applied in the market. And there’s some obvious targets.

00:06:37 Gunnar Hellekson

For this right?

00:06:38 Gunnar Hellekson

First thing and what every customer asks about is performance, right? So just baseline performance, right? If we can do all the performance tuning.

00:06:46 Gunnar Hellekson

Ahead of time with all the partners necessary.

00:06:49 Gunnar Hellekson

Then great right now.

00:06:51 Gunnar Hellekson

Everybody is kind of putting their best.

00:06:52 Gunnar Hellekson

Foot forward when we release the.

00:06:54 Gunnar Hellekson

For the second area, where we’re collaborating is on AI, right? So this is a again lighting up the GPU is one thing, making sure we have the right drivers in place. That’s one thing, but actually making sure the GPU is in a usable state when we actually ship the product production back here is not vigorously because she’s responsible for making sure this happens.

00:07:14 Gunnar Hellekson

Making sure the GPU is actually in a usable state for developers and for partners is extremely important and not work that you want to start after you ship the product, because once you ship the.

00:07:22 Gunnar Hellekson

Product it is.

00:07:23 Gunnar Hellekson

Too late. You need to do it beforehand.

00:07:25 Gunnar Hellekson

Send to a.

00:07:25 Gunnar Hellekson

Stream is where.

00:07:26 Gunnar Hellekson

We’re going to do that. The second area of collaboration is on Edge. And so here this.

00:07:33 Gunnar Hellekson

Here I would say a lot of.

00:07:34 Gunnar Hellekson

This work is.

00:07:34 Gunnar Hellekson

Around open keynote and making sure that we have the Red Hat device Edge products can work together with open vino and actually create good components that can be used in a broader AI solution on the on the device edge.

00:07:49 Gunnar Hellekson

I should mention too that all this work on AI, all this work on edge accrues not just to Red Hat Enterprise Linux, but also to all the products that rely on Red Hat Enterprise Linux, including open shift and open shift.

00:08:00 Gunnar Hellekson

Yeah, right. So as long as we’re doing it on as long as it’s all figured out on rail, then it becomes a lot easier to go figure it out on the higher order products, right? The last place where we’re collaborating is on security and this is one that’s close to my heart. I spent ten years in red hats, public sector Group and became endlessly frustrated with.

00:08:17 Chris Tobias

All of this.

00:08:18 Gunnar Hellekson

Great hardware that.

00:08:19 Gunnar Hellekson

Intel was shipping in order to enable.

00:08:20 Gunnar Hellekson

Security and nobody could figure out how to use it right? And I’m considering kind of my personal mission like this is the opportunity to take all these great new security capabilities like TDX and put them in the hands of the ecosystem, put them in the hands of developers so that we actually have something that customers can use on the first day of these products being available.

00:08:40 Gunnar Hellekson

Does all that make sense?

00:08:42 Gunnar Hellekson

OK, good. Cause I think I.

00:08:43 Gunnar Hellekson

Just stole Thunder from my.

00:08:45 Gunnar Hellekson

Last, OK, here we go.

00:08:45 Gunnar Hellekson

Yeah. Good. Because I just.

00:08:47 Gunnar Hellekson

Talked to this entire slot.

00:08:48 Gunnar Hellekson

That’s how good I am. It’s.

00:08:49 Gunnar Hellekson

All in my head, I don’t even need this one.

00:08:52 Gunnar Hellekson

So did I touch on everything? Oh, yeah. On the foundational performance, so in sent to a stream, the magical thing about sent to a stream is that it is a community that is building the next version.

00:09:01 Gunnar Hellekson

Of Red Hat Enterprise Linux.

00:09:02 Gunnar Hellekson

And you say, well, that’s fine, gunner.

00:09:03 Gunnar Hellekson

But doesn’t Red Hat completely control?

00:09:05 Gunnar Hellekson

Everything that goes into sent to a.

00:09:07 Gunnar Hellekson

Stream, the answer is yes, because it is our product and we are responsible for it when it ships, but sent to a stream has the ability to create forks of itself in the form of special interest groups, and this is where the collaboration happens. We have a hyper scaler group, Special interest group, we have a virtualization.

00:09:23 Gunnar Hellekson

Special interest group we have now created an ISA Special interest group, so folks who are interested in doing early enablement on the next version on the next ISA version, they can actually work in that form of sent to a stream and any improvements can be folded back into.

00:09:38 Gunnar Hellekson

Main line any improvements on virtualization? Any improvements on security et cetera, et cetera, you get the idea. So this is a, this is a place and what’s cool is this is the kind of collaboration we’ve always wanted to do and now not only do we get to do it, but we get to do it in public in a completely open and transparent way where everyone can watch us work and everyone has an opportunity to help.

00:10:00 Gunnar Hellekson

Which is exactly what you need if you’re trying to build out a larger ecosystem around a lot of these capabilities.

00:10:03 Speaker 7


00:10:05 Gunnar Hellekson

OK, that was a lot.

00:10:06 Gunnar Hellekson

Of words I feel like.

00:10:07 Gunnar Hellekson

I should pause there.

00:10:09 Gunnar Hellekson

Mark, do you have anything to improve, Chris, anything to improve?

00:10:11 Gunnar Hellekson

On that.


That’s great.

00:10:14 Gunnar Hellekson

Hear that? It was great.


One important question.

00:10:19 Speaker 8

From the back of the room.

00:10:30 Speaker 6

One more time. Hardware. Everything you’re doing here.

00:10:36 Speaker 8

For indoor platform you should also work on.

00:10:39 Speaker 8

Two or three other things.

00:10:41 Speaker 7

So how I’m?

