The Sci-Fi Vision of Telecom Operators

I have had a few conversations last week about the future of Telecom Operators. It started when Benedict Evans trolling Twitter with this comment:

Benedict Evans (@BenedictEvans)
The global SMS system does around 20bn messages a day. WhatsApp is now doing 42bn. With 57 engineers.

Of course, I am joking, he was not really trolling. He was making the point that modern software engineering is so advanced that it vastly improves the efficiency of communications services. I responded that WhatsApp rides on top of global telecom systems which employ considerably more than 57 engineers, and without those systems WhatsApp would not work. His response basically said that the carriers’ value proposition has been vastly eroded, to the point that they are becoming dumb pipes. (I am oversimplifying here, but the whole exchange was limited to 140 characters, so bear with us.)

Then I spoke to a friend at T-Mobile and another at Verizon, both of whom lamented that their companies have become alarmed at their declining hope for ever growing again.

All of this is common fare for the Tech Industry. The Valley wants to believe that carriers are just ‘dumb pipes’, with the real money to be made in software and applications. I have never been a big fan of the operators, nor do I believe that they can really be home to much innovation. Nonetheless, I think it is a HUGE mistake to overlook the importance of their role in modern life. The plumbing beneath all of the Internet’s ‘Tubes’ is essential, but too easy to ignore.

All of these exchanges took place under the shadow of a bunch of carriers reporting bleak earnings numbers. Sprint also announced a major retooling of their network, which analyst site Wireless Estimator called “Network Suicide”. That post is a good view on Sprint’s proposed changes, and what seems a fair assessment for how hard those changes will actually be. But it also introduced an old theme “if you were going to build a network from scratch today, you would build it very differently than everything out there.”

That last point really resonated when I spoke to a super-interesting start-up that is building telecom equipment. (No names now, but hopefully I can get the go-ahead to profile them soon.)

I think if you look at the trends in the hardware industry, it is pretty clear that big changes are coming to Telecom Operator networks. In that report on the Hardware Industry I noted that Enterprise networks and Operator networks are increasingly converging. Meaning that both camps are starting to use the same equipment and software. This was not true even a few years ago, but it is hard to escape this conclusion today.

When you boil it down, Telecom Operators today really have three competitive advantages:

  • They own the ‘last-mile’ either through wires in the ground or ownership of attractive spectrum.
  • They have a relationship with end-users and retail distribution.
  • They have preferential positions through regulatory conditions.

Of those three, only that last one looks secure to any degree.

This is all a long preamble to what I think the future of Telecom networks will look like. Today, operators are largely vertically integrated Institutions. They build vast networks, they have consumer relationships, they provide all kinds of services on top of their core competency of moving digital traffic around.

In my opinion, the changes in the underlying hardware is pushing the industry towards a greatly unbundled role for the operators.

The most obvious example of this is Ben Evans tweet. Text messaging and voice communication are the key services that operators once monetized, but the whole world is moving towards messaging apps for both of those. There is no obvious structural reason why operators should provide any of these services, or any service not directly tied to the underlying network. And thus the shift to pricing plans focussed on data consumption with cheap voice and free messaging tacked on.

A common defense of the operators’ position is that since they own the best spectrum and all that copper in the ground, they will continue to dominate the last mile of communication. But this one is much more threatened than people realize. Today, you can buy a box that provides a full telecom network in its entirety. The entire stack in one box, replacing literally several million dollars in equipment. Put up an access point, ‘borrow’ some spectrum, run some software in AWS and you too can be a wireless carrier.

Admittedly, that spectrum borrowing part is a little tricky, but is not impossible. Imagine a future in which one of the wireless operators struggles with its debt and legacy networks, to the point where it just gives up the network and ‘wholesales’ its spectrum. It is not as far-fetched as it sounds.

This is already happening in ‘un-licensed’ spectrum bands where there are literally thousands of wireless Internet Service Providers running gear from companies like Ubiquiti in the unlicensed spectrum band also used for Wi-Fi band.

