A new partnership between InfiniG and Nokia upgrades in-building neutral-host coverage to ‘carrier-class’ standards while quietly positioning enterprise networks for AI-driven RAN innovation – which ties to both the Finnish firm’s ‘super-cycle’ AI vision and its strange enterprise 5G strategy.
In sum – what to know:
Neutral host credibility – InfiniG’s adoption of Nokia’s AirScale RAN elevates its CBRS-based MOCN offer from enterprise-grade to carrier-class, targeting large venues with future-proofed 4G/5G infrastructure.
AI-RAN in the enterprise – The deal aligns InfiniG with Nokia’s Nvidia-backed AI-RAN roadmap, opening a path to on-prem GPU-powered inferencing and automation – the first time Nokia has addressed the space.
Shifting CBRS dynamics – US operator support for indoor MOCN is uneven, but large-sized enterprise-funded deployments are driving new pragmatism – not just from AT&T, but also from T-Mobile and Verizon.
Here’s the full discussion about neutral-host specialist InfiniG’s new RAN deal with Nokia to offer a higher-grade in-building cellular coverage to enterprises on CBRS spectrum in the US, with a view also to make a home for the the vendor’s AI-RAN pitch inside enterprise premises. It is a notable channel tie-up for InfiniG in its own right, but it has intriguing relevance for a whole bunch of other industry narratives besides: about the perceived MOCN / CBRS jeopardy in the US, about Nokia’s own strategy in the enterprise space, including the sale of its campus (DAC/MPW) unit, and also about its grander plans with Nvidia.
As such, the story provided a minor footnote appearance in a two-part review of Nokia’s recent fortunes in the public and private 5G markets last week (versus Ericsson; see here and here). But this is a somewhat straighter treatment, focused on InfiniG’s particular vision in the market, based on an interview with the firm’s chief executive, Joel Lindholm. Which doesn’t mean all of these other talking points are not here, of course, but they are rather bubbling under the surface, to be discerned separately. For his part, Lindholm is candid, and conversation skips around these stories all the same.
But let’s deal with the news: California-based InfiniG, selling a cloud-managed neutral-host service with a multi-operator core network (MOCN) gateway for in-building enterprise coverage in CBRS spectrum, has added Nokia’s “carrier-class” AirScale radio infrastructure to its portfolio. It is the “first in the industry” to do this, it says – specifically to off Nokia radio access network (RAN) infrastructure for indoor neutral-host coverage in shared CBRS spectrum. The big point for InfiniG is the Nokia RAN addition gives it new “carrier-grade” credibility.
Until now, InfiniG has primarily sold Airspan RAN systems with its MOCN system, classified as “enterprise-grade” by comparison. Nokia makes radios for big public networks, and small-but-serious private networks – is the message; it is the same distributed RAN (D-RAN) architecture that is “deployed across hundreds of mobile operator and private network environments worldwide”, says InfiniG. It will sell the Nokia units as ready-to-go 5G-capable, so enterprises are not required – when US operators (MNOs) get their 5G SA integration work done – to swap out 4G Airspan units.
More than this, InfiniG has jumped on Nokia’s developing AI-RAN narrative, which has followed Nvidia’s $1 billion investment in the Finnish firm – about placing graphics processing units (GPUs) in or near RAN equipment, ostensibly in public 5G macro networks, plus about driving new 5G-Advanced and 6G network capabilities. For InfiniG, the arrangement with Nokia means it will, in time, be able to take advantage of new AI efficiencies in RAN operations, passed to enterprises, and also offers an on-prem GPU resource for inference workloads.
Lindholm discusses all of this in the Q&A below, including the mechanics (mostly in concert with third-party private networks) and timelines (six months for pilots, longer for commercial rollout) for such a vision. A press note says: “InfiniG’s network design will leverage Nokia’s roadmap and the evolution toward AI-RAN. This integration will enable InfiniG and its enterprise partners to facilitate AI inferencing and intelligent automation directly within the RAN. These capabilities will facilitate real-time analytics, deep operational insights, and autonomous network optimization.”
Lindholm says more besides, including about the state of CBRS-based indoor MOCN support in the US from the big three mobile operators. AT&T’s support is unflinching; T-Mobile’s support, as memory serves, has recently wavered, as it has sought to take control of its local SA rollouts; Verizon’s opposition has been total. But T-Mobile has remained pragmatic, and Verizon is increasingly so; money talks, implies Lindholm, and InfiniG is chasing bigger enterprises, harder-to-refuse, with its new Nokia proposition. Plus, the indoor MOCN model is enterprise-funded.
The statement says: “As mobile operators continue to prioritize macro network investments, many enterprises and venues are facing growing challenges in ensuring reliable indoor mobile coverage. InfiniG’s platform addresses this gap by enabling enterprise-funded, carrier-grade mobile coverage that seamlessly connects to all major US mobile operators.” Its solution is easier and faster to deploy than DAS, and better anyway, suggests Lindholm; it can also slot into aging or end-of-life DAS infrastructure, he says. Best of all, it is built to last – for 5G, AI, and whatever.