00:10:42 Speaker 8

Just curious, so you have this collaboration and how does it work the other way where the core red hats going to?

00:10:48 Speaker 8

Learn on three or four other.

00:10:51 Gunnar Hellekson

This is a fantastic question.

00:10:54 Gunnar Hellekson

Because this is this, you are describing the nightmare that is my life. This is the way I said on the panel earlier that it was Chris you.


And then.

00:11:02 Gunnar Hellekson

Were on the panel, yeah.

00:11:05 Gunnar Hellekson

I said in the last 10 years.

00:11:07 Gunnar Hellekson

The hardware market has gotten the way I describe it in every conference is like the.

00:11:10 Gunnar Hellekson

Hardware market has gotten.

00:11:11 Gunnar Hellekson

Deeply weird and the there are a lot of the standards that we used to be able to rely on a lot of the norms in the hardware.

00:11:19 Gunnar Hellekson

Market that we.

00:11:19 Gunnar Hellekson

Used to be able to rely on no.

00:11:20 Gunnar Hellekson

Longer existing, that’s.

00:11:21 Gunnar Hellekson

Because the market will not tolerate that, they need the innovation.

00:11:25 Gunnar Hellekson

Right. And so we need to go the investments that everyone is.

00:11:28 Gunnar Hellekson

Making in GPU.

00:11:29 Gunnar Hellekson

The investments we’re making at PGA 6, all these accelerate all these accelerators are completely changed in the hardware land.

00:11:35 Gunnar Hellekson

And I am in the unenviable position of hiding that from all the.

00:11:39 Gunnar Hellekson

End users.

00:11:40 Gunnar Hellekson

A lot of people misunderstand the like. What makes something Red Hat Enterprise Linux compatible as being like? Well, the entire stack like all the way from the kernel on down The Dirty little secret is that we actually have 17 kernels running at any given time. A comedy is all different kinds of hardware and all different kinds.

00:11:55 Gunnar Hellekson

Of life cycles and all the rest of it.

00:11:56 Gunnar Hellekson

But we make sure that it is all compatible.

00:11:58 Gunnar Hellekson

With itself, right and so.

00:12:01 Gunnar Hellekson

No, I’m not saying.

00:12:02 Gunnar Hellekson

The job is getting easier. I’m saying the job is getting a lot harder, but we do know how to do this, which is and the way that we do it is we declare compatibility.

00:12:09 Gunnar Hellekson

At the what we call the Avi layer application.

00:12:11 Gunnar Hellekson

Binary. Yeah. And so as long as we can guarantee ABI compatibility, then that allows for portability of the application across different clouds, different hardware profiles.

00:12:21 Gunnar Hellekson

Et cetera, right.

00:12:22 Gunnar Hellekson

Yeah, it’s not easy. I don’t recommend anyone else.

00:12:25 Gunnar Hellekson

Try this, it’s very expensive. It’s very unpleasant to maintain.

00:12:30 Gunnar Hellekson

But it is important for the.

00:12:32 Gunnar Hellekson

Yeah. OK, I’m going to.

00:12:34 Gunnar Hellekson

Sit down.

00:12:35 Gunnar Hellekson

But I’m ready for the.

00:12:36 Speaker 5

Next question. All right, well, there’s.

00:12:41 Chris Tobias

No immediate questions and I forgot to introduce myself.

00:12:44 Chris Tobias

So I managed.

00:12:44 Chris Tobias

Technology leadership, so collaboration with wonderful companies like Red Hat all the way through we have.

00:12:51 Chris Tobias

A bunch of solution.

00:12:52 Chris Tobias

Next, on my team to help.

00:12:54 Chris Tobias

The boy, in a heterogeneous compute environment downstream, which is one of the reasons when you when you heard one API be.

00:13:02 Chris Tobias

Talked about earlier, that is a tool no matter what environment people are using. If we go implement that, we know they’ll get their best experience and we can start building up from there.

00:13:12 Chris Tobias

So so just to.

00:13:13 Chris Tobias

Give a little framework gunner gave a great kickoff. There’s kind of three areas of of of big importance that that we’re looking at right now in the market, artificial intelligence. It’s like locating just wake up fall out of bed, the other ones.

00:13:32 Chris Tobias

Our security, which you highlighted, so thank you, Gunner and then sustainability because.

00:13:39 Chris Tobias

Power budget is becoming just as important as the dollar budget nowadays in any sort of any sort of cloud, multi cloud, hybrid cloud nor edge to cloud Edge. However you want to deploy it, that’s just as important as the as the.

00:13:53 Chris Tobias

Dollar budget. So I don’t know if you want to give a few thoughts on those and we.

00:13:57 Chris Tobias

Can open up some more general questions.

00:14:00 Gunnar Hellekson

I guess, but I just finished talking for only.

00:14:01 Mark Skarpness

Half the.


Market well, yeah, I.

00:14:03 Mark Skarpness

Think hopefully if you saw Greg’s keynote and Pat yesterday, you know we’re we’re continuing to build new hardware capabilities. You know, Gunner was talking about, I think together the hardware capabilities are getting more complex. TDX is a good example of that. In terms of the different touch points in the stack.

00:14:20 Mark Skarpness

So I think just the ability for us to together build this in an early form, go out into the ecosystem and find you know joint customers that want to work with us on making sure that it can be deployed, it could be supported in production. I think that’s going to be a super.

00:14:35 Mark Skarpness

Valuable part of this collaboration, you know, obviously when you mentioned sustainability, you know performance and sustainability.

00:14:42 Mark Skarpness

Sort of deeply related. So I think getting the best out of the hardware that people have in terms of performance leads to better power consumption. Obviously you can do work faster so it.