I think that there are really three parts to a carrier’s vertical stack – the last mile, long-distance transportation and central offices all over the world to connect everything. Once upon a time, there were reasons for all those to be owned by one company, but those reasons are largely no longer relevant.

The long-haul portion of the network, transporting bits form city to city or across oceans, has already largely been unbundled. There are lots of (mostly) rural areas where there is only one fiber backbone, but transport from city to city is a competitive business with many companies playing.

The last mile, as I mentioned, is getting unbundled in some areas, but will likely remain a key competitive barrier, as spectrum is limited and most cities are unwilling to have streets torn up again to lay fiber. However, it is important to untangle this area a bit. If we were starting from scratch today, all we would want to build is a data connection. A big reason that operators emerged the way they did is to ensure interoperability. When you fly from your home to some other city, you want your cell phone to work in that other place. This federation of geography, allowing for pervasive connections is the true competitive barrier. So while I could theoretically launch a network in one city, building it out in hundreds would cost billions of dollars. Personally, I think someone will solve this problem, likely through ‘wholesaling’ some spectrum. There is no fundamental reason why this part of the network has to be connected to the other layers of the network.

The Central Office or local telephone exchange is also a tricky to reconcile. In practical terms the Central Office and the Last Mile are both part of metropolitan rings, networks that tie a city together. Again, if we started today, we would likely have one set of companies focused on this whole system, while others would handle the long-haul portions.  However, when viewed from a technological position, those Central Offices are really just data centers – big rooms filled with computing equipment. It is well-known that many Central Offices are largely empty today, hollowed out by the ever-shrinking nature of computers.

When it comes right down to it, the real trick is providing service that works across countries, federating metro networks. My vision of the ideal “Network on the Hill” would start here. If I were to start from scratch, I would look to build out a nationwide network, I would start by building out LTE wireless coverage in urban areas. This would be a ‘data-only’ network. The devices I would run on the network, would not have voice service. Customers could use Skype or Kik or WhatsApp or WeChat or any of the other apps that offer voice messaging. The problem is that this network is still expensive. I once estimated that building out a nationwide US LTE network would cost $10 billion to build. Prices have come down since then, and by avoiding voice services, I could buy a less expense set of network gear. Still, this is a $5 billion exercise.

A simpler way to do this would be to use some of the equipment I mentioned at the opening of this post. There seems to be room for a new business model here. Instead of building out my own network, it will soon be possible to provide low-cost network-in-a-box access points to building owners, perhaps at subsidized prices. Have them connect it to their wired data networks. Then the challenge would only be to provide macro base stations for outdoor coverage, a far simpler (i.e. cheaper) proposition. This would become a complicated network in many senses, stringing together tens of thousands of small cells. But that complexity can be managed in the cloud.  This model is similar to what European company Fon is trying to achieve with Wi-Fi. Fon has not been a huge success, but it is an example worth exploring further, and becomes more feasible as the technology advances.

Another approach would be a franchise model, letting local owners of spectrum (e.g. Television stations) build out local networks, with a central franchiser aggregating and managing those networks from a remote data center. This is effectively how we manage international roaming with bilateral agreements among operators. Admittedly, that system is such a mess that it is not a shining example. But starting from scratch we could apply a more sensible model. My New Model Network would focus on regional development, finding regional partners and most of the rest of the network handled through a centralized data center or two.

I would say there are two big unknowns in this vision. The first is the fact the there are a lot of unseen ‘hooks’ protecting the carriers. Some of these emerged with the US Telecoms regulatory ‘reform’ in the 1990’s. It turned out the big operators had many ways to legally blunt competitors (remember ILECs?). The big operators still have a lock on big parts of the world’s networks that they could use to once again beat back the competition. However, offsetting this is another big unknown – new ideas. There is a lot of ferment and experimentation in the networking world today. I hinted at one start-up, but there are many. One example would be Artemis, a company I profiled a few years back. I called this a Science Fiction post, because there is room to let the imagination expand a bit.

 It sounds crazy, but it is becoming clear that the time for crazy ideas may be upon us.

Leave a Reply