Intelligence Brief: What are the top tips for TIP?

First off, a disclaimer. It’s an important one (not like the warning that your McDonald’s coffee may be hot).

Last week, I spent some time at Facebook’s Telecom Infra Project (TIP) Summit and put together a blog post titled The Problem with Facebook’s Telecom Infra Project. If you haven’t seen it yet, you can check it out here [1].

If you did read it, you probably wondered why it took me so long to get to the actual problems I see TIP facing. The bulk of the blog was focused on a handful of more positive messages: that initial concerns about industry support for TIP were quickly put to rest; that an initially vague focus quickly sharpened; that the core value proposition of connecting the unconnected is a real issue which aligns with our own work here at GSMA Intelligence.

Was it a love letter? Not quite. But was it clear to anyone reading beyond the headline that I really like TIP, what it hopes to accomplish and the progress it is making? And within that context, the point of highlighting the problem referenced in the headline was less about disparaging TIP than it was flagging a challenge that needs to be focused on to create success.

Of course, with that as a set-up, it only makes sense to talk about a few other problems: if we want to successfully navigate the challenges, then we need to recognise them. And to be completely honest, the one flagged last week might not be the most pressing.

Broad versus deep
When I highlighted that OpenRAN, mmWave and optical transport programmes were being complemented by new TIP work streams, the point was to recognise forward progress. But there’s a potential downside to this progress: lack of focus.

There is no doubt that vRAN fronthaul; efficient power technologies; artificial intelligence and machine learning; end-to-end network slicing; solution integration; and edge networking are all critically important telecoms technologies. And there is no doubt they can all benefit from the TIP treatment.

What is less certain is whether TIP can focus on all of these simultaneously and show progress with scaling attractive solutions which are more than one-off demos or trials. As the scope of any company or organisation expands, there is always the risk of spreading its energies too thin. TIP is not immune to this.

Network supplier expertise versus inertia
If TIP is successful, it will bring new technologies into operator networks. And with those technologies will come new vendors. Hence, we saw the best performing vendors of the OpenRAN RFI including a broad set of players extending beyond the traditional end-to-end network players: Mavenir; Parallel Wireless; Altiostar; Fairwaves; Radisys; BaiCells; NEC; ASOCS; Phluido; Comba; Dali Wireless; and VANU (in no particular order).

From an innovation and pricing perspective, competition is a good thing. But, for the most part, today’s wireless networks have been built by a handful of network vendors. In many cases this includes the deployment and management of network gear. And, over the years, this gear has come to include technologies which can be upgraded to support new generations of wireless access. It all adds up to an inertia that favours working with incumbent vendors: they are easier to work with because they know how the existing networks run and how to keep them running.

Commercial wrangling versus open commitment
The inertia I just mentioned stretches beyond network operations and deployment. It is a core dynamic in how networks are purchased. Setting up new commercial arrangements with small, relatively unproven vendors can be risky: the type of risk procurement departments loathe. And, while a new set of vendors might promise a focus on innovation and openness, larger vendors are often in a position to cut costs in order to avoid losing out on business. The question for an operator, then, is how much that focus on openness is really worth? Can an investment in the future benefits of openness and supplier diversity be justified, especially when service pricing and margins are under pressure?

None of this is news to the people who run TIP, including its vendor and operator members. They all get it. And that helps to explain a vision Telefonica outlined of a new telecom infrastructure value chain which includes hardware vendors, software vendors, node integrators, network integrators, support vendors, and service vendors. Breaking down the end-to-end value chain helps to break some of the inertia (operational or commercial) that accrues to end-to-end players.

It’s a smart move. It’s necessary. But it’s not sufficient. Maintaining progress will require a few other things.

TIP will need to make a few bets. Rather than spreading its energies around, it will need to focus on a few technologies (like it did with optical transport and OpenRAN) where it can show progress and gather momentum. These bets need to touch incremental network opportunities, places where it is easier for an operator to work with new vendors because incumbency isn’t an issue yet.

Do you need an example? Consider the new TIP focus on edge networking, including the Edge Application Developer Project Group. While edge promises latency improvements, transport efficiencies and new operator revenues beyond connectivity, it’s still a very nascent space. It’s still very much anyone’s game.

The most important requirement, however, stretches beyond TIP itself. If operators are committed to open networking innovations, it must be reflected in their purchasing. They can’t always be swayed by incumbent vendors offering the same solutions at a discount.

This means new procurement thinking. It requires faith in the long-term benefits of open network solutions, even if reliability or standards support doesn’t compare with traditional network vendors. It may require co-investment or involve working with incumbent vendors, driving them further and further towards openness.

I understand that this is a lot to ask but change won’t come from building, and buying, networks the same way they have for years. Trials and demos are a good start, but need to be followed by deeper actions and commitments.

– Peter Jarich, head of GSMA Intelligence

The editorial views expressed in this article are solely those of the author and will not necessarily reflect the views of the GSMA, its Members or Associate Members.


