Could there be good news for telcos staggering under declining profit pressure? NVIDIA wants to come to their rescue! Of course so does everyone else, and so far “rescue” seems to consist of wanting telcos to spend on the rescuers’ products without any real proof it will help. The question is whether NVIDIA is any different.
NVIDIA does have a telco industry initiative and web presence, and the opening sentence makes it clear what they want telcos to believe: “The nexus of AI, the Internet of Things (IoT), and accelerated computing at the edge is turbocharging network performance and efficiency.” At the moment, at least, they’re doing no such thing, but hey, hyperbole is the heart of the industry these days. They then say: “The world’s leading telcos are turning to NVIDIA technologies to create AI-enabled solutions, build software-defined and accelerated infrastructure for 5G, and bring connected intelligence to smart devices at the edge.” Also not strictly true, but maybe the point is that it should be. Let’s see how well NVIDIA makes that point.
There are three points you can dig out of NVIDIA material to boost their claim. One is that you can link AI to the standards evolution in the mobile space that’s the only credible path to getting any significant telco budget money, and that includes both 5G and 6G. The second is that IoT is perhaps the foundation of any credible new service set, and the final one is that to make AI and NVIDIA credible to telcos, you need to have an alliance, an industry body to unify players and build media momentum. NVIDIA aligns its AI into three areas, accelerated RAN, AI-powered operations, and generative AI infrastructure. We’ll work these three areas with the three points here.
I think that NVIDIA’s first point is a mixed bag as far as value is concerned, but that they try hard to justify it with those three credibility areas. It’s surely true that the only thing that has firm budget support in telco-land is mobile standards evolution. The problem is that telcos got burned on 5G, and while there’s surely some interest in redeeming 5G, it’s been tried and has failed so far. 6G is obviously tainted at the moment, but because it probably won’t see much light for another five years or so, all that may be forgotten.
The problem I see, though, is that the great majority of 5G buildout is mature, which means that the accelerated RAN story isn’t helpful to most telco engagements. That leaves AI operations as the network/5G story that has to carry the water. NVIDIA’s example of the chatbot is hardly revolutionary in that space, and the biggest problem is that telco operations software is all over the place; not the least because so much of it is from network equipment vendors.
The notion that IoT would have to be the foundation of any credible new service story is a good one, but one that’s particularly difficult to push from the chip side, and to a telco audience. NVIDIA’s own theme seems linked more to IoT connectivity than anything else, and we’ve already proved that telco initiatives in 5G connection and device control have little impact.
That doesn’t make IoT any less than critical to both NVIDIA and telcos. There is no area, none at all, where so much infrastructure, meaning telco cloud edge computing, infrastructure could be expected at in IoT. The problem is that NVIDIA chips have to go into servers that then run some sort of credible IoT middleware and application set, and we don’t have that today. NVIDIA doesn’t promise it, either, and you can surely argue that as a chip vendor that’s not a surprise.
It’s still a problem, though. Why have a telco sector strategy whose strongest pillar isn’t really in your own wheelhouse? NVIDIA has worked to develop a software ecosystem for GPUs in graphics, photography, and video. Why not IoT? The reason, I think, goes back to the telcos’ own focus on IoT, and its limitations.
When IoT was first suggested as a telco opportunity, an operator planner (cynical about his company’s plans) called their initiative “refuge from the saturation of the human market for mobile services, in a new market where you can manufacture consumers instead.” This was a clear reference to the notion that telcos could connect IoT devices with their own mobile plans.
Some of that is reasonable. There are applications for telco-connected IoT in security, and even in network management, as a backup should primary network connectivity fail. There are also applications in the utility, energy, and transportation, industries. The problem is that all of these predate 5G and are already being exploited. What NVIDIA needs to do is to assess what winning roles for telcos in IoT would look like, then present both a vision and the tools needed to realize one or more of those roles, more through vendor relationships than directly with the telcos.
Nowhere would that be more important than in the AI-RAN alliance, but I think the linkage of the group to RAN, which links it to 5G (where it’s too late) and 6G (where it’s too early) is a serious mistake. It would have been far smarter to have created the “AI Factory Alliance”, but that notion is linked to telco edge-cloud, and I wonder if NVIDIA fears alienating the cloud providers whose use of AI is NVIDIA’s main revenue driver.
Several years ago I modeled the impact of a telco edge cloud deployment, and it generated a hundred thousand new data centers globally, with IoT/metaverse being the major long-term driver. That’s surely behind NVIDIA’s carefully worded promotion of the AI factory concept, and a big part of the linkage of edge computing with telcos is the fact that telcos have the real estate to house these new data centers. Public cloud providers could also deploy toward the edge, but at a higher new cost to house their equipment.
NVIDIA needs to do more for telecom, if it expects any useful results for itself, for the telcos, or for the vendors who serve as the bridge between chips and services. Despite a lot of natural advantages, and despite a lot of potential to play a game-changing role, they’re not there yet.
