Is 5G Really a Fiber Replacement?

I recently saw a short interview in FierceWireless with Balan Nair, CTO of Liberty Global. In case you haven’t heard of the company, they are the biggest cable company in the world with over 28 million customers.

One of the things he discussed was the practical widespread implementation of 5G gigabit technology. He voiced the same thing I have been thinking for the last year about the economics of deploying 5G. He was quoted as saying, “5G will be a ‘game-changer’ in its superior ability to transfer data, but the technology will not replace fixed-network broadband services anytime soon. The economics just aren’t there. You’re talking about buying hundreds of towers and all of that spectrum. And on the residential end, putting a device outside the window and wiring it back into the home. It’s a question of business model and if you plan on making any money. The economics benefit fixed.”

The big telcos are making a big deal out of 5G, mostly I think to appear cutting-edge to their investors. And I have no doubt that in certain places like dense urban downtowns that 5G might be the best way to speed up gigabit broadband deployment. But I look at what’s involved in deploying the technology anywhere else and I have a hard time seeing the economic case for using 5G to bring fast broadband to the masses.

5G will definitely make an impact in urban downtowns. You might assume that cities already have a great fiber infrastructure, but this often isn’t the case. Look at Verizon’s FiOS deployment strategy in the past – they deployed fiber where the construction was the most cost effective, and that meant suburban areas where they had existing pole lines or conduit. Verizon largely avoided much of the downtowns of eastern cities because the cost per mile of fiber construction was too expensive.

Now, 5G can be deployed from the top of high-rises to reach the many downtown buildings that never got fiber. New York City recently sued Verizon since the company reneged on its promise to build fiber everywhere and there are still 1 million living units in the city that never got fiber broadband. Verizon, or somebody else is going to be able to use 5G in the densely populated cities to bring faster broadband, and as Nair said, this might be a game changer.

But as soon as you get out of downtowns and high-rises the math no longer favors 5G. There are three components of a 5G network that are not going to be cheap in suburbia. First, 5G needs fiber. You might be able to use a little wireless backhaul in a 5G network, but a significant portion of the network must be fiber fed. And in most of the country that fiber is not in place. Deloitte recently estimated that the cost for just the fiber to bring 5G everywhere is $130 billion. There is nobody rushing to make that investment.

5G then needs somewhere to place the transmitters. This is more easily achieved in a downtown where there are many tall rooftops and existing towers. But the short delivery distances for millimeter wave frequencies mean that transmitters need to be relatively close the end-user. And in suburban areas that’s going to mean somehow building a lot of new towers or else placing smaller transmitters on existing poles. We know suburbia hates tall towers and it’s always a struggle to build new ones. And the issues associated with getting access to suburban poles are well documented. An ISP needs to affordably get onto poles and also get fiber to those poles – two expensive and time-consuming challenges.

And then there is the economics of the electronics. Because millimeter wave spectrum is easily disrupted by foliage or any impediments it means that there won’t be too many homes served from any one pole-mounted transmitter. But the 5G revenue stream still has to cover both ends of the radios as well as wiring into the home.

I build a lot of landline business plans and I can’t see this making any economic sense for widespread deployment. In many cases this 5G network might be more expensive and slower to deploy than an all-fiber network.

I instead envision companies using 5G technology to cherry pick. There will be plenty of places where there is existing fiber and poles that can be used to serve suburban apartment complexes or business districts. I can see strategic deployment in those areas and the technology used in the same way that Verizon deployed fiber – 5G will deployed only where it makes sense. But like with FiOS, there are going to be huge areas where there will be no 5G deployment, even in relatively dense suburbia. And the business case for rural America is even bleaker. 5G will find a market niche and will be one more technology tool for bringing faster broadband – where it makes economic sense.

5G Needs Fiber

I am finally starting to see an acknowledgement by the cellular industry that 5G implementation is going to require fiber – a lot of fiber. For the last year or so the industry press – prompted by misleading press releases from the wireless companies – made it sound like wireless was our future and that there would soon not be any need for building more wires.

