GenAI Taking 5G by Storm: Amdocs, NVIDIA, Ericsson & AT&T Moves – 5G Factor – EP73

Hello and welcome everyone to the 5G Factor. I'm Ron Westfall, research director here at the Futurum Group, and today I am focusing on the major 5G ecosystem developments that have caught my eye, including Anox and Nvidia working together on you guest IT AI capabilities, in particular j ai. Also, Ericsson has launched new, uh, WNI, I guess that's one way of describing it, uh, products.

And also I think it's important for us to touch on what at and t is also doing with ai. So with that, let's jump right in. Uh, this week at TM Forums Innovate America's 2024 event.

I believe that Amdocs and NVIDIA showcased how strong their partnership is in potentially unlocking gen AI for communication service providers. And that, of course impacts their 5G networks. And what they're looking to do is optimize large language models to speed adoption of gen AI apps and services.

And what that means is that the delivery of real world apps of AI that are designed specifically for the unique needs of the telecom industry, and this is where, uh, where AM's DNA is, it certainly has proven very adept at meeting the needs of telcos throughout the world. They count up to 400 CSPs as customers, um, as of, uh, this last quarter. And also Nvidia, as we know, is definitely interested in expanding its presence and influence throughout the telecom realm.

As we saw with the recent AI ran launch, that also included a partnership with T-Mobile, Ericsson and Nokia to basically use T-Mobile's Gen AI innovation labs to really test, or I should say the AI ran innovation labs to test the emerging capabilities of AI in terms of enabling much better ran outcomes. Now, with that in mind, to achieve this, and Docs and Nvidia are customizing enterprise grade LLMs running on Nvidia accelerated computing as part of the Amdocs a Maze framework. And this is important because what it's doing is it's aiming to empower CSPs to efficiently deploy gen AI use cases across their business from customer experience to networking, uh, provisioning.

And I think we're seeing a real progress. Basically, the entire Amdocs portfolio is using gen ai, uh, capabilities. And this is something that I think the company has made strides in terms of its positioning at demonstrating, and I'll touch on this, how this can positively impact these outcomes for CSPs.

Now, what is going on is that Amox is using Nvidia, GGX cloud ai, supercomputing, and Nvidia AI enterprise software to support flexible adoption strategies and help ensure that CSPs can simply use gen AI applications, uh, in a safe manner naturally. Now, notably the collaboration with NVIDIA builds on the Amdocs Microsoft, uh, partnership. And I think this is something that's important because basically Amdocs is bidding that its relationship with Microsoft can help them win more business, particularly when it comes to what can be traditionally described as customer relation management applications.

But we're seeing that's expanding into really a wider range of requirements and capabilities. And so this is enabling Amdocs, uh, fundamentally to compete more directly against rivals, uh, such as Salesforce as well, ServiceNow, which is up and coming, I believe, within, uh, the telco realm. And so this is something that I think is going to create new competitive dynamics, but also the fact that Amdocs is aligned with Nvidia and Microsoft definitely will help its channel relations and also, you know, its overall credibility in terms of why Amdocs, when it comes to the question of using AI capabilities, including particularly gen ai.

Now I find that CSPs are beginning to benefit from enhanced performance, optimized resource, uh, utilization and the flexible scaling to support their emerging, uh, capabilities, but also meeting future needs. So that is both on the customer facing side as well as on the back office side. Now, to back up the vision, AMOC spotlighted how this telco specific knowhow has already led to significant results.

And what they cited is that when it came to using Gen ai, a leading North American service provider, and we'll, you know, keep it a non naturally, uh, but still it's demonstrating the point very effectively saw a 60% reduction in token consumed in data processing, a 30% increase in response accuracy due to the use of telco specific knowledge and a nearly 80% reduction in query latency. Now, I find that such gen AI metrics are gonna become increasingly important in understanding who's providing the best solutions for their needs, but also a better understanding 5G customer and workforce experience, uh, outcomes. And also just be a better understanding how an operator A CSP, uh, can, um, definitely deliver a better 5G service quite simply.

And so this is going to impact everything from price optimization, value packaging, you name it. And this is, I think, going to be good news for CSPs as they figure out, okay, how are we going to energize, you know, our revenue streams, diversify them more effectively and so forth. As you know, basically the overall industry is kind of going through a lull right now, but we've been here before and I'm anticipating that ai, uh, just through partnerships such as Nvidia and Amdocs, will play an integral role in stimulating new CSP activity, including particularly in the 5G side.