00:14:52 Mark Skarpness

We’re again excited about the ability to drive faster use of the hardware that’s there. The hardware evolves faster in terms of the ice sometimes than the software ecosystem can absorb. So I think working together to increase the pace of that so that everybody can get the benefit out of the hardware. You know there’s performance left on the table because.

00:15:12 Mark Skarpness

For compatibility reasons, as Greg mentioned, it’s easier to build to the lowest common denominator which is very old. You know it’s 1010 plus years old, but if you want to get the best out of the latest hardware, you have to make that compromise. If I’m going to optimise for the latest generation, get better performance, won’t that particular binary won’t run?

00:15:29 Mark Skarpness

In an old.

00:15:31 Mark Skarpness

But you know, that’s another thing that we can do together, you know, in the ISIS say, for example, where we can go explore like, what are the workloads?

00:15:37 Mark Skarpness

Where this really benefits again with the customer saying I’m going to get a big benefit out of this workload if we do this ISO work together. So I think that’s going.

00:15:45 Mark Skarpness

To be really exciting and.

00:15:47 Mark Skarpness

Then on AI, of course, there’s tremendous innovation there in terms of hardware.

00:15:51 Mark Skarpness

Lots of new software, so the need to do that early fast integration is really evident like that world is evolving so fast.

00:16:00 Mark Skarpness

So being able to do that quickly get it out into the hands of developers, get feedback, iterate on it. You know, we can’t. We can’t work at the old sort of very long enterprise cycle for that kind of a workload because it changes so fast. So I think it’s a lot of really great potential in this collaboration from our point of view.


That’s great.

00:16:23 Speaker 8

I’m just wondering.

00:16:25 Speaker 8

How mature are your customers to kind of recognise the value that they are getting from this partnership and you know, sort of access and you know where you know what?

00:16:39 Speaker 8

Do you see? You know, within the, you know, sort of your end user organisations.

00:16:45 Speaker 8

You know, have you seen a kind of maturity and evolution and the type of roles that are coming up that now sort of are much more kind of laser focused on that kind of getting, you know sort of getting performance at a lower level than perhaps in the old days where you know the development audience might have just?

00:17:03 Speaker 8

But I don’t really care what goes.

00:17:04 Speaker 8

On beyond the.

00:17:05 Speaker 8

Thing on yourself. Do my application code.

00:17:07 Speaker 8

And make it run.

00:17:08 Speaker 8

Whereas now I get the impression both from a sustainability and obviously from an AI workflow that you know there may be a group of you know roles and if you can talk to those, you know both whether within your organisation.

00:17:23 Speaker 8

But also within your customer organisation, how far are we, you know, to kind of getting that kind of maturity or is it?

00:17:30 Speaker 8

Really, still early days.


OK. Yeah.

00:17:35 Mark Skarpness

Well, I can, yeah.

00:17:36 Gunnar Hellekson

Go ahead. No, this is a. So this is a great question. I made a joke earlier about like the in the security category, right, that there have been security capabilities in the chips and they’ve been in there for decades, right. And we haven’t actually made full.

00:17:52 Gunnar Hellekson

Use of them.

00:17:54 Gunnar Hellekson

And I think with the benefit of hindsight, I think if I could go back in a time machine and do.

00:17:58 Gunnar Hellekson

This again, I think that I think I figured.

00:18:00 Gunnar Hellekson

Out what the problem is the problem.

00:18:02 Gunnar Hellekson

Was we built?

00:18:03 Gunnar Hellekson

It and we lit it up and we didn’t tell anyone.

00:18:05 Gunnar Hellekson

How to?

00:18:05 Gunnar Hellekson

Use it.

00:18:07 Gunnar Hellekson

Right. And and that’s and that’s on both of us like that’s a that’s a fault of the partnership, right. And one of the things I heard in all the keynotes and all the kind of keynote highlights today and one thing that I’ve noticed in the collaboration we have with Intel now is that we are all now focused much more on, I’m going to call them 3 way partnerships or three.

00:18:27 Gunnar Hellekson

Collaborations. Whereas before we kind of looked at it as a as a two way colour.

00:18:30 Gunnar Hellekson

Duration right. And to the extent that we can invest in these three way collaborations with an ISV with a customer with a partner and figure out like, OK, well they have this great capability, we figured out how to light it.

00:18:41 Gunnar Hellekson

Up efficiently and.

00:18:42 Gunnar Hellekson

Make it consumable and So what are you going to do with it? And then they say, well, I could do this, but I need the button to be blue instead of purple and make OK, great. We’re going to make the button blue instead of purple.

00:18:50 Gunnar Hellekson

And now you can. Now you can make use.

00:18:53 Gunnar Hellekson

So I think that the it’s less about, it’s true that AI is creating some like very specific needs. Security is creating some very specific needs. Certainly sustainability is creating specific needs. But I think that the problem that we are solving here has actually been in general and it has been endemic, you know for it has been, this has been going on for a long time, good news is.

00:19:11 Gunnar Hellekson

The answer is the same.

00:19:12 Gunnar Hellekson

Is partnering with the right set of folks in our ecosystem and both of us leaning on the.

00:19:17 Gunnar Hellekson

Ecosystem that we.

00:19:18 Gunnar Hellekson

Share in order to help a customer make the best use of the of the work that.

00:19:21 Gunnar Hellekson

We’ve already done right? Does that make sense?

00:19:23 Speaker 9

Yeah, that just makes sense. I just wonder.

00:19:26 Speaker 8

I mean, you know, I I guess you’re saying that that’s great. I mean, how much will you then go? Because I know there was. Yeah, that, you know, all those talk about you know.

00:19:35 Speaker 8

Sort of. The training and having, you know it’s incredibly.

