To embed our video on your website copy and paste the code below:
<iframe src="https://www.youtube.com/embed/f2q8Hv_sOKA?modestbranding=1&rel=0" width="970" height="546" frameborder="0" scrolling="auto" allowfullscreen></iframe>
So we're in Dublin for fyuz 24. I'm here with Udayan Mukherjee, who is Senior Fellow Network and Edge Group at Intel. And Matt Conrod who is Global vRAN Business Development Manager at Intel. So thanks so much for joining us today. We're going to be talking about Open ran. Now Udayan, I'm going to start with you because major shifts in technology and business such as the one from traditional RAN to open, ran, that's a major undertaking, a major transformation, a really big journey. Where is the industry on that journey and how is the industry performing?
Udayan Mukherjee, Intel Corporation (00:43):
Yeah, it's actually a journey. As you said, we first implemented a v, an open ran version back in 2018, started LTE in Japan and then moved to 5G and others. So every incremental step of the way, we are improving the technologies, the platform, silicon software and the platform and then the overall TCO aspects of it. Where is going right now? You see a lot of announcement today and before as well. Every major vendor, every operator has an open and plan. Now every one of these elements that is getting there, the technologies are getting perfected and remember where it came from, it was a black box. You can't just switch it on. It's a black box with one vendor controlling all the stack. Now you have du, your CU, you have cas and pass multiple different vendors supporting to this. You have whole ecosystems going out. It took some time, but the journey has already started and there is no stopping back.
Ray Le Maistre, TelecomTV (01:43):
Absolutely. And Matt, from your perspective, how do you see things developing? Do you get a sense of acceleration in the open ran space?
Matt Conrod, Intel Corporation (01:51):
Well, I would say the one thing that has happened is there's a shared vision in the industry. I think it's undeniable. I talked to many operators around the world. Open and virtualized networks is the future. Some operators like Verizon and Vodafone are early adopters. They have the technology deployed and for those guys that have those operators that haven't scaled yet, there are better and better practices and better and better solutions emerging. So the journey for those that haven't done it yet is improving with every deployment with every year. But it is a journey. But I would say that the vision is aligned also. We've demonstrated in cities like New York that virtualized grand implementation with an open radio interface can compete with the best traditional baseband in the world in terms of performance. So now it's just a matter of time of the global operator community doing the scale implementation and every operator has their own timeline for introducing technology and technology risks. But we do have some early adoption and it's very exciting to see these add up good early starts and we need more to happen.
Ray Le Maistre, TelecomTV (03:05):
Okay. Now you mentioned TCO there and that is obviously one of the
Udayan Mukherjee, Intel Corporation (03:12):
Absolutely
Ray Le Maistre, TelecomTV (03:12):
Big talking points the whole time and there are still questions being asked around TCO related to open ran and with reports saying that operators are questioning the value of Open ran. Is that a fair accusation? A fair question, Matt?
Matt Conrod, Intel Corporation (03:31):
Yeah, I think it is a fair question. We've come a long way on TCO. Firstly, every generation of Intel, silicon and implementation of vRAN has improved with respect to TCO. We started with an external fixed function accelerator and a PCIE card. Then we integrated it on dye with our software, rapid key CPU and the GNR timeframe. We're integrating NIC silicon. All these steps have greatly improved the TCO and we're going to continue so it gets better over time. I think one of the challenges that I'm seeing right now on Global Open RAN RFPs is our ability or the traditional telco procurement team's ability to quantify the benefits of the technology, the ability in a hardware software disaggregated platform, it is easier or should be easier to automate and it's difficult to quantify that benefit in these TCO calculators. Likewise integrating another function other than VDU on the same server as the video, like a virtual routing function as an example, that's another more advanced concept. It's future. It's a little bit more challenging to quantify upgrade ability to AI and six G or other examples. But I would say that we've gotten a lot better. The TCO challenge and question will continue to come up and if I had any advice for the telecom community is challenge your team to quantify the benefits, the future benefits of the technology and you'll start seeing these TCOs converge between traditional RAN and open ran
Ray Le Maistre, TelecomTV (05:08):
And Udayan, how do you see this developing from a technology perspective? Is the TCO constantly improving?
Udayan Mukherjee, Intel Corporation (05:14):
Right. Any workload when you have a software defined workload is virtualized in the platform, it opens up the opportunities for a lot more things than like for, give an example, used to have those cell sites, those virtual routers used to be a standalone equipments along with a du. Now if it's a software defined aspects of it, the routing function you can integrate in the platform itself. A lot of companies are started looking very seriously. So there you go, another 5,000, $10,000 worth of box you can just put in because it's a pure software. In future look into this, we talk about a lot of AI aspects of it. The moment you have a software defined aspects of this you can add. So MIMO channel, somebody can create a much better function. You can slide that in into these functions and operators are going to demand it. So it opens up a whole bunch of novation that actually translates to better TCO in longer term. But the shorter term, I mean the capital outlay is a key element of it and something that Matt talked about that we are constantly innovating. We are adding more cores in the same power footprint. So adding more cores in means more number of sales I can actually accommodate into that. So single server is important. So I think that's the way we are contributing to this.
Ray Le Maistre, TelecomTV (06:30):
Okay. Now Matt mentioned AI there before and of course there isn't a conversation to be had in this industry right now or pretty much any industry without talking about ai and of course AI is going to play a really important role in open RAN and across telecom in general. What are your thoughts about AI in the ran? What are kind of impact does the implementation of AI applications in the radio access network going to have?
