Your browser is not supported. Please update it.

18 May 2020

Before Boost deal even finalized, Dish and TMO are in race to the 5G core

The expanded T-Mobile and Dish Network are still hammering out the final details of their deal, which will see Sprint’s Boost prepaid business sold to the satellite TV operator for an initial $1.2bn. But while this arrangement helped secure regulatory approval for T-Mobile USA’s merger with Sprint, once it is activated, the companies will become competitors – even though, until it rolls out its own network at scale, Dish’s main coverage will come from its seven-year MVNO arrangement with TMO.

The new TMO has a significant advantage over Dish in terms of established networks, services and user bases, but it has to go through a difficult integration process (see separate item). Meanwhile Dish has spent years collecting spectrum but failing to build out more than a sparse NB-IoT network (to satisfy an FCC mandate). But now it is brushing aside speculation that its real agenda is to sell on its spectrum, perhaps to midband-starved Verizon; or that its ailing pay-TV business means it cannot afford to finance a build-out.

On the contrary, its executives say, it will harness the benefits of being a mobile greenfield with no legacy equipment or customers, and will disrupt the US market – including its MVNO host – by deploying a fully cloud-based network with very low total cost of ownership (TCO) compared to others. That will give it the elasticity to undercut competitors on price where necessary.

This is a formula seen in France (Free), India (Reliance Jio) and Japan (Rakuten), although the last of these is Dish’s real role model, because of the cloud-native nature of its new network. Adopting a fully virtualized architecture, and O-RAN interfaces to encourage a competitive supply chain, will enable Dish to build its network for about $10bn in capex, it claims – a figure that some analysts and rivals have dismissed.

But Dish will not have the cloud-native upper hand, necessarily. TMO also says it will deploy a cloud-based 5G core this year and will adopt a heavily virtualized approach to reduce costs. Both companies aim to use the new generation of cloud-native, containerized technologies, to steal a march on AT&T, whose early moves into cloudification and software-defined networks mean it will be stuck with some older generation, virtual machine technologies for years to come.

Not that cloud-native is a magic wand. For a greenfield, it can bring down capex, especially if – as Rakuten did and Dish plans to do – it is introduced alongside open multivendor platforms with competitive ecosystems. But in these early days, cloud-native end-to-end networks require a major investment in people, integration resources and organizational change to address all the challenges of this difficult deployment.

Of course, such challenges will be even greater for TMO, which will have to migrate not one, but two, existing networks to the cloud over the coming years, picking up the threads of very different existing virtualization approaches, and supporting legacy and non-cloud elements for years ahead.

In the past week, both operators have spoken aggressively about their 5G core network plans, though TMO – predictably, given it is an established MNO – scored higher on actual detail. It set out several milestones on its roadmap to deploy 5G New Radio Standalone (NR SA), which requires a 5G core, rather than using an enhanced LTE core like NR Non-Standalone. The company said it already has a multivendor 5G production core working on its commercial 5G network, paving the way for the launch of NR SA later this year.

Unlike Dish, it is sticking with traditional suppliers for now – named so far are Cisco, Ericsson and Nokia, plus chip vendors MediaTek and Qualcomm, and interestingly, as the handset partner for initial tests, a Chinese firm, OnePlus.

Its tests of the new core included several firsts for 5G SA working with a commercial smartphone (from OnePlus) on a production, multivendor core.

These included:

  • SA and NSA 5G devices operating with active sessions at the same time in the same cell.
  • An SA data session between commercial modems from two suppliers, Qualcomm and MediaTek.
  • An SA 5G voice call in 600 MHz, using Evolved Packet System (EPS) fallback to VoLTE, and working with Cisco, Ericsson, MediaTek and Nokia. That will enable high quality voice services using VoLTE in the SA architecture while the industry develops full 5G voice over new radio (VoNR) technology.
  • A video over new radio (ViNR) test call on a production network with Cisco, Ericsson, Nokia and Qualcomm.