00:19:38 Speaker 8

You know good.

00:19:39 Speaker 8

The certification, but whenever we speak to Angie’s organisation part, the fact is and and I think you’re right.

00:19:45 Speaker 8

What lighting up and making it easy for you know, sort of them to really understand that a lot of them are, you know, sort of looking to getting you know.

00:19:54 Speaker 8

From opinionated you know strategies they they don’t know which roles that they should, you know? Yeah. Have in place. A lot of them probably have some of those roles already in place but they think they are you know deployed maybe not deployed in the right place. And so I wonder how much of this could be seen as you you know?

00:20:14 Speaker 8

Of raising and putting spotlight on that, but also putting the.

00:20:18 Speaker 8

Training, you know, sort of.

00:20:20 Speaker 8

You know, sort of the framework in.

00:20:22 Speaker 8

Place whether whether you do it.

00:20:23 Speaker 8

Yourself or you do it for the third party.


Yeah, yeah, yeah, I’ll, I’ll.

00:20:28 Chris Tobias

Take a take a stab at this one cause I spent a lot of time in my.

00:20:32 Chris Tobias

Role downstream and this is where the keyword partnerships and this the technology partnerships upstream on developing the technology.

00:20:40 Chris Tobias

Making sure right.

00:20:42 Chris Tobias

Goes well and historically every company kind of had their target segment, but today with Edge to cloud, you have smart Edge.

00:20:51 Chris Tobias

I mean everything.

00:20:52 Chris Tobias

‘S intertwined, so you have to think not only about, hey, what is? What is my hardware build out look like from servers? It’s going to be what’s my networking, what’s my switching?

00:21:02 Chris Tobias

OK, when should I move? You know, when should I move the data off the sensors at?

00:21:07 Chris Tobias

The edge to.

00:21:08 Chris Tobias

A local server.

00:21:09 Chris Tobias

Sure. OK. It’s sitting there. How much compute do I need? Should I pay the latency penalty, you know? And now people are starting to consider the power penalty of how much energy it costs to move that back to the data centre. The the dollar costs just the economic factors. And then there’s a technology factors of is it secure when I move it back and forth between between those locations.

00:21:31 Chris Tobias

And guess what? This is a huge opportunity for all the partners, the system integrators, because they people here like, hey, I need to do generative AI for example.

00:21:42 Chris Tobias

So I need to go spend a bunch of money on on systems that are going to take me.

00:21:46 Chris Tobias

A year to get because that’s what I’m hearing. I need. They don’t quite know what the problem is they’re trying to solve. They just know they’re at, you know, hey, everybody else is doing this and be at risk. What’s your recommended architecture I could use and we start to ask them questions like, OK.

00:22:02 Chris Tobias

Well, what problem you’re trying to?

00:22:04 Chris Tobias

12 I don’t know, I just need generative AI. What are you looking at? A chat bot to help you know, prevent calls. Go to your actual live customer service Rep, are you training it on your own internal data? How many parameters you think that is? There’s other ways to solve that problem with the existing technology and that’s where we’re really partnering up with system integrators.

00:22:25 Chris Tobias

We actually partner up with red hats doing.

00:22:27 Chris Tobias

The same thing.

00:22:29 Chris Tobias

So we have our own direct sales channel. We also have our partnership channel.

00:22:33 Chris Tobias

And recommended architectures just to get started on the basic problems is hey, solve this problem first. Everybody needs to get a little bit understand.

00:22:43 Chris Tobias

Thing, because with all these changes and now we got multiple nodes, technology is changing faster than I’ve ever seen it since I’ve been in this industry and it’s ridiculously fast. Nobody can keep up. You know, the person on my team who leads AI. She was an expert in deep learning for 15 years on computer vision. She’s like Chris.

00:23:03 Chris Tobias

Nobody says they’re AI expert is a liar because it’s being made up right now. So.

00:23:10 Chris Tobias

People were practising it. They’re the experts, so it part of it is the learning and then how do we get the the how do we work with partners to get something that’s consumable to the end user and using the latest and greatest technology and it’s it’s actually a very, very big challenge, but it’s a huge business opportunity.

00:23:30 Chris Tobias

Solve people’s pain points, so it’s a great question, yeah.

00:23:34 Chris Tobias

All right. Other questions we were talking about AI.

00:23:38 Speaker 5

Yeah. So maybe this isn’t the most appropriate question, but you know you’re talking about Intel talking about the the developer cloud.

00:23:47 Speaker 5

And how it’s going to put everything?

00:23:48 Speaker 5

Some people up front and and be able to.

00:23:51 Speaker 5

To get things developed quickly and like how does this relationship fit into that structure?

00:23:57 Mark Skarpness

Yeah, I mean you can imagine, take TDX for example. We work with the joint customer and and they want to try it out on some.

00:24:03 Mark Skarpness

Early systems.

00:24:05 Mark Skarpness

So they can get access to Emerald Rapids, which is our next Zeon, which has TDX in it and they can go try that workload out in the cloud, you know, and get familiar with it. We can bring the Red Hat. You know, the CentOS Stream build in there. So it gives.

00:24:18 Mark Skarpness

Gives us a place to do that sort of. Try it out live without having to ship them a box. You know, it’s a much faster.

00:24:25 Mark Skarpness

Easier way to get that hardware in the hands of developers users. Same for AI with our latest you know accelerators, we’ve got those in the developer cloud, so it’s.

00:24:35 Mark Skarpness

I think it accelerates that whole cycle of.

00:24:38 Mark Skarpness

I’ve looked at something I’m interested in and I want to try it out. OK, here you go. You can get access to a system or set of systems. Go try out your use case. I think that will be a critical part of speeding up the whole cycle. Having access to that.