But Lindhom tells it best; all the quotes below are from him.
…
What’s new with this? The press note talks about ‘carrier-grade coverage’ and a pathway to 5G and AI-RAN, but there’s lots in there that has always been part of the InfiniG proposition (‘less complexity’, ‘faster’ deployments, ‘investment protection’). What changes with Nokia?
“Yes, in a way, not much changes – except that we’re now putting carrier-grade infrastructure into enterprises. Which is important. The point is that, if you’re a million square-foot hospital, say, bidding on infrastructure, you’ve now got integrated Nokia hardware – which would’ve been a challenge for an enterprise to figure out. It does both LTE (4G) and 5G, simultaneously, at a cost point that’s manageable – where we’ve integrated and negotiated so medium and bigger buildings can get this without a price premium at all. Because these are superior radios – the same as the NDAC team has been selling with private networks, the same that are in just about every mobile operator.
“But we are taking on that lift for the enterprise. We integrate all the hardware components and management components, and we make it an automated process – so we reduce the barrier to entry; so this carrier-grade solution is deployable by enterprises. And the hospital, call it, gets highly-reliable connectivity with a path to 5G. We have avoided the 6G story, but it will go there too – but 6G is also too futuristic for this market. But it will last 20 years – kind of thing. It is future-proofed. There’s no gray area. It’s rock solid.”
Explain the 5G upgrade path for a MOCN-based neutral-host set-up in the US – if the enterprise RAN is already 5G-enabled?
“So the US carriers are still on LTE – as certified for that [MOCN] infrastructure. This year, early next year, we’ll move them to 5G, and the Nokia infrastructure is ready for that… The radios are capable, the gateway is capable, but the carriers have to certify it – because [the traffic goes] to their SA core, and they need SA devices, capable of voice over NR; which are moving targets for some of the MNOs. T-Mobile has done a good job to push ahead, but the others are catching up. But the infrastructure itself is software-upgradable.
“So it is an investment protection story for the enterprise, and a pathway to the future – while they are trying to define what that future looks like. And we know, based on the Nvidia investment (in Nokia), that it is going this way – an Nvidia blade goes into their BBU, and the story goes there. The point is they’re not spending money on stuff that is going to be obsolete in a few years. It will last 20 years – which is a huge deal in our world. For an enterprise spending its precious dollars, that is an important thing.
“It is an investment protection story as well because we will feed these radios into end-of-life DAS. We are targeting greenfield customers, but we also offer an easy DAS upgrade path. The carriers have abandoned DAS; they’ve said, ‘You’re on your own Mister Customer’. But customers have coax in their buildings – where it is super disruptive to redo the cabling. So we will rip out the headend from the DAS and stick our network in, using the same cabling. So we’re taking them from a single carrier DAS, maybe, to a multi-carrier neutral-host system – pretty easily, just by plugging in radios. So there’s a massive upgrade capability.”
And what’s the MNO’s latest position on MOCN in CBRS spectrum in the US – because there has been a notable push-back from T-Mobile and Verizon?
“AT&T is on all of our networks, and there’s not a question asked. AT&T has it all figured out. T-Mobile, despite its bluster nine months ago, is pragmatic. To say it’s pivoted back would be an overstatement, but there are chinks in the armor, as it were. If there’s a business case, then, yeah, it’s enough justification to do something. Which means we’re not chasing little schools; we’re chasing bigger environments. Which is clearly applicable for this [work with Nokia]. And Verizon, given everything, is also increasingly pragmatic.
“The thing is that our neutral-host model using CBRS is good for all of them. They get coverage paid for by the enterprise, their cost of provisioning is minimal, there are no lengthy legal agreements or complicated spectrum licensing, and they get a consistent view of the KPIs – and no risk of interference with their local macro network. So if the enterprise customer wants to pay for coverage, then they should join – to satisfy their subscribers.”
So if InfiniG writes a piece of business with an enterprise tomorrow, based on a Nokia RAN system, would you expect to have all three carriers connected through your MOCN gateway?
“If it is a small building, AT&T will show up, and the other guys might not care because it’s a small building. But when it’s a big and important customer, then, yes, all three of them will show up. Our focus is with these large enterprises. We work with a lot of household names; we are taking major deals from our competitors; we have got a couple national retailers on board; we’ve got some military stuff. These are all customers that the MNOs care about – and they don’t want to be the only ones in there without coverage.”
Is InfiniG the biggest neutral host provider in the US?
“We are probably the biggest, but we are also the only pure-play neutral host provider. All the others are either 80 percent or 90 percent about private networks, with this neutral host thing on the side, or else doing a bit of everything – Wi-Fi, DAS, private networks. We are the only one that is just focused on neutral-host, and on doing it really well.”