Now, when it came to real world business value, it, uh, brought to customer care. This same provider sought incredible 63% reduction in average handling time, a 50% improvement in solving problems on the first try, and a 49% increase in net promoter score, which is certainly very important to, uh, an ECSP out there. And this again, is reinforcing that gen AI is improving business outcomes across the telecom industry.

And these are, I think, very impressive figures. I mean, we're seeing them in many parts, for example, in, uh, proved deficiency and customer support, but also in terms of, uh, you know, naturally, uh, you know, customer call, uh, support environments and so forth. And, uh, so stay tuned.

I think we're gonna see more of these results coming to the forefront and quite simply improving the efficiency of operators in the back office, which I touched on, but also just making things better for their customers across the board on the consumer side, but also on the business side because as, as we know, the telcos are looking to expand their ability to meet, uh, business enterprise requirements, including in particular B2B capabilities. And so AI will be a valuable tool in meeting that particular objective. Now, let's turn, and this is still related to ai, uh, you can be rest assured to Ericsson announcing the launch of two new routers, the Ericsson Cradle point R nine 80 and S 400, which are designed to provide enterprises with secure and scalable connectivity for enabling ai.

I see, I touched on it again, computer vision data analysis of other advanced applications across IOT environments and vehicle networks. Now, with the integration with EE Ericsson's net cloud platform, including net Cloud, SSE Secure Connect, these routers can offer smooth deployment management and robust zero trust security across large scale IOT and vehicle endpoints. Now, the announcement is important not just because of these new products, because it also ushered in a new branding approach for Ericsson with the formation of the Ericsson Wireless Solutions unit, combining Ericsson CSP focused 5G infrastructure, DNA with cradlepoints enterprise focused portfolio assets and channel prowess.

And certainly the launch of these two new products are aimed primarily at enterprises, but it it's showing that Ericsson brand name is out in front and it's leveraging really the trust that Cradlepoint has established throughout the enterprise channels. Now, uh, specifically the Ericsson Cradlepoint R nine 80 is a ruggedized 5G router that is designed to deliver high performance connectivity to vehicles enabling the deployment of applications such as AI driven video recognition and real time data analysis. With the increased use of these technologies in public safety and smart cities, 5G especially 5G standalone is providing what I believe is more reliable and agile connectivity.

And why is this so, you know, important? Well, I think we all probably have friends or colleagues who're dealing with Hurricane Helene and its wrath throughout the southeast us. And so this is connected.

This product is designed for first responders near shore vessels, and as well as other important vehicles such as school buses. And what is going on is that the compact R nine 80 delivers secure and persistent 5G connectivity even in the most demanding conditions. So while it's certainly Hurricane Helene, it's generating headlines right now, but this is gonna be true for any public safety challenge that arises.

And as a result, it can also be used in iot deployments where video and mission critical applications may require higher performance and lower latency. So this is really good news or certainly public safety, but also for just making enterprises smarter about what's going on at their facility or, you know, improving, uh, their overall uh, security profile. Now with this third gen 5G architecture, the R nine 80 is three GPP release 16 compatible, and that is aligning with the readiness for standalone networks.

I touched on that and network slicing, including public safety slices and low latency network slices for IOT. And now we've been hearing about slices for a while and they have actually been coming on board, but I think this is gonna be important for us CSPs and their ability to again, generate those new revenue streams and differentiated offerings. Now the routers new embedded eim can enable over the air carrier switching and future net cloud releases, adding deployment flexibility and operational efficiency.

I think, you know, carriers are now are very adept at being able to use the cloud in order to take advantage of capabilities such as this. Now it's also I think, important to note that emerging AI applications also rely on gathering data from distributed widespread IOT devices, really across the edge now as one of the first zero trust IO OT appliances, at least that that is from I view Ericsson's other offering the Cradle point S 400 enables secure connectivity for other diverse IOT lease, uh, use cases including light industrial digital signage and building controls. As a semi ruggedized compact IOT VI IOT device, its design can offer field expandable interface options such as additional ethernet ports, a serial port, and a general purpose input output switch.

Now this adaptability can allow i it specialists to tailor the devices to specific deployment needs and to ensure that even small vulnerable IOT devices which cannot support traditional security methods remain secure. The S 400 includes net cloud, sass, e secure connect as part of its subscription license, and I think we'll be seeing more of this quite simply. The SASS e approach or architecture will be a fundamental in terms of being able to meet these types of use cases that is combining the cloud with a a physical product in order to meet the specific needs of the customer.