00:24:51 Speaker 5

Hardware and the reason I ask it is because I mean Intel has so many partners in so many different areas.

00:24:57 Speaker 5

The developer club is terrific, but how all encompassing is it? If I really want to try something out, you know, without having to go buy, have to go, go to buy licence from Red Hat. But you know half a dozen.

00:25:08 Speaker 5

Other people as well.

00:25:09 Mark Skarpness

Yeah, yeah.

00:25:10 Speaker 5

How how accelerating is it really? If I can’t get everything?

00:25:13 Mark Skarpness

I need but you can sort of bring your I think in the early development stage you you bring that with you you know we give you access.

00:25:19 Mark Skarpness

To a system.

00:25:19 Mark Skarpness

You can provision it with bring your own, you know early stack and there are things that are pre configured. But I think for something like you know a CentOS stream build would probably be something we would.

00:25:20 Speaker 5

Bring your own.

00:25:28 Mark Skarpness

Going in, say we’re going to bring in this pre-release software and we’re going to use the dev cloud as a means to test it out and try the use cases. So it kind of spans the range of I want to use some already release stuff to. I’m using very cutting edge pre-release kind of early things.

00:25:45 Gunnar Hellekson

And this is the value sent to a.

00:25:46 Gunnar Hellekson

Stream, by the way.

00:25:47 Gunnar Hellekson

Somebody could come in and do it.

00:25:49 Gunnar Hellekson

We’re doing early enabling on something for something real, 10 for even REL 11, right, which is just feels like just around the corner from my point of view. Yeah, we could actually bring in a Fedora release that we think is going to be close to round 10 and let them try it out and maybe it works and maybe it doesn’t, but at least we’re finding out several years ahead of the actual.

00:25:55 Speaker 4


00:26:07 Gunnar Hellekson

Of of them going into production, right? So in the same way that the developer Cloud provides early access to the hardware, we have software communities that provide early access to the next version of Raul and it’s a nice, you know, it’s a.

00:26:18 Gunnar Hellekson

Nice complementary.

00:26:19 Speaker 5

Thank you getting feedback as.

00:26:19 Speaker 5

Well, that’s right. Exactly right.

00:26:20 Mark Skarpness

Yeah, yeah. What are we missing? What have we thought about? Yeah, that’s.


The good news is.

00:26:28 Mark Skarpness

Blame the heart.

00:26:30 Chris Tobias

Any question in the back?


Yes, Sir.

00:26:32 Speaker 6

Yeah, noting of course that this.

00:26:35 Speaker 6

Question is more completely out of it because I don’t know.

00:26:38 Speaker 6

Anything about it?

00:26:39 Speaker 6

But the fact that it takes years to get hardware features and they maybe software scares the hell out of me as I think about Cent OS on the security guy.

00:26:50 Speaker 6

Something. Well, how?

00:26:51 Speaker 6

Do you do patching what happens in?

00:26:53 Speaker 6

A. If you can’t light up.

00:26:55 Speaker 6

Hardware features in three or four years.

00:26:57 Speaker 6

Yeah. What happens when the log 4 chain happens?

00:26:58 Gunnar Hellekson

Oh, yeah, yeah. So first of all, I owe you 20 bucks at the end of this.

00:27:02 Gunnar Hellekson

Because I have, I have.

00:27:03 Gunnar Hellekson

Already built answer for you so so.

00:27:08 Gunnar Hellekson

The expectations are going to change if you’re working in an upstream project that is going to move extremely quickly. A lot for Jay shows up, they’re going to patch it within hours or probably, you know, at worst a couple of days, right? Because that’s how upstream communities work. That’s the good news. The bad news is that they don’t care whether they broke anything in fixing it.

00:27:28 Gunnar Hellekson

Right. And so in Fedora sent to a stream. What they’re doing is they’re pulling down from the upstream communities, right. And if there’s a lock for Jay, if there’s some security vulnerability, they will probably pull that in very quickly, right. They, in fact, they will probably pull that in as soon as possible.

00:27:45 Gunnar Hellekson

When you move into a.

00:27:46 Gunnar Hellekson

Product that is being.

00:27:47 Gunnar Hellekson

Supported for a long period of time like Right enterprise Linux, right? I’ve got a.

00:27:50 Gunnar Hellekson

12 year life.

00:27:51 Gunnar Hellekson

Cycle right and one of the reasons why Red Hat has thousands of engineers is because, yes, we’re innovating upstream, but we are also taking those fixes and then backporting them to the software that we have in a way that won’t disrupt.

00:28:05 Gunnar Hellekson

The workload that is already running and so it’s a it’s a different risk profile, right? Like I’m going to favour stability like I yes, I want to go patch the vulnerability, but I don’t want my database to go under. Right. And so that’s the kind of that is the kind of security.

00:28:18 Gunnar Hellekson

Response. You’re going to get from a Red Hat Enterprise Linux subscription.

00:28:21 Gunnar Hellekson

But if you’re working in the upstream, if you want it extremely quickly, you’re pulling directly from the upstream. If you want it a little bit more curated, you’re going to get it from Fedora. If you want to make sure that if you want to have good hygiene on a system that you’re testing with a partner, then you’re going to be using sent to a stream. So it’s just the responses change.

00:28:38 Gunnar Hellekson

Depending which community you’re working in, yeah.

00:28:43 Chris Tobias

Are we been neglecting this side of the room recently? Any any questions over here?

00:28:49 Chris Tobias

Yes, in the back.

00:28:51 Speaker 4

I just wondered about one API.

00:28:53 Speaker 5

Whether you.