As always, when there is talk about 5G there is a need to make sure which 5G we are talking about, because there are two distinct 5G technologies on the horizon. One is high-speed wireless loops send directly to homes and businesses as a replacement for a wired broadband connection. The other is 5G cellular providing bandwidth to our cellphones.

It’s interesting to see the term 5G being used for a wireless microwave connection to a home or business. For the past twenty years this same technology has been referred to as wireless local loop, but in the broadband world the term 5G has marketing cachet. Interestingly, a lot of these high-speed data connections won’t even be using the 5G standards and could just as easily be transmitting the signals using Ethernet or some other transmission protocol. But the marketing folks have declared that everything that uses the millimeter wave spectrum will be deemed 5G, and so it shall be.

These fixed broadband connections are going to require a lot of fiber close-by to customers. The current millimeter radios are capable of deliver speeds up to a gigabit on a point-to-point microwave basis. And this means that every 5G millimeter wave transmitter needs to be fiber fed if there is any desire to offer gigabit-like speeds at the customer end. You can’t use a 1-gigabit wireless backhaul to feed multiple gigabit transmitters, and thus fiber is the only way to get the desired speeds to the end locations.

The amount of fiber needed for this application is going to depend upon the specific way the network is being deployed. Right now the predominant early use for this technology is to use the millimeter wave radios to serve an entire apartment building. That means putting one receiver on the apartment roof and somehow distributing the signal through the building. This kind of configuration requires fiber only to those tall towers or rooftops used to beam a signal to nearby apartment buildings. Most urban areas already have the fiber to tall structures to support this kind of network.

But for the millimeter technology to bring gigabit speeds everywhere it is going to mean bringing fiber much closer to the customer. For example, the original Starry business plan in Boston had customers receiving the wireless signal through a window, and that means having numerous transmitters around a neighborhood so that a given apartment or business can see one of them. This kind of network configuration will require more fiber than the rooftop-only network.

But Google, AT&T and Verizon are all talking about using millimeter wave radios to bring broadband directly into homes. That kind of network is going to require even more fiber since a transmitter is going to need a clear shot near to street-level to see a given home. I look around my own downtown neighborhood and can see that one or two transmitters would only reach a fraction of homes and that it would take a pole-mounted transmitter in front of homes to do what these companies are promising. And those transmitters on poles are going to need to be fiber-fed if they want to deliver gigabit broadband.

Verizon seems to understand this and they have recently talked about needing a ‘fiber-rich’ environment to deploy 5G. The company has committed to building a lot of fiber to support this coming business plan.

But, as always, there is a flip side to this. These companies are only going to deploy these fast wireless loops in neighborhoods that already have fiber or in places where it makes economic sense to build it. And this is going to mean cherry-picking – the same as the big ISPs do today. They are not going to build the fiber in neighborhoods where they don’t foresee enough demand for the wireless broadband. They won’t build in neighborhoods where the fiber construction costs are too high. One only has to look at the hodgepodge Verizon FiOS fiber network to see what this is going to look like. There will be homes and businesses offered the new fast wireless loops while a block or two away there will be no use of the technology. Verizon has already created fiber haves and have-nots due to the way they built FiOS and 5G wireless loops are going to follow the same pattern.

I think the big ISPs have convinced politicians that they will be solving all future broadband problems with 5G, just as they made similar promises in the past with other broadband technologies. But let’s face it – money talks and these ISPs are only going to deploy 5G / fiber networks where they can make their desired returns.

And that means no 5G in poorer neighborhoods. It might mean little or limited 5G in neighborhoods with terrain or other similar issues. And it certainly means no 5G in rural America because the cost to build a 5G network is basically the same as building a landline fiber network – it’s not going to happen, at least not by the big ISPs.

