- Verizon has been talking about MEC for years
and years and years and years - But wait. Verizon is partnering with Nvidia on this deal, which adds AI to the mix, so it’s all good, right?
- Sorry if we sound a little jaded
Remember when it was cool to talk about mobile edge compute (MEC) like it was going to change the world — or at least make 5G private networks more attractive by boosting download speeds and reducing latency?
OK, reality check. We’re not sure those days ever existed. But guess what? The “catchy” MEC acronym has been replaced by artificial intelligence (AI). All you have to do is add a little “AI” to spice up your press release and journalists are bound to bite.
That’s exactly what we did when we read this week’s announcement about Verizon and Nvidia.
Wait, you didn’t hear? Verizon and Nvidia developed a new solution that enables a range of AI applications to run over Verizon’s 5G network using private MEC. By blending Verizon’s private 5G know-how with Nvidia’s snazzy software and inference microservices, Verizon is betting enterprises will take the bait.
Asked what spectrum will be used for these private networks, a Verizon spokesperson said that it’s a spectrum-agnostic architecture. “We can build this private network solution to spec to meet the needs of the solution it will be running using any of the spectrum we have available,” the spokesperson told Fierce.
When it comes to spectrum, we had our money on C-band and Citizens Broadband Radio Service (CBRS) spectrum, which Verizon has been using in NFL stadiums and other venues across the country. But given that Verizon used 28 MHz spectrum when it announced its MEC integration with Amazon Web Services (AWS) back in 2019, the multi-pronged spectrum strategy makes a lot of sense.
That AWS deal came around the time Verizon CEO Hans Vestberg was singing the praises of MEC like nobody’s business. Spoiler alert. It didn’t turn out that way. Industry analyst Dean Bubley of Disruptive Analysis called it when he described the MEC market more like “MEH,” for Mobile Edge Hype.
Needless to say, Vestberg has since come to acknowledge that “probably I was early out on mobile edge compute.”
AI adds sizzle
Do the prospects look any better today? After all, Nvidia is the hottest AI ticket in town. Surely Verizon’s collaboration with Nvidia to power AI workloads on private 5G networks must represent a big advancement in enterprise AI capabilities.
Analysts contacted this week were both cautiously optimistic and skeptical.
AvidThink’s Roy Chua recalled use cases, like robotics, and discussed during the first round of MEC go-to-market hype in the 2018-2019 timeframe. “I think it was a case of Hans and Verizon (and AWS) being early to market and betting that if they built it, the use cases and customers would come,” Chua told Fierce. “We've all collectively learned a lot since then and recognize there's value but the market and use cases needed time to catch up.”
With the generative AI buzz around Nvidia and its aggressive push of the Nvidia AI Enterprise software platform and Nvidia Inference Microservices (NIM) into the market, “I think both Verizon and Nvidia see a new window of opportunity to jumpstart the market again for MEC and the same use cases we were hoping would take off back then,” he said.
While the traction for these workloads is getting better, it's not clear yet that many of them will immediately succeed, although language models with computer vision use cases and robotics could be valuable, he noted.
Chetan Sharma, CEO of Chetan Sharma Consulting, said things are moving in the right direction. “Having compute capability at the edge was always desirable especially if you can build a national fabric, but it remains to be seen if this will ignite the demand for private networks/edge in the U.S.,” he said.
“China is much farther ahead in executing on private/edge strategy compared to the U.S., but we could potentially see some pickup as the markets get educated about the available solutions,” Sharma told Fierce.
Recon Analytics founder Roger Entner was harsher. He’s not a big believer in private networks. “I think there are now more press releases than actual private networks,” he quipped.
“I’m still a skeptic on private networks because big businesses don’t want to be telcos. That’s what it comes down to. They don’t have the time to be telcos,” Entner told Fierce. “They try to outsource all IT functions. Why would they insource the most complicated thing when they have no expertise in it? Doesn’t make sense. Some companies, yes. The vast majority, no.”
Verizon isn’t saying exactly when the Nvidia-powered solution will be commercially available. It described it as a “proof of concept” and said Verizon engineers will begin demonstrations in February.
Until then, try to hold onto your seats.