During the last World Mobility Conference (MWC) held in Barcelona, Intel spokespeople repeated a phrase that, for some, could be cryptic. They talked about modernizing the networks, monetizing the edge and taking AI everywhere. But Intel is not the first brand you think of when you talk about network providers, obviously there is something else. Presentations by Alexander Quach, Vice President and General Manager of Intel’s Core and Fixed Networks division, along with Pallavi Mahajan, Corporate VP and General Manager of Intel’s Network and Edge Group, helped to clarify the ad chain and its impact on the telecommunications service provider sector.
“We’re really excited to provide some updates at this MWC, within the context of modernizing networks: a journey that Intel started a decade ago,” Quach said at the opening of the roundtable that Intel was part of. The executive clarified that Intel was one of the pioneers in what gave the name NFV (Network Function Virtualization), a technology that helped open the game of telecommunications networks to information technology providers. A decade ago, much of the infrastructure came from proprietary closed boxes, and the virtualization of network functions contributed to those same functions being performed on standard IT equipment. For example, servers that use Intel Xeon processors.
“The arrival of 5G really accelerated that movement. So virtualization probably covers between 80% and 90% of wireless call sessions today,” he added. “And we are expanding virtualization within the RAN (radio access networks) as you have seen in recent years”. When Intel talks about vRAN it refers specifically to the virtualization of RAN capabilities.
The announcements on networks
At MWC Intel announced the next generation of the Intel Xeon platform, which is advancing at leaps and bounds in both performance and function, but especially in energy efficiency. “Operators around the world are very concerned about getting the best power efficient solutions for their networks. This delineated how we think of the next generation of processors, codenamed Sierra Forest, which delivers a performance that is 2.7 times superior in the same kind of power envelope we’re deploying today,” Quach said.
This means that any operator can take a rack, and measure what Sierra Forest delivers versus what you have installed today. The result is at least two and a half times (2.7). “This is absolutely huge, and it’s what service providers are looking for,” Quach sums up.
“OPERATORS AROUND THE WORLD ARE VERY CONCERNED ABOUT GETTING THE BEST POWER-EFFICIENT SOLUTIONS FOR THEIR NETWORKS. This delineated how we think of the next generation of processors, code-named Sierra Forest…»
“In addition to this, we are announcing the commercial availability of the Intel Power Manager for 5G Core, which we announced last year”. Companies like Nokia are already integrating this Intel management software into their own software to take advantage of all the capabilities in the processor, and save 40% power in their networks. “If we consider that this consumption represents between 20% and 40% of the operating expenses of service providers, we will see that this is a very important achievement,” said the executive of Intel’s Core and Fixed Networks division.
Moving from the core of the network to the RAN space, Intel made big announcements last year about vRAN Boost and the Sapphire Rapids-EE processor platform. “This year we are announcing the future generation of processors that can be used in vRAN, which has the code name Granite Rapids-D, which promises more performance and savings in consumption. They will be released sometime in 2025, after the launch of Granite Rapids processors later in 2024,” said Quach, who added: “The ecosystem with which we work to bring these solutions into reality already has samples of the Granite Rapids D. Such the case of Samsung and Ericsson who already made the first calls on the Granite Rapids-D”.
Artificial Intelligence in the edge
In his presentation, Pallavi Mahajan summarized the different waves of computation that, this time, are moving from the core to the edge (core to edge). That edge, he explained, is governed by what AWS calls the Three Laws of Distributed Computing: the law of Physics, which limits latency, performance and availability of network connectivity; the law of the Economy that determines the profitability of the transfer of increasing volumes of data; and the law of the Country that regulates how the data are handled and where they can be stored. The other point to take into account, as already seen, is energy consumption.
What Intel is aiming for is continuity between the clouds and the edge. “Intel has had over 90,000 real-world deployments and we sold over 200 million processors, which gives us a unique experience to understand the edge. Now we wanted to make it simple so our customers can move forward and deploy more edges,” she said. It is in this context that Intel presented its Edge Platform, designed to develop, deploy, secure and manage edge and AI applications on standard hardware, but can scale in a simple way, as if it were a cloud. All of this is AI-optimized and delivered in a modular and open way for greater compatibility of components from multiple manufacturers.
Mahajan explained that the platform will allow more devices to be deployed on the edge, and that this can be managed centrally (a single glass panel, in technical jargon). It is worth remembering that this is an evolution of the solution first presented at the end of September 2023 at Intel Innovation under the key-name Project Strata.
“Intel has had more than 90.000 deployments in the real world, and we sold more than 200 million processors, which gives us a unique experience to understand the edge…»
The missing piece to take AI everywhere, especially in the context described above, was also announced in the framework of the MWC: the Intel vRAN Developer Kit. This set of tools and platforms will allow developers and operators to build, train, optimize and deploy AI models for vRAN-related use cases. This includes software with AI capabilities and the fourth-generation Xeon processor that has built-in AI acceleration power management, plus enhanced telemetry capabilities. With all this, operators can use the kit to dynamically reconfigure their network on the fly, save costs, get more value from their infrastructure and support new revenue streams. As mentioned, all this without resorting to infrastructure