Cellular Networks and Fiber

We’ve known for a while that the future 5G that the cellular companies are promising is going to need a lot of fiber. Recently Verizon CEO Lowell McAdam verified this when he said that the company will be building dense fiber networks for this purpose. The company has ordered fiber cables as large as 1,700 strands for their upcoming build in Boston in order to support the future fiber and wireless network there. That’s a huge contrast from Verizon’s initial FiOS builds that largely built a network using mostly 6-strand fibers in a lot of the Northeast.

McAdams believes that the future of urban broadband will be wireless and that Verizon intends to build the fiber infrastructure needed to support that future. Of course, with that much fiber in the environment the company will also be able to supply fiber-to-the-premise to those that need the largest amounts of bandwidth.

Boston is an interesting test case for Verizon. They announced in 2015 that they would be expanding their FiOS network to bring fiber to the city – one of many urban areas that they skipped during their first deployment of fiber-to-the-premise. The company also has engaged with the City government in Boston to develop a smart city – meaning using broadband to enhance the livability of the city and to improve the way the government delivers services to constituents. That effort means building fiber to control traffic systems, police surveillance systems and other similar uses.

And now it’s obvious that the company has decided that building for wireless deployment in Boston is part of that vision. It’s clear that Verizon and AT&T are both hoping for a world where most devices are wireless and that the wireless connections use their networks. They both picture a world where their wireless is not just used for cellphones like today, but will also be used to act as the last mile broadband connection for homes, for connected cars, and for the billions of devices used for the Internet of Things.

With the kind of money Verizon is talking about spending in Boston this might just become the test case for a connected urban area that is both fiber rich and wireless rich. To the extent that they can do it with today’s technology it sounds like Verizon is hoping to serve homes in the City with wireless connections of some sort.

I’ve discussed several times how millimeter wave radios have become cheap enough to be a viable alternative for bringing broadband to urban apartment buildings. That’s a business plan that is also being pursued by companies like Google. But I still am not aware of hardware that can reasonably be used with this same technology to serve large numbers of single family homes. At this point the electronics are still too expensive and there are other technological issues to overcome (such as having fiber deep in neighborhoods for backhaul).

So it will be interesting to watch how Verizon handles their promise to bring fiber to the homes in Boston. Will they continue with the promised FTTP deployment or will they wait to see if there is a wireless alternative on the horizon?

It’s also worth noting that Verizon is tackling this because of the density of Boston. The city has over 3,000 housing units per square mile, making it, and many other urban centers, a great place to consider wireless alternatives instead of fiber. But I have to contrast this with rural America. I’m working with several rural counties right now in Minnesota that have housing densities of between 10 and 15 homes per square mile.

This contrast alone shows why I don’t think rural areas are ever going to see much of the advantages of 5G. Even though it’s expensive to build fiber in a place like Boston, the potential payback is commensurate with the cost of the construction. I’ve always thought that Verizon made a bad strategic decision years ago when they halted their FiOS  construction before finishing building in the metropolitan areas on the east coast. Verizon has fared well in its competition with Comcast and others.

But there is no compelling argument for the wireless companies or anybody else to build fiber in the rural areas. The cost per subscriber is high and the paybacks on investment are painfully long. If somebody is going to invest in rural fiber they might as well use it to connect directly to customers rather than to spend the money in fiber plus adding a wireless network on top of it.

We are going to continue to see headlines about how wireless is the future, and for some places like Boston it might be. Past experience has shown us that wireless technology often works a lot different in the field compared to the lab, so we need to see if the wireless technologies being considered really work as promised. But even if they do, those same technologies are going to have no relevance to rural America. If anything the explosion of urban wireless might further highlight the stark differences between urban and rural America.

Broadband Shorts – March 2017

Today I’m writing about a few interesting topics that are not long enough to justify a standalone blog:

Google Scanning Non-user Emails. There has been an ongoing class action lawsuit against Google for scanning emails from non-Google customers. Google has been open for years about the fact that they scan email that originates through a Gmail account. The company scans Gmail for references to items that might be of interest to advertisers and then sell that condensed data to others. This explains how you can start seeing ads for new cars after emailing that you are looking for a new car.