The MNO’s president of technology, Neville Ray, said in a statement: “Standalone 5G, paired with the broad and deep network we’re building by combining the assets of T-Mobile and Sprint, will accelerate 5G adoption and services and transform wireless!”

So far, TMO has the broadest 5G coverage in the USA, because it has started in the low band 600 MHz spectrum, rather than midband or millimeter wave options, with its NSA network reaching 200m people and more than 5,000 cities and towns. It will now be able to add capacity using Sprint’s plentiful 2.5 GHz spectrum, plus millimeter wave assets, as well as supporting new services via the 5G core.

AT&T says it will begin deploying 5G NR SA this year, and Verizon aims to roll out a standalone core in 2020 or 2021.

It is unclear whether Dish will be going live in the same timeframe, but its greenfield status will enable it to deploy NR SA from scratch, and it looks likely to have the most open, multivendor architecture for its cloud-native core and its vRAN.

On its earnings call, Dish’s chairman Charlie Ergen said the timing for the core deployment would partly depend on how far TMO cooperated but he set a target of one year for Dish to have transitioned from a prepaid, MVNO-based mobile business to one weighted towards network owner economics and a postpaid (higher value) user base. Of course, the MVNO deal will remain an important element of the Dish mobile model -with its favorable (though undisclosed) terms, it will last for seven years, and there is still a strong logic behind a deeper deal between Dish and TMO, involving long term network hosting, network sharing or ‘heavy MVNO’ arrangements, perhaps.

Meanwhile, the closing of the Boost deal and transfer of the users is likely to happen at some point during June.

Tom Cullen, Dish’s EVP of corporate development, set out a roadmap for Dish’s journey towards being a full network operator. The integration of the Boost services on TMO’s new network and with Dish’s operational systems, is the first step.

Next, Dish aims to finalize its section of vendors for the 5G RAN and core; then it will announce a detailed 5G deployment plan. The aim is to have a 5G Standalone platform on trial in one market by the end of the year.

Cullen said RF planning had begun and Dish was “far down the road” in tower discussions, though decisions about decommissioning Sprint sites can only happen when the acquisition of Boost closes.

On the vendor side, it has already issued three requests for proposal (RFPs), covering equipment, software and roll-out services, and announced one vendor, Mavenir. It has halted work on its fledgling NB-IoT network, which was being built with Ericsson in order to meet FCC build-out requirements for some of Dish’s spectrum. This will be transitioned to the 5G platform. This will result in a $253m non-cash impairment charge from certain assets, including satellites, that won’t be used for the 5G network.

Mavenir will be one of the suppliers of cloud-native open RAN software, and its CEO, Pardeep Kohli, has hinted that he is confident of getting some of the core networks awards as well. Dish is aiming for a multivendor, open supply chain like that of Rakuten in Japan (whose main supplier of vRAN software was another US start-up, Altiostar). Ergen said Mavenir was the first vendor to comply with all Dish’s technical requirements, but “by no means will they be the only vendor that we use” – he added that he expected to announce more multiyear contracts during the third quarter and to have a detailed update on the 5G network plan in time for the next earnings call.

The announcement of the Mavenir deal sent out strong signals that Dish would stick by its promises to use open platforms to reduce its spending and introduce new economics to US wireless networks; and to have a more US-centric supply chain than most MNOs, as the development of open RAN and core ecosystems has enabled several US start-ups to gain credibility.

Some of those US independents were involved in the Rakuten cloud-native network which Dish sees as its template – as well as Altiostar, Airspan provided radios and small cells. And the move to the cloud brought in giant US firms from the IT sector, such as Cisco, HPE, IBM/Red Hat and Intel.

“The open and intelligent architecture of our greenfield network will give us the ability to source a diverse technology ecosystem, including US-based solution providers,” said Marc Rouanne, Dish’s chief network officer, announcing the first contract award. “Mavenir will help us lay the foundation for an innovative software-defined network with the flexibility, intelligence and scalability to deliver applications that will redefine the US wireless industry.”