00:28:56 Speaker 4

Consider that an Intel initiative.

00:28:59 Speaker 4

Or whether you consider.

00:29:00 Speaker 4

What open standard across the industry because you have parts of, you know, video as well. Just wonder how size things and whether you really think.

00:29:09 Speaker 4

One API, something that will take.

00:29:11 Speaker 4

Off and you’ll have a more open interface for most of the accelerators and.

00:29:19 Mark Skarpness

Well, you you probably heard this morning, you know, the creation of the new foundation under the Linux Foundation to really bring it to be, you know, a true industry effort. So I think that’s a big step in that direction of, you know, really moving it out under neutral governance. We’re going to put all of our sort of open source.

00:29:37 Mark Skarpness

Project assets under that same governance and you saw, you know, pretty wide breadth of industry partners stepping in to participate there.

00:29:46 Mark Skarpness

So I certainly our intention is that it become an industry standard that is an open place for everybody to build sort of this interoperable layer that can span different hardware from different vendors. And I think there’s a really strong desire. We hear it from our customers every day to get out of a proprietary gaol they’re in. It’s a very painful place to be.

00:30:07 Mark Skarpness

So I think that that’s a step in that.

00:30:09 Mark Skarpness

Direction from our point.


Yeah, yeah.

00:30:11 Gunnar Hellekson

Yeah, and I have.

00:30:12 Gunnar Hellekson

To get I have to give Intel credit for the for. For doing that right now, because moving that into the foundation is exactly what you do when you want something to be an industry standard, right? I mean, they’ve one API wasn’t invented yesterday, it’s been around for a while and the act of putting in the foundation was a great demonstration of their commitment to actually making it an industry wide.

00:30:32 Chris Tobias

Yeah, and and I’ll.

00:30:33 Chris Tobias

I’ll just add like because my my team works downstream deploying with customers and one of the costs they face depending on is you know they have to hire elite developers. If you’re going with a very narrow industry solution and we’ll show them performance on one API.

00:30:52 Chris Tobias

Almost every case it’s as good as or better than what they’re using, and then they could free up those developers to to work on higher value problems than just, you know, getting the getting the basic hardware running. So it’s very positively viewed in the industry and is viewed as a a support for heterogeneous compute. And I think this is a very logical next step.



00:31:12 Chris Tobias

To get complete industry wide participation. So I’m I’m personally very excited.

00:31:17 Chris Tobias

About it, it’s.

00:31:21 Chris Tobias

Other questions from this side of the room.

00:31:23 Speaker 9

Yes, possibly with surprising us yesterday with a 288 core is here on for.

00:31:29 Speaker 9

Example, yeah, one.

00:31:30 Speaker 9

Of the slowest moving things in the industry is the commercial model and licencing, et cetera. How do you think you can encourage your partnership?

00:31:38 Speaker 9

With ISV’s and.

00:31:39 Speaker 9

Others who are all trying to build on this to move with the times and have a simpler.

00:31:42 Speaker 9

Way to engage rather than.

00:31:44 Speaker 9

Having a licencing model built for the 90s.

00:31:48 Mark Skarpness

Oh, per core licencing model you mean? Is that what?

00:31:50 Mark Skarpness

You’re talking about.

00:31:50 Speaker 9

Or just things like that, but there’s.

00:31:52 Speaker 9

A lot of.

00:31:52 Speaker 9

Licencing still built around consumption.

00:31:54 Mark Skarpness

That’s true. Yes. Yeah, yeah, it’s something we’re certainly talking with our ecosystem partners about in terms of how to think about as I mean, it’s not just on our platform, everyone’s platform core count is going up dramatically, so.

00:32:06 Mark Skarpness

They have to sort of move with the times. I think in terms of what’s the value they’re delivering and is it aligned with, you know, customer expectations.

00:32:14 Mark Skarpness

In terms of, you can’t have a backwards looking per core licencing model and apply that to a dramatically higher core count platform. I mean those economics just they don’t work. So I think the market will fix itself for the most part there my, my view.

00:32:30 Mark Skarpness

We’ll see.

00:32:34 Chris Tobias

I’ll just Jake, so since I work with many, many software companies, the good news is everybody has to have a cloud, you know.

00:32:44 Chris Tobias

Consumption model pricing.

00:32:46 Chris Tobias

And because they’re evolving, everyone’s evolving their pricing and how they price it means they can look at, OK, the beginning to end of how to do this and what everyone’s realising is the market is changing super fast and they’re trying to build in some, some boundaries to look at. OK, I got to be in a multi cloud hybrid cloud.

00:33:07 Chris Tobias

Edge to cloud environment cores are going like crazy. You know, some people look at memory or whatever and everyone’s reevaluating how they price.

00:33:19 Chris Tobias

And it’s got all to be up their own software companies decision on.


How they do that?

00:33:24 Chris Tobias

But they’re looking at to how could I provide the engineering service?

00:33:30 Chris Tobias

Hey, the old the old.

00:33:32 Chris Tobias

20 year old models like I’ll sell you this licence around this software drop which supports this hardware.

00:33:40 Chris Tobias

It’s all kind of lying, but now it’s like I just just, I give you some money, take make sure.

00:33:45 Chris Tobias

I don’t have.

00:33:45 Chris Tobias

Problems and so how that works is everybody’s looking at. So I I feel like it’s a solvable problem and people once you once you open up that process to fix, you can come up with solutions.

00:33:59 Chris Tobias

Don’t know if you have anything to add.

00:34:00 Gunnar Hellekson

I have nothing to add. They won’t incriminate me.