There are no specific numbers available for how much they make from scanning Gmail, but this is part of their overall advertising revenues which were $79.4 billion for 2016, up 18% over 2015.  The class action suit deals with emails that are sent to Gmail users from non-Gmail domains. It turns out that Google scans these emails as well, although non-Gmail users have never agreed to the terms of service that applies to Gmail users. This lawsuit will be an important test of customer privacy rights, particularly if Google loses and appeals to a higher court. This is a germane topic right now since the big ISPs are all expected to do similar scanning of customer data now that the FCC and Congress have weakened consumer privacy rights for broadband.

Verizon FiOS and New York City. This relationship is back in the news since the City is suing Verizon for not meeting its promise to bring broadband to everybody in the city in 2008. Verizon has made FiOS available to 2.2 million of the 3.3 million homes and businesses in the city.

The argument is one of the definition of a passing. Verizon says that they have met their obligation and that the gap is due to landlords that won’t allow Verizon into their buildings. But the city claims that Verizon hasn’t built fiber on every street in the city and also that the company has often elected to not enter older buildings due to the cost of distributing fiber inside the buildings. A number of landlords claim that they have asked Verizon into their buildings but that the company either elected to not enter the buildings or else insisted on an exclusive arrangement for broadband services as a condition for entering a building.

New Applications for Satellite Broadband.  The FCC has received 5 new applications for launching geostationary satellite networks bringing the total requests up to 17. Now SpaceX, OneWeb, Telesat, O3b Networks and Theia Holdings are also asking permission to launch satellite networks that would provide broadband using the V Band of spectrum from 37 GHz to 50 GHz. Boeing also expanded their earlier November request to add the 50.4 GHz to 52.4 GHz bands. I’m not sure how the FCC picks winners from this big pile – and if they don’t we are going to see busy skies.

Anonymous Kills 20% of Dark Web. Last month the hackers who work under the name ‘Anonymous’ knocked down about 20% of the web sites from the dark web. The hackers were targeting cyber criminals who profit from child pornography. Of particular interest was a group known as Freedom Hosting, a group that Anonymous claims has over 50% of their servers dedicated to child pornography.

This was the first known major case of hackers trying to regulate the dark web. This part of the Internet is full of pornography and other kinds of criminal content. The Anonymous hackers also alerted law enforcement about the content they uncovered.

Are Cable Companies Winning the Speed War?

Polk County SignThe latest news about Google Fiber slowing on their metropolitan fiber builds got me to wondering if perhaps the cable companies are starting to win the speed wars. Are we getting to a time when a fiber overbuilder is going to have trouble competing with them?

After many years of being stingy with bandwidth the cable companies have now largely adopted the opposite strategy and increase household speeds over time without raising prices. I can remember quotes from several big cable companies a few years ago where the cable companies claimed they were giving households all the speed that they need. And this was back at a time when they were experiencing a significant amount of network congestion during the peak evening hours. But my reading of many different customer reviews tells me that the cable companies have largely solved the congestion issue.

This is not to say that there are not places where the cable networks are still not up to snuff, but compared to ten years ago, a lot more cable networks seem to be delivering the speeds that customers want. Of course, there are still plenty of small town where the rural cable networks are not up to snuff, but metropolitan areas seem to have improved a lot.

The FCC reported in their 2015 Measuring Broadband America Fixed Report that Comcast customers got between 109% and 119% of the speeds that they paid for. I know personally that my speed tests often shows at least 5 Mbps better performance than what I am paying for with Comcast.

But the question that has been nagging me is if a new fiber provider can really thrive in a metropolitan area? Can they get enough customers to be profitable? It’s been widely reported that Google and other fiber overbuilders need at least a 30% market share to succeed, and that’s a tall order in a city where everybody already has broadband.

People need a compelling reason to change providers, because it’s a process that nobody enjoys. It means staying at home to meet an installer, returning settop boxes and modems, and worrying about the billing transition.

