How Nvidia is helping telcos take advantage of artificial intelligence

The infusion of artificial intelligence (AI) into everything is literally changing our lives from day to day. People use voice interfaces, purchase articles based on recommendations, and routinely log into facial recognition phones because it makes their lives easier.

One of the few industries yet to jump on the AI ​​train is the telecom sector. This is a huge opportunity because AI has the potential to transform telecom operations and improve efficiency in areas such as call center automation and much more. In the cloud era, telcos are in a unique position to offer new services that differentiate their networks from competitors. AI for wireless is now the most transformative technology for telcos.

Nvidia, the industry leader in all things AI, has invested billions of dollars in telecom-focused computing technologies. One of Nvidia’s differentiators is that it builds full-stack, fully integrated products that are easy to deploy, allowing them to make quick money. Telcos are uniquely positioned to take advantage of AI-based technologies because they collect a massive amount of data, process millions of transactions per second, and manage tens of millions of devices simultaneously.

“We want to help the telcos create AI centers of excellence where they can maximize productivity, shorten the path to intelligence and create significant total cost of ownership for their businesses. We can partner with the telcos on AI operations, especially to help them speed up processes and expose inefficiencies,” said Chris Penrose, head of telco sales and business development at Nvidia.

Digital twins can transform telco business

An example is the optimal route determination of trucks. Telcos can use Nvidia’s AI capabilities to send engineers to the field more efficiently. Another example is building digital twins for telco networks to optimize scheduling and performance, using the Nvidia Omniverse virtual simulation platform. The latter can save a lot of time and money, as this is usually done by building a second physical network.

For example, Ericsson is building in a digital twin for 5G networks Nvidia’s Omniverse to optimize tower clients for maximum performance and coverage, based on real city environments. With more than 15 million microcells and towers planned to be deployed worldwide over the next five years, the use of digital twin is an optimal way for telcos to test their networks before they are rolled out to the public.

“By bringing together Nvidia’s AI computing platforms and the vast array of data telecoms, we can really help transform and unify the customer experience across all operations. The AI ​​models can also be used to reduce customer churn and generate more revenue through recommendations,” said Penrose.

AI can help telcos provide better customer service

During COVID-19, T-Mobile’s call centers were inundated with payment requests due to pandemic financial difficulties customers were experiencing. Nvidia partnered with T-Mobile to automate the process by rolling out a chatbot that handles payment arrangements. The side project grew into a frequently used tool among T-Mobile customers. T-Mobile also implemented Nvidia’s speech-to-text tools to transcribe conversations between agents and customers, which would normally be a time-consuming task at the end of each call. Telcos have received subpar customer satisfaction scores regarding call centers in the past, and the use of AI could help reverse this trend.

Success at the Edge Requires AI

AI-powered telco edge solutions are another area of ​​focus for Nvidia. Telcos spend billions of dollars deploying 5G networks, and Nvidia wants to help telcos capitalize on their investments by using AI-powered edge solutions. A realistic scenario where such solutions can be used is traffic management, i.e. using AI to optimize real-time traffic flows and traffic lights over 5G.

Telcos have transformed their infrastructure from purpose-built hardware to cloud-native and virtualized. It turns out that central processing units (CPUs) are not very efficient for data center infrastructure.

“What if every multi-access edge computer (MEC) could also host a 5G virtualized radio access network (vRAN). What if telcos could build edge data centers and host 5G vRAN as the anchor tenant in data centers? enable AI to create edge apps and generate new revenue streams, we believe this will be a game changer for telcos: building edge AI data centers, running 5G as another cognitive function,” said Soma Velayutham, global industry business development lead for telecom at Nvidia.

In 2021, Nvidia introduced its AI-on-5G card, which the vendor describes as a “hyperconverged, graphics processing unit (GPU)-accelerated, edge data center-in-a-box.” The card can be used for both high-performance computing AI workloads and 5G RAN, which is implemented as an additional software stack.

“When delivering 5G and AI in business apps, the two are very different. 5G is extremely latency sensitive – we are talking milliseconds, while AI is more prioritized in terms of latency. 5G timing is critical when you enable a virtual local area network (VLAN). That’s why we created the AI-on-5G map. You can run any of the five functions or all five functions on one card, which is quite unique,” Velayutham said.

With this innovation, enterprises can add and manage 5G connectivity as part of their IT infrastructure, while telcos can transform all 5G base stations (gNBs) into edge data centers. Nvidia envisions AI-on-5G, as well as the other AI-based capabilities mentioned above, powering the data centers of the future.