Now, secure connect, uh, enables the deployment of a zero trust network instantly providing easy to deploy, uh, as well as comprehensive security for large scale deployments. And that certainly applies to IOT devices and fleet vehicles or, uh, vehicle networks in general. Now, overall, I find that the launch aligns with the edge representing the next best opportunity for CSPs, telco pops, uh, brick and mortar prime real estate assets can also play a role in this as AI capabilities are being increasingly distributed across the IO OT environments.

As we saw with this launch, but also further out, I anticipate that the operators will need to look at ways to offer AI interesting services using its vast edge resources that hyperscalers even cannot emulate today. So today the new Ericsson devices are well suited to advancing AI capabilities of the CSPs. Today, it would be across iot environments and certainly public safety environments, but also I think it can als, uh, serve as a linchpin for the carriers to be smarter about how they support AI and deploy ai, including ultimately them being able to offer their own native, uh, AI inferencing services as a service, but also, uh, you know, partner with hyperscalers in a more flexible fashion rather than just leasing these A assets outright to the hyperscalers and letting them, uh, in effect that the lion share of these, uh, revenues.

Now speaking of AI and operators, let's turn to what A&t is doing. And at and TI thought that, that this is, uh, very interesting reports that ask at and t is driving business value by making employees data experts. And I think, uh, we understand that AI has the potential to really democratize knowledge across organizations, and this I think is being demonstrated by at and t.

Now, there's always gonna be room for some expertise. However, what this is, I think going to be a difference maker is enabling, uh, say low to no coding capabilities, streamlining really workflows across organization. And these are, you know, just some examples, but I think, uh, there are some numbers here that have been impressive.

Now, at t is generating a billion tokens per day right now, and a two a review a token and Gen ai, whether powered by an LLM or increasingly popular small language models, it is essentially equivalent to a generated word. So fundamentally, gen AI is cranking out the equivalent of about a million a billion words per day at at and t. Now, these are everything from automated summaries of inbound customer calls, which saves between 30 seconds to several minutes per call.

We've already touched on that in terms of the Amdocs Nvidia Alliance, but also, uh, to suggestions on relevant products and services for customers to lines of computer code. Again, the low code, no code paradigm and well stay tuned for more. Now that's a level of information and automation that simply wasn't possible before.

Uh, I think we all understand that now, at t's asked data tool currently ranks number one in the world on the gen AI bird leaderboard. And this is constantly updating scoreboard tracks, which j AI platforms around the world are the best at translating, playing text queries into SQL programming language and SQL is within the, uh, the telco realm, but other industries the standard for querying in manage, uh, and managing large databases. So this is really good news.

Certainly at and t is reporting that this is making a difference for them today, and it's really laying the foundation for even, I believe, more impressive results. Now, what that ranking means is that ask at and t is exceptionally good at taking standard plain English questions and turning them into computer code to find insights and data. And I'm finding that at and t has put a lot of work into refining is ask at and t capability off the shelf gen AI platforms could be useful.

But what at and t is pointing to here is that they've shown that accuracy and value increase substantially when you fine tune models on your own data and also use approaches such as retrieval, augmented generation or rag along with vector databases. Now, from, uh, my understanding, I think these capabilities are gonna be important not just today, but for say the next year or two, but ultimately the LLMs could in essence take on the vector database and RAG capabilities natively. So, uh, that those capabilities won't have to be add-ons, if you will, but it's kind of a moot point because right now it's making all that difference in terms of enabling the CSPs to become better in terms of operational efficiency, in terms of proving customer experience, et cetera.

You heard all the data, all the results, and this is all something I, I think, uh, we can believe will make a difference in stimulating, uh, CSP innovation and investments over the next, well 12 months, but also further out. And, uh, kind of as a coda here at and t is driving now to move from as data to explain data automatically building models to explain what is going on, you know, with sales, with, you know, churn rates, uh, with fraud detection, you name it, this is going to get better and better, at least that is the trend line here and then act on data. So this is really, I think, you know, taking it to the next level, they're just learning more.

The training is paying off, and now the inferencing is gonna be able to allow employees, the workforce to really run with it. So folks, stay tuned through your j ai prompts on what's going on with gin AI throughout the CSP world, let alone the overall digital e ecosystem. And with that, everyone have a great j AI and 5G Day.

And also before I sign off completely, don't forget to bookmark 5G Factor. It's available on, uh, the future and group, uh, website, but also available at techron tv. And uh, certainly I appreciate those folks who take the time to tune in and listen and looking forward to our next 5G factor.

Again, great 5G Day, everyone.