Intel's Nick McKeown: Defining the edge to lead the frontier

Intel’s ambitious goal at the edge is to provide customers with the best programmable platform in the industry.

summary:

• From wired data center networks to telecommunications networks to the edge—Intel is a leader in network transformation.

• The Internet is undergoing a paradigm shift from centralized computing and data storage to distributed computing and network architecture.

• The resulting explosion in edge computing presents tremendous opportunities for Intel and the industry as a whole.

• Intel’s edge strategy aims to provide customers with the industry’s best programmable platform and transform global networks and enterprise operating systems into software-defined and programmable forms.

Nick McKeown, Senior Vice President and General Manager of the Network and Edge Group at Intel Corporation

Edge computing solutions based on Intel technology have supported thousands of deployments, creating tremendous value for Intel customers and billions of dollars in revenue for Intel. Our customers use Intel products to build IoT solutions, 5G infrastructure, and enterprise network products deployed at the edge. We are working with partners and customers to deliver and deploy edge solutions based on our portfolio of Intel® Xeon®, Intel® Core™, Intel Atom®, FPGAs, vision processing units (VPUs), and network silicon, and using open containerized software frameworks such as Smart Edge, the OpenVINO™ toolkit for AI inference, and our Infrastructure Programmer Development Kit (IPDK). 1 At Intel, this business is booming and growing.

Shifting to a distributed architecture

But I believe this is just the beginning. We are experiencing a paradigm shift in how we use the Internet and deliver software applications: from centralized computing and data storage clusters to a more distributed computing and network architecture. This "cloud-to-edge" infrastructure model combines the cloud's unparalleled scale and capacity with the edge's faster response times and closer data storage. At Intel, we see this model as a "superpower" that can drive digital transformation across industries.

In addition, various social changes are also driving this transformation: the rapid development of remote and hybrid work modes; the gradual digitization and automation of work in physical operations such as factories, retail stores, and healthcare services; and the increasing privacy issues and related government regulations require data to be processed and stored close to its source. These changes require us to deploy a large amount of computing infrastructure to the edge, allowing edge artificial intelligence platforms to interpret the massive amount of generated data, while promoting intelligent automation and transforming the network into a software-defined and programmable form, so that workers connected to the network anywhere can dynamically connect to the distributed computing infrastructure.

My 25 years in the technology industry, both as a professor at Stanford and as a founder of several startups, have convinced me that this paradigm shift and the resulting explosion in edge computing is a significant opportunity for Intel and the industry.

But where exactly is the “edge”? The term is actually a broad umbrella for many locations that all have in common that data computation and storage are located close to where the data originates or where the data is processed (and/or close to the end user). The edge includes the “on-premises edge” (where most data is generated and will host a large portion of AI workloads), the “telecom network edge” with carrier data centers and central offices (where software-defined 5G/6G networks are deployed), and the “hosted edge” where computing facilities are owned by a hosting or cloud service provider (where software-defined enterprise networks, security applications, and latency-sensitive applications such as gaming are deployed).

Intel's Edge Strategy

Intel’s ambition at the edge is to provide customers, especially their software developers, with the industry’s most advanced and broadest programmable platforms. Our edge strategy is two-fold: providing programmable platforms for:

• Transform the operational systems of each enterprise location - whether it is assembly line monitoring on a factory floor, POS machines in a retail store, patient monitoring systems in a hospital, or supply chain management in a logistics center - into software-defined automated systems and use artificial intelligence technology on a modern cloud-native software platform.

• Transforming the global network into a software-defined and programmable network, enabling these distributed enterprise locations to flexibly and securely connect remote and onsite workers and mobile consumers.

Intel has world-class process, manufacturing scale, and software capabilities, and we are using these capabilities to support the implementation of the above strategy and accelerate the deployment of edge computing solutions to help achieve these transformations. At the same time, we will also give full play to our leadership and experience in the network field in this process.

Network Leadership

For more than a decade, Intel has been leading the global transformation of networks from legacy fixed-function hardware to networks defined by open, interoperable software, a transformation that spans both wired data center networks and cellular wireless networks. In 4G/5G networks, we are driving the core network’s transition to virtualization with network function virtualization and radio access networks with virtualized RAN and open RAN capabilities. We are making significant progress in this area—more than half of new 5G core network deployments in 2021 are virtualized, and the vast majority of commercial 5G vRAN deployments are running on Intel platforms, including 140 licensees using the Intel FlexRAN™ software reference architecture. A great example is Japan’s Rakuten Mobile, which is delivering world-class 5G services with an end-to-end fully virtualized, open, cloud-native network built entirely on Intel platforms .

Similarly, we are leading the transformation of wired data center networks - transforming them into software-defined fabrics based on Ethernet adapters, infrastructure processing units (IPUs), and P4 programmable intelligent fabric processors (IFPs). Once you have the ability to program devices in the network, you can give unprecedented flexibility to network innovation without changing the hardware. Having this capability end-to-end means that all network devices in the data path are programmable, which will further enhance the flexibility and scalability of the network.

Empowering edge partners through AI and 5G

Intel and its partners are working together to provide solutions to a wide range of customers in vertical industries such as retail, banking, hospitality, education, manufacturing, energy, healthcare and pharmaceuticals to drive digital transformation and automation in the enterprise. We are helping these partners build edge platforms that enable data processing, analysis and storage closer to its source. To do this, we provide partners with powerful AI and 5G building blocks to enable fast, real-time analysis and response. AI—especially edge inference—can provide useful insights in real time, right where and when data is generated. As a result, it is becoming the most common use case in edge computing, enabling transformation and automation in factories, smart cities, hospitals and other places. Programmable 5G networks enable our partners to connect a large number of sensors and smart devices (such as robots) with software and AI running on Intel programmable platforms through high-bandwidth, low-latency, reliable and deterministic wireless connections. Intel is working with partners to achieve our shared vision: autonomous operations at the edge of the enterprise.

We don’t stop at building programmable hardware platforms, we also provide developers with open software building blocks (such as IPDK) that enable them to access our hardware capabilities on modern cloud-native software platforms (such as Kubernetes). In addition, we also provide toolkits to help developers develop optimized inference models for the edge using OpenVINO. Intel will continue to be committed to creating the most exciting development experience for edge software developers.

With these platforms, Intel is creating a dynamic and reliable path to ubiquitous computing, from the cloud to the intelligent edge. In turn, our customers, software developers, technologists, and end users will innovate on these platforms to create amazing applications that will transform lives and businesses in ways not yet imagined.

1 Note : Revenue from FPGAs and a significant portion from Xeons is included in Intel’s Data Center and Artificial Intelligence (DCAI) business segment.

2Rakuten Mobile uses Intel Xeon processors, FPGAs, network interface cards (NICs), FlexRAN, and Smart Edge Open.

Intel, the Intel logo, Xeon, OpenVINO, and the OpenVINO logo are trademarks of Intel Corporation or its subsidiaries.

No product or feature is completely secure.