It is not clear whether Mavenir will be the only vRAN software provider, as Altiostar is at Rakuten, or which companies will supply other key elements like the radios, the servers to support the VNFs, or the core. But the fact that the first announced vendor is a US start-up, and a supporter of cloud-native open architectures, ticks all the right boxes in a USA which wants to be as advanced in its 5G platforms as Japan, while promoting a homegrown alternative to Chinese equipment makers.

Dish’s build-out will include between 30,000 and 50,000 cell sites with a mixture of fiber and microwave xHaul. Ergen has said that he could roll out a national 5G macro layer network for $10bn (excluding spectrum) by going cloud-native from day one; by taking advantage of Dish’s lack of legacy networks to maintain and integrate; and by harnessing the economics of open, programmable and automated platforms.

However, as with Rakuten, critics have dismissed the capex targets, pointing out that Verizon spends about $18bn and AT&T over $20bn each year (though not all for mobile). But Dish insists capital and operations costs really can be transformed by a combination of cloud economics and extensive use of shared or third party infrastructure. “Because we’ll be cloud-native in that network we can just do things faster, better cheaper,” Ergen said in November. “We don’t have the legacy of 2G, 3G and 4G. The cost structure goes down across the board. The vast majority of capex for the incumbents is not for 5G, it’s to maintain the legacy. We don’t have that cost going forward.”

Dish claims it can raise the money for its build-out, despite the impact of Covid-19 and the decline of its pay-TV business. TV and Boost revenue will help, and it has raised $1bn in equity already, but it may have to pull back some expenditure in 2020. It is currently still forecasting wireless capex this year of $250m to $500m, but expects to be at the lower end of that range because of the impact of the pandemic.

Ergen noted that Dish has paid down about $2.5bn in debt in the past 13 months. “We’re not standing still … the funding part is not the part that’s keeping us up at night at this point,” he said.

Another factor in how far Dish can meet its low cost targets is partnership. There have been persistent rumors that the firm will attract co-investment in its build-out, with Amazon or even Google the prime candidates (though Microsoft, given its acquisitions of Affirmed and Metaswitch, is also looking keen to have greater direct control of a full network platform to support enterprise 5G/cloud services – see separate item). Ergen would only say there has been plenty of third party interest, especially in the past six months, from “people who know what a modern network should look like”.

And at least one Wall Street analyst firm, New Street Research, is positive about the plans and wrote in a client note: “Dish truly has the potential to disrupt the US wireless market”, calculating that its 5G network could deliver a cost per gigabyte a full 3,550% lower than that of wireless incumbents.

In the first phase, Dish aims to keep the network asset-light, saying that, since it would not have large numbers of customers on day one, as a new entrant, small cells will be a second-phase priority. The $10bn capex bill for 5G is for a “total macro layer” that will provide Dish with nationwide coverage “well beyond the FCC requirements”. It will look to minimize equipment on macro towers – and therefore costs – by moving a substantial proportion of baseband processing to the cloud and stealing a march on rivals by creating a cloud-centric, highly virtualized platform that can behave in an agile manner.

The virtualized nature of the network enables Dish to support open platforms and shake up its supply chain, and it seems to be taking on the ‘America-first’ mood of the current USA, whose government has been calling for a homegrown ecosystem to be funded and supported as an alternative to Chinese equipment. “We do think it brings into play American vendors that have been left out of the traditional networks,” Ergen said in an interview last year. “So we know we’ll have a much more American-centric set of vendors than the traditional incumbents.” He offered some names that would be considered – all of them already part of Rakuten’s cloud-native 4G/5G network plan. They were Cisco, Intel, Red Hat and Altiostar.

Ergen talked up the most prominent open network initiative, O-RAN, whose most impactful work so far is on an open fronthaul interface between the baseband unit and radio unit in a disaggregated virtual RAN. “We now know O-RAN is real; it’s not pie-in-the-sky,” Ergen said, referring to a comment by some US politicians, that open RANs were for the future. He added: “Huawei is really good. The Chinese equipment is really good. It’s best in class. We shouldn’t just try to be as good as the Chinese manufactures – we should be better.”