And what’s the private-networks angle here – because yours is offered like a gateway infrastructure, essentially, for a private core network to piggyback on? Is that right?
“Correct. We work with anybody’s EPC – a Druid EPC, a Pente EPC, a Cisco EPC; whoever’s. We don’t really care. Our focus is the neutral-host piece. But we have an ecosystem of private network partners via our InfiniG Connect model. So we put in the neutral host, and if that hospital – again – wants to pick this [core network] vendor or that vendor for a private network, then no problem – we will point the radios right to them, and they have their private network. It is a lot easier than putting a square peg in a round hole.
“Because if you break down the private network into components – the radios, the core, the SIM cards in the devices, as the primary things – then we probably take 70 percent of the complexity out of the equation. Because the radios are paid and managed for the neutral-host network anyway. So we are lowering the barrier for private-network adoption, as well – because the enterprise only has to worry about the core network and some SIM cards. And that’s not that complicated of a thing to do – a piece of software on a VM somewhere, which you add devices to it.
“But there is also a straddling of the market. Not all private networks are created equal. There’s Wi-Fi, and then there’s a-little-better-than-Wi-Fi, for basic PTT or barcode scanning in warehouses, which is a lot of what gets written about; and then there’s the five-nines (99.999-percent reliability) Industry 4.0 stuff in airports and shipping yards, which is a whole-other planet – which is, actually, what Nokia has been selling. There are a thousand enterprises on Nokia, and on NDAC specifically, and those aren’t going anywhere. Because it’s operational, based on an ROI.
“We’re not providing [the RAN] infrastructure for them, but we are connecting to them in our cloud – no problem, within weeks. But as you move down [the criticality/performance scale], a hospital is a little bit more of a jump ball, where 70 percent of its requirements will be served by neutral-host infrastructure, typically, and 30 percent will require a private network – at some point. Same for some of these light warehouses, whether they want their phones to work for safety purposes or they want the barcode scanners.”
So explain this AI-RAN story with Nokia in the enterprise?
“Well, putting AI at the cell tower – what are you going to do? Better spectrum management and better quality of service. Great; but I’m still not paying more than 40 bucks a month. So that’s just table stakes for the MNOs to be better at their normal job. And that tower can’t reach inside the building, as it is today. You can’t run a robot on that. You need the network inside the building. So we are going to show up with the infrastructure for the neutral-host bits – which is the me-too table stakes for the enterprise – and we are going to move them to 5G, with infrastructure that is not only optimized to leverage the AI RAN stuff, but also has [a slot for] a GPU to do inferencing.
“So that robot that’s running down the hallway into a wall, right now, will say, ‘Oh crap, there’s a wall ahead’, and take new instructions – because there is [on-prem RAN] infrastructure for inferencing, at performance and scale. The enterprise is the one that needs the applications that can take advantage of AI – and it has to be inside the building for that kind of performance. Even if the signal gets into the building, it is one thing if you want a phone to work – 50 milliseconds, whatever; it is another thing to drop that down to single-digit millisecond inferencing. You can’t do that from the macro network.”
But those kinds of cases, like with the robot, will be attached to a private network, and not a neutral host, surely? So in the end, is this a private-networks play – like a Trojan horse via a neutral host gateway? Are there IT-style use cases in an office building that will justify a GPU on a neutral-host RAN, or will most of GPU requirements be attached to a private core that is paired with your neutral-host system?
“Exactly. And with Infinity Connect, you can pick that core; when that robot needs to come to life, you can choose whichever vendor. Some of these buildings will take some open source thing on GitHub and do their own internal core – because it’s customized and proprietary for what they want to do, and it drops their costs down to nothing. We’re saying, ‘Fine; you have the RAN infrastructure, already, and we can point it your way, and you can stick an [AI] blade in there whenever you want’. They’ve got the flexibility to go do that.”
How soon do you expect this on-prem AI-RAN story to land with enterprises? Is it something that will come around next month, next year?
“Customers buying a neutral host system (in the first instance), and not a private network, will need to take a journey, and it is probably a year from now – six months, maybe, until we have a pilot of something. But they will play with it, and there’s a roadmap, quote/unquote, of Nokia/Nvidia infrastructure that still needs to come together. It takes time for all that to evolve.”
So it’s not like an urgent message to the sales team to sell AI-RAN to customers?
“No. We definitely don’t want to say, ‘Buy this today and buy AI tomorrow’. That is not it. The message is to buy infrastructure that is carrier-grade, and which will last – which will take you on a path to 5G and AI. And the fact it is with Nokia and Nvidia makes that a very believable path. I mean, it is much better signposted than by anyone else, right? With AI RAN, there are benefits for carrier efficiencies, but the same efficiencies will make enterprise RAN better too. And the application of inferencing needs to happen inside the enterprise – on an enterprise network, inside the four walls.”