00:34:04 Gunnar Hellekson

It is a. It is an extremely hard problem and I will say, and there is nowhere to hide like I haven’t seen a given unit of measure like there is no, as you say, the market will work it out. It is absolutely not worked it out yet, right? I’ve seen experiments with core counts and memory and like I do socket pairs.

00:34:20 Gunnar Hellekson

But yeah, there’s nowhere to hide it.

00:34:23 Gunnar Hellekson

All it’s all going to change.

00:34:25 Gunnar Hellekson

Right. Because it’s obviously not sustainable for all the.

00:34:27 Gunnar Hellekson

Reasons you just talked about, right? Yeah, yeah.

00:34:30 Speaker 8

I’m just going.

00:34:30 Speaker 8

To add a little something because it’s on the screen as well about open shift when.

00:34:34 Speaker 8

You think about Red Hat. Think about real when.


You think about open?

00:34:38 Speaker 8

Shift just any.

00:34:39 Speaker 4

Enhancements there for how those guys work together.

00:34:42 Gunnar Hellekson

Yeah. Thanks for asking. So I’m the I’m the real guy, so of course I’m going to tell the real part of the story. So thanks for correcting me, but the.

00:34:52 Gunnar Hellekson

All of the.

00:34:53 Gunnar Hellekson

All of the benefits that I just that.

00:34:55 Gunnar Hellekson

I just described immediately.

00:34:56 Gunnar Hellekson

Accrue to open shift as well, right? So as I light up as I light up gaudy, right, that becomes now available to open shift as a like as a as a computing resource, right? This is. This is kind of how we’ve constructed the, the the Red Hat.

00:35:12 Gunnar Hellekson

Environment. I think that there is.

00:35:14 Gunnar Hellekson

This we talked about sustainability earlier. We do have a project called Kepler which I.

00:35:18 Gunnar Hellekson

Think we’re working together on right. OK, good, good.

00:35:19 Mark Skarpness

We are, yeah.

00:35:23 Speaker 5

Because that would be awkward. We’re working.


On it.

00:35:26 Gunnar Hellekson

But so we have this this project called Kepler, which is not meant to solve the sustainability problem, but at least provide visibility into which is actually the first problem and the sustainability question is.

00:35:35 Gunnar Hellekson

People have no idea what even what instrumentation they should be using, right? Let alone like where to set targets and how to manage them. Right? And so Kepler is is specific.

00:35:44 Gunnar Hellekson

For open shift, of course, there’s benefits to Red Hat Enterprise Linux as well, and that provides more visibility into the open shift. Kubernetes environment helps people do power management across their Kubernetes clusters, right?

00:36:00 Speaker 11

How’s the feedback open window, the Red Hat from about the.

00:36:10 Speaker 11

The feedback from users.

00:36:15 Gunnar Hellekson

Well, so the so I mentioned my good friends at guys AI who are actually in our booth right on the floor. You can ask them. Enjoy it. It’s right there in their architecture diagram. The feedback has been extremely good. And like I said, we’ve seen.

00:36:31 Gunnar Hellekson

I we talked about this earlier today, I think we hear AI and there’s all this talk about, oh, the AI and the GPUs and the.

00:36:37 Gunnar Hellekson

Large language models.

00:36:38 Gunnar Hellekson

And all.

00:36:38 Gunnar Hellekson

The rest of it for me as the as the REL guy the the action is like, yeah, that’s all interesting. And that’s all super important. But the real action is actually on the inferencing side, right? That’s where a lot of the really interesting applications are happening.

00:36:52 Gunnar Hellekson

That’s where, that’s where the action is.

00:36:54 Gunnar Hellekson

And so we’re really excited about being able to. That’s one of the reasons why I said I was excited about open vino is because it makes the destructive workloads practical for a lot of folks.

00:37:03 Gunnar Hellekson

In a way that.

00:37:04 Mark Skarpness

They weren’t before, right?

00:37:06 Chris Tobias

Yeah, I think I think it.

00:37:07 Chris Tobias

Also highlights, you know, open veto kind of start at the edge, but it’s also not being applied in the data centre.

00:37:14 Chris Tobias

But one of the things that tends to get ignored in our work together with with Red Hat and and Gutter touched on earlier.

00:37:23 Chris Tobias

Was hey, at the Edge, people would had this direct software stack so you could have billions of dollars of hardware. That is, you have to replace because some control boards and software run it were written in the 80s or even worse the 70s or whatever. Everybody’s retired. Nobody knows how to run it the three.

00:37:43 Chris Tobias

Consultants left. You know I. Yeah, I’ll do that for $10 million or whatever, but eventually they’re all gone. So going to, you know, containerized infrastructure at the edge.

00:37:55 Chris Tobias

And then using some of these tools helps helps facilitate that problem. So open Vino has actually kind of helped accelerate that kind of architecture and it and it works where we’re all trying to go. So it’s very, it’s very kind of synergistic.

00:38:11 Gunnar Hellekson

Sorry, I want to improve on that I want.

00:38:13 Gunnar Hellekson

To connect it back to the open shift.

00:38:14 Gunnar Hellekson

Question one thing I’ve absolutely seen from.

00:38:18 Gunnar Hellekson

Let’s say internally, I think we say I would say the majority of customers now are I talked about that weird how weird the.

00:38:24 Gunnar Hellekson

Hardware environment is.

00:38:26 Gunnar Hellekson

One of the ways that customers are managing this is they’re using containers to insulate themselves from any of the hardware questions, right? Of course they want to descend from the container and down and go take advantage of other capabilities they want. But to the extent.

00:38:38 Gunnar Hellekson

Possible they are abstracting themselves away from their infrastructure choices for exactly those reasons, right? They’re trying to create some. They’re trying to draw a line and say, OK, I’m going to standardise everything underneath here and then I’m going to innovate everything on top and the dividing line for most customers is about at the container boundary, right? That’s where it’s.