Udayan Mukherjee, Intel Corporation (06:57):
Yeah, the gen AI part of is recent, but the AI is not recent and machine learning aspects are there for a long time and we are taking a holistic view. We are taking from what is it going to do in platform, whether it's in DU and CU software and then overall what you can do in the operation side of it. So take an example on the platform. Intel Silicon has lot of the hooks in the power management C statistically sleep state, P state where you can actually do frequency changes and safe power. What if this tracks the actual workload, your D workload? If you look at TT, I mean there's a lot of times there is not much is happening. Why am I burning power in this? So AI is a perfect applications. This thing is dynamically tracking the workload and actually reducing the power as we can.
(07:50):
And then you see that right now this is the number one thing that every operator is looking into. So we already implemented that. We showed that in MWC earlier this year that if you go up in the DU rank in the digital unit, the layer one there, I talk to a buy more channel. A lot of innovations are actually being looked into. How do you do beam management 3G PP talks or beam management and couple of other use cases. We are implementing those now You have a software, you can virtualize platform, you can do a whole bunch of stuff. So your companies like IRAs or the world are innovating startup companies looking. Then you have this platform, what you do put a gene AI on top of it to actually do a lot of these diagnostics in future, like for example, when you do the operations set, some people call it copilot for operations. So this is where gen AI comes into play. So every layer the stack, we are looking seriously into what AI can. But the one of the most important things is that why am I using the AI is it either has to improve the KPI spectrum utilization or improving the uplink or downlink or you simplify it just for the sake of ai. So people are asking questions that, okay, is it going to increase my power - one? Yes. What benefit I'm going to get out of it? So we need to show those assets.
Ray Le Maistre, TelecomTV (09:08):
Sure.
Udayan Mukherjee, Intel Corporation (09:09):
So that journey has started.
Ray Le Maistre, TelecomTV (09:11):
Matt, from a business perspective, our customers getting excited about how AI might further improve the KPIs.
Matt Conrod, Intel Corporation (09:20):
Well absolutely. I think in the industry is a little bit caught up in marketing of ai. There is a lot of noise around the possibilities of gen AI and ML AI for RAN and telco. But if we park the marketing for a second, I think if we can improve spectral efficiency or if we can reduce power consumption, even maybe look at capacity optimization, a certain amount of capacity on fewer cores, if we can touch those TCO levers with the AI assist, then we should pay attention. Not only should we pay attention, we should invest resources on innovating piloting. I think the underlying, we must keep, we must make sure we don't forget the TCOV ran, especially in a distributed ran location, has a certain power envelope in a certain space and a certain amount of cooling needs as well as a certain cost structure. And if we bring AI at such an expense that it blows up that environment where you deploy a DR RAN server, then it may not be worth it. So innovation in this area should be done with TCO in mine day one. The last thing we need to do is have a full forklift change out of all the RAN with limited improvement in whether it's spectral efficiency or power saving. So let's innovate, but let's do it with TCO in mind.
Ray Le Maistre, TelecomTV (10:50):
Absolutely. Now, I mean the whole transformation that the telecom operators are going through, it's a tough one. People, processes, tools, all of these eat up resources and there might be a temptation for some operators to hold off, wait and see. But of course the danger then is that you miss out on the benefits. What would be your advice to operators that are considering not investing now and therefore maybe not getting the benefits as soon as others?
Matt Conrod, Intel Corporation (11:27):
Well, yeah, I mean what we're seeing in the market is there are learnings going on. So if you haven't started the virtual and open ran journey yet, talk to people that have being a fast follower or a second adopter, you have an opportunity to obtain a tremendous amount of knowledge that's been learned the hard way. Talk to the ecosystem that have deployed at scale, talk to other operators that have deployed at scale and track your own journey. Secondly, most organizations learn when they have hands-on experience. So get to pilots, get to deployments in the field and bring in the operations team. Don't let engineers manage these pilots. Make the operations guys manage the pilots and that will force the learning to start the people process tools you mentioned. You need to start that. And then I would say lastly, every two years Intel comes up with a new X 86 piece of silicon. There's new technology being driven by the hyperscaler community. It's, it's rapid. But my recommendation to the telecom community is not wait. There's a risk that you're always waiting for a better widget and you can take advantage of future technologies, but you need to start. So start small, start with today's technology. By the time you're ready to scale, there will be a better product. But that start needs to happen and do it early.
Ray Le Maistre, TelecomTV (12:50):
Okay, Matt, well that sounds like sound advice for the industry. So thanks very much for joining us today, both of you, and enjoy the rest of the show.
Matt Conrod, Intel Corporation (12:58):
Thank you.
Udayan Mukherjee, Intel Corporation (12:58):
Thank you.
Please note that video transcripts are provided for reference only – content may vary from the published video or contain inaccuracies.
Matt Conrod, & Udayan Mukherjee, Intel Corporation
Udayan Mukherjee and Matt Conrod from Intel discuss the progress being made in the development of Open RAN, the improving technology and total cost of ownership (TCO) metrics related to Open RAN, the role of AI in the RAN, and the benefits of hands-on experience with open, disaggregated architectures. All in all, say the Intel executives, the business and technology case for Open RAN stacks up and has never been stronger.
Recorded November 2024
Speakers

Matt Conrod
vRAN Business Development Director, Intel Sales and Marketing, Intel Corporation

Udayan Mukherjee
Senior Fellow, Network and Edge Group, Intel Corporation