Intelligence Brief: What lessons do 5G auctions offer?

Everyone is getting excited about the brilliant future that 5G offers, but let’s face it: without a lot more spectrum dedicated to mobile services, dreams of faster speeds and greater capacity will just remain dreams.

But only last week the world saw two more small steps to fulfil those 5G dreams with two 3.5GHz auctions which offered spectrum discussed for 5G services coming to an end. And, with them, came a very noticeable disparity in final results between the two.

While three Finnish incumbent operators went just slightly above the starting prices, Italian bidders splashed out [1] on their investment, paying almost 11-times more than the reserve price (and 56 times more than Finnish bidders) for nearly half the amount of spectrum. With a few 3.5GHz auctions completed in the past 18months and several more on the way, it’s a good moment to look at what happened so far and anticipate what may happen in the future.

Of course, there are other bands important for 5G deployment as well, like mmWave bands (especially for densely populated areas). However, only Italy and South Korea have organised 5G-specific mmWave tenders so far (with operators paying only slightly more than starting bids), making 3.5GHz the most instructive as far as 5G goes.


Less (spectrum offered) is more (in prices paid)
Across these recent auctions, Finland offered the most spectrum: 390MHz. That is almost twice as much as offerings in Spain and Italy, with the UK auctioning even less (150 MHz). This was reflected in final prices: in terms of €/MHz/Pop (PPP adjusted), Spain paid almost twice; the UK four-times; and Italy nearly 13-times as much as Finland did.

It’s noteworthy that none of the recent 3.5GHz auctions ended with unsold spectrum.

Less spectrum offered means scarcity, but this does not necessarily mean it was artificially imposed by national regulators. In some countries, spectrum has been reserved to vertical industries or there already have been incumbents in the band, as licences have been assigned to fixed wireless access (FWA) or WiMax services (Spain, Italy, and the UK). Regardless, it creates the pressure for bidders to fight fiercely for any piece of spectrum left, which naturally pushes the prices up. And MNOs will need a lot of spectrum to deploy 5G properly; at least 80MHz to 100MHz of contiguous spectrum per operator in the mid-band, according to ITU.

Offering a sufficient supply of spectrum in the 3.5GHz band will take some of that pressure from the bidders and leave them with more resources to invest in actual 5G deployment.

Noteworthy, some mobile operators have been able to get access to the 3.5GHz band through the secondary market rather than government auctions, acquiring spectrum licences from incumbent FWA providers: Masmovil acquired Neutra and Eurona in Spain; Three UK purchased UK Broadband; and Fastweb bought spectrum licences from Tiscali in Italy. This may be a way of gaining access to some of the much-needed spectrum in the band without having to wait for government action. However, acquiring the spectrum piecemeal may mean that some band pre-planning will have to occur.

Lot size matters…a lot
Countries assigning spectrum in 3.5GHz have shown two different approaches so far when it comes to block sizing: offering a small amount of larger blocks (Finland, Italy, Czech Republic); or offering large amounts of small 5MHz to 10MHz blocks (South Korea, UK, Spain, Ireland).

Large contiguous blocks of spectrum ensure the higher quality of future 5G services. However, offering lots that are too big during tenders may also cause spectrum scarcity and lead to higher prices if the number of potential applicants exceeds the number of blocks. This can be illustrated by the recent example of Italy, where only two lots of 80MHz and two smaller lots of 20MHz were offered, forcing operators to fight for the bigger lots. Appropriately sized lots may allow operators more bidding flexibility and ensure more participants will end up with winning spectrum, and that more end customers will be offered 5G services.


5G hype is attracting new players
In Finnish and South Korean auctions, only incumbent operators participated and the price increase compared to starting bids was rather low (19 per cent and 13 per cent respectively), as applicants had a lot of spectrum to bid for (130MHz and 93MHz respectively). In other markets, however, participation of potential new entrants increased the competitiveness, which left the bidders with less spectrum available, on average, per bidder (typically only 30MHz to 40MHz in the UK, Czech Republic and Italy). This fuelled the struggle to secure as much spectrum as needed.

While allowing new entrants shouldn’t be avoided, their presence within the bidding process has to be taken into consideration, in terms of auction elements such as reserve prices, lot size or overall auction design, especially regarding three- or four-player markets.


Everyone is closely watching the forerunners
It’s no surprise that 5G spectrum auctions are garnering more interest and analysts are trying to use those results to predict the future pricing trends. While benchmarking to other markets should not be the only element considered in spectrum valuation, results from previous auctions are carefully studied by regulators in terms of setting their own reserve prices and the temptation to set them on a higher level may be strong.

However, every country is unique, with its own market forces and conditions to consider, so benchmarking at this stage, especially given the wide variety of outcomes, is risky. So far, 5G auctions have happened only in developed markets and the dynamics and conditions are very different from those in the developing world.