00:38:57 Speaker 9

So in conjunction with that, is there any role for?

00:38:59 Speaker 9

Some sort of intermediary?

00:39:01 Speaker 9

Language like Microsoft has with its common language runtime.

00:39:08 Gunnar Hellekson

I don’t know, Mark, we think.

00:39:09 Mark Skarpness

Well, I think there’s there’s lots of like in the AI space. You’ve probably seen Mir and you know there’s tonnes of new language framework things are going on. I don’t think there will be one. You know, there are many.

00:39:21 Mark Skarpness

And there will be more.

00:39:23 Mark Skarpness

And I think the containerization is one way to.

00:39:25 Mark Skarpness

Help manage like.

00:39:26 Mark Skarpness

I’m going to curate a specific set of things for the moment that rapid because it’s changing so fast and I can run that container on top of a much more stable realm and I can just iterate inside the container that I’m sort of managing myself. So I think that’s we’re.

00:39:40 Mark Skarpness

Seeing that out for sure.

00:39:42 Speaker 10

I think we have one more, one more.

00:39:44 Speaker 10

I have.

00:39:46 Speaker 10

Question about. I heard a lot about how this the developer side you need to provide.

00:39:54 Speaker 10

A lot of new tools for for them because of the AI, I think and also how to make clear for them that what you have and on the other side is for you guys it’s I think this business, this industry is already very fragmented. So after the the AI thing that.

00:40:14 Speaker 10

It it I feel that it becomes becomes more and more fragmental. So is this something that will bring some impact on the business model that?

00:40:24 Speaker 10

That you shared.

00:40:26 Gunnar Hellekson

That’s a huge question.


That’s a big one.

00:40:30 Gunnar Hellekson

Like I said before, I.

00:40:31 Gunnar Hellekson

Think there is fragmentation in the market right now, but that is because the market is innovating in a way that it hasn’t in a long time.

00:40:38 Gunnar Hellekson

You mentioned this.

00:40:39 Gunnar Hellekson

Earlier, Chris.

00:40:39 Gunnar Hellekson

I mean things.

00:40:40 Gunnar Hellekson

Are things are moving faster than we’ve ever seen before, so naturally it’s.

00:40:44 Gunnar Hellekson

Going to be fragmented, right?

00:40:46 Gunnar Hellekson

As an operating system guy with an ecosystem of hardware and ISP’s to take care.

00:40:51 Gunnar Hellekson

Of this is my. This is kind of my central challenge. Right is finding a way to make everything stable, consumable and Cohen.

00:41:00 Gunnar Hellekson

Rent so that people don’t get lost in the chaos of the hardware, innovation and chaos of the software innovation. The rest of they need more than any other time in the past. They need a stable platform that they can rely on and let you know. And this is one of.

00:41:14 Gunnar Hellekson

The reasons why in the.

00:41:16 Gunnar Hellekson

Red Hat Enterprise Linux 8.

00:41:18 Gunnar Hellekson

Why we announced a stable?

00:41:20 Gunnar Hellekson

Release cycle. So we’re releasing a new version of Rell every three years. Whether we need it or not. And and the reason why we’re doing that is because we need a predictable cadence for innovation, so that people can plan their data set. And now I’ve I’ve talked to, you know, I talked to large customers who are now have.

00:41:38 Gunnar Hellekson

Because we have made that decision, they can now actually make plans despite the innovation, despite the chaos, they can plan their investments out 1015 years in the future and they find that extremely comforting, right?

00:41:51 Speaker 10

So you need to set a.

00:41:51 Speaker 10

Time hold, I’m.

00:41:52 Gunnar Hellekson

Sorry, you need to set the.

00:41:54 Gunnar Hellekson

Time hold. Yeah, that’s right. Yeah.


From your.

00:41:56 Gunnar Hellekson

Yeah, that’s right.

00:41:56 Gunnar Hellekson

And that allows the innovation to happen on top without sacrificing the stability.

00:42:01 Speaker 5

Anybody is setting a plan 10 or 15?

00:42:03 Speaker 5

Years in advance is not.

00:42:07 Gunnar Hellekson

Tell that to my life.

00:42:08 Gunnar Hellekson

Cycle product manager.

00:42:10 Speaker 4

You may not know.

00:42:10 Mark Skarpness

What’s going to be in the?

00:42:11 Mark Skarpness

Release, but there will be.

00:42:12 Gunnar Hellekson

I’ve got spreadsheets that would make your eyes bleed.

00:42:17 Speaker 6

Well, I think that’s.


All we have time.


For maybe we have time for one more.


Let me get a good one.

00:42:23 Speaker 7

Do you do we do we think?

00:42:25 Speaker 7

That the key level of level of interoperability.

00:42:29 Speaker 7

For AI AI.

00:42:31 Speaker 7

And ML is more likely to be like the language level at the high level framework level or like this translation tool construct Compiler SDK.

00:42:40 Gunnar Hellekson

Right. What level? So put another way. You’re asking what layer is AI going to standardise on? Is it going to be the run time layer?

00:42:48 Gunnar Hellekson

At the layers layer at the.

00:42:51 Mark Skarpness

All of them.

00:42:52 Gunnar Hellekson

It’s going to be and I think.

00:42:55 Gunnar Hellekson

For the foreseeable future, I think it’s going to be different answers to different problems at different levels in the stack, and this is one of the reasons why the world is so chaotic and fragmented right now is because I don’t think anybody knows what how this is going.


All right.



00:43:09 Chris Tobias

To resolve. Yeah, I totally agree with that.

Leave a Reply