Designing auctions with sufficient spectrum offered, appropriately sized lots and reasonable reserve prices is important and also complicated: it’s possible that some countries may even turn away from auctions as the assignment mechanism for some of the 5G spectrum. The Hong Kong Secretary for Commerce and Economic Development suggested spectrum could be assigned to MNOs for free in order to ensure access to spectrum and result in boosting national competitiveness. Additionally, many countries have already assigned fixed wireless/WiMax licences in the 3.5GHz band, often to MNOs.

The question remains whether the operators would be able to convert their licences to tech neutrality and offer 5G services, without having to undergo a long and costly process of reapplying and competing for the same band they already use. The ultimate goal is putting spectrum to good use and, clearly, regulators are recognising that auctions are not the only approach for spectrum assignment in the 5G era.

– Robert Wyrzykowski, analyst – Mobile Spectrum, GSMA Intelligence

The editorial views expressed in this article are solely those of the author and will not necessarily reflect the views of the GSMA, its Members or Associate Members.


Intelligence Brief: The problem with Facebook’s Telecom Infra Project

This week, Facebook hosted its TIP Summit 2018 – the third time it has convened Telecom Infrastructure Project (TIP) members and other interested parties since the project’s kick-off in 2016.

When launched back at MWC 2016, the TIP buzz was undeniable. Just as Facebook was disrupting the businesses of so many operators, it was now promising to disrupt the telecom networks business with a focus on open networking technologies.

It was a direction in which the telecom infrastructure space was already moving and an extra push from a name like Facebook was more than welcome, especially since it was all in the name of bringing network costs down and connecting the unconnected.

What’s not to like?

Well, for starters, it wasn’t quite clear why vendors would get with the programme. Nokia was a founding member, but a plan to open up “what is a traditionally closed system” would obviously threaten the business of anyone currently selling into that closed system.

Of course, when queried about their support for network innovation, vendors are fond of saying that they develop and build what their customers are willing to pay for. Would it be safe to assume, then, that operators which have often been slow to change the way they do business would be slow to ask their supplier to follow suit? Or that, if these operators did ask vendors to innovate in the name of openness, they might not actually buy the resulting solutions – especially if they were less reliable or more expensive in the near-term?

And yet, the momentum behind TIP has continued to build. Operators see an opportunity to drive network innovation. Vendors (especially non-incumbents) see an opportunity to break into the network business formerly closed off to them. Facebook sees an opportunity to build a tighter relationship with the service providers it relies on for connectivity, while driving the altruistic aim of increasing emerging market Internet usage.

Growing support
A handful of members at launch became “more than 300” by the first TIP Summit and “over 500 member organisations” today. An early, somewhat vague, focus on “access, backhaul, core and management” morphed into concrete work around OpenRAN, mmWave and open optical technologies. These, in turn, were joined by additional work streams around: vRAN fronthaul, power and connectivity, Artificial Intelligence and Machine Learning, end-to-end network slicing, and solution integration.

This week, we saw the announcement of further work around system disaggregation (optical and cell site gateway), the results of vRAN fronthaul trials, OpenCellular trials, and the OpenRAN RFI launched in June by Vodafone and Telefonica [1]. By all accounts, the results have been impressive.

Job done, right? Maybe not.

The problem comes back to the core TIP value proposition of connecting the unconnected. It was the proclaimed focus of the first day of the 2018 TIP Summit. It was called out as the whole point of TIP in a fireside chat between Facebook’s Jay Parikh (VP of Engineering) and Ina Fried of Axios [2]. And, Facebook’s motives aside, it’s a real issue. Just take a look at our Mobile Connectivity Index (overview below).


In some regions (like Sub-Saharan Africa), nearly 40 per cent of the population isn’t covered by a mobile network. And even where network coverage tops 80 per cent across other regions, the percentage of people who are covered but not connected – thanks to cost, content availability or consumer readiness – is often more than 40 per cent.

There’s no doubt connectivity matters. Whether or not Facebook and TIP can change the connectivity landscape is another matter.

Sure, bringing down network infrastructure costs could make it easier to roll out networks to the uncovered. But conflating network infrastructure with networks is a mistake. Networks require siting infrastructure. They require spectrum. They require people to keep them running. TIP’s ability to impact those things is limited.

If there’s any doubt about the magnitude of the costs that go beyond network infrastructure, consider what operators spend on opex vs capex. We’re not talking an even split. Or even two-to-one. In many cases, it’s closer to five-to-one, or even more skewed towards opex.

TIP may be able to drive network infrastructure scale and cost efficiencies, but that’s only one part of the connectivity equation.

Ultimately, this is why it’s encouraging to see TIP messaging around rural connectivity accompanied by progress on innovations like network slicing, edge networking, 5G NR and other technologies that can be deployed everywhere. Not because rural connectivity isn’t important, but because the ability of TIP innovation to drive success may always be limited by dynamics beyond technology.

– Peter Jarich, Head of GSMA Intelligence

The editorial views expressed in this article are solely those of the author and will not necessarily reflect the views of the GSMA, its Members or Associate Members.