I have some anecdotal evidence about the way at least one group of people buy broadband. I’ve been a member of several active Maryland sports message boards for over two decades and broadband is a periodic topic of conversation since sports fans these days care about watching sports on the Internet. The majority of the people on these boards happen to live in neighborhoods that have both Verizon FiOS and a cable company – mostly Comcast, but sometimes somebody else. These are folks who have had the choice between fiber and coaxial cable networks for a long time.

What I’ve seen over the years is that there are a few people that are big fans of either the cable company or Verizon. But the vast majority don’t seem to really care as long as the broadband works well enough to watch their sports and the other things their families do on the Internet. Probably half of the people on these boards have moved back and forth between the providers during the last decade. I’ve seen evidence that content matters more than speeds when over the years there were occasions when one provider or the other did not broadcast a Maryland football or basketball game. At least among this one large group of I don’t see any major affinity for fiber over coaxial cable networks. These folks just want something that will work.

A new fiber provider has to provide a compelling reason for people to change. Certainly having lower prices could be a compelling reason, but most metropolitan fiber providers are not much cheaper than the cable company (and sometimes they are more expensive). And while a fiber provider might offer gigabit speeds, I wonder if that is enough to get people to change if they are happy with the speeds they have had for the last few years?

I’ve always said that there is some percentage of any community that will change to a new provider because they dislike the current provider for some reason. But those are rarely enough customers to justify a business plan, and so being successful with fiber also means persuading customers that are not unhappy to change. And perhaps, as Google has found out, that is not as easy as fiber proponents have assumed. Certainly, the cable company tactic of greatly improving the performance of their data products is making it harder and harder for a new overbuilder to thrive.

Thoughts on the Google Fiber News

800px-OSU_Bucket_TruckIt was recently reported that Larry Page, the CEO of Alphabet told Google Fiber to cut their staff in half from 1,000 to 500 and to also cut the cost of building new fiber. That certainly is going to slow or even stop Google Fiber’s expansion plans. There are a few lesson to be learned from that announcement for all fiber start-ups.

It’s expensive to build in cities. Building new fiber networks from scratch is expensive. It’s very doubtful that Google has found any magic that would let them build fiber networks for much less than anybody else. Verizon always said they stopped expanding FiOS due to the construction costs. Yet Verizon built most of their network in places where the construction was relatively inexpensive. They lashed fiber to existing phones cables where there was poles or installed underground fiber where there was existing conduit. But Google Fiber didn’t have either of those advantages and so they were spending a lot more than Verizon.

The company is now looking at wireless technologies to cut construction costs. Unless the company has some really good millimeter wave technology ready to roll out soon, this could cause a big delay on expansion. Lots of companies are thinking about wireless loops, but we know from past experience that there is a big transition with any wireless technology when moving from a lab into the real world.

You have to sell what customers want to buy. The news articles I saw say that Google Fiber had 200,000 customers at the end of 2014. One would expect that they have up to twice that by now. But considering their initial goal to have millions of customers those numbers are low.

My guess is that the company has had trouble convincing enough households to buy their $70 broadband. I would buy that in a flash, but I expect a lot of homes found the product to be out of reach for their budgets. I see that in Atlanta that Google Fiber has introduced a 100 Mbps broadband connection for $50, and that has to open up a lot more sales opportunities for them.

This highlights one of the biggest challenges for a fiber overbuilder. The biggest expense by far of getting into the business is to build the fiber network. Once you’ve sunk that money the goal is to get as many paying customers as possible onto the network, and that means having a wide enough array of products (which are still profitable) to generate revenue.

Delays are to be expected. There has been news over the last few years of delays in Google Fiber construction, much of it having to do with access to poles. Google Fiber learned the same lesson that had deviled many of my clients: when you need to get poles from the company you will competing with they are going to use every legal trick in the book to slow down the process. I’ve written about pole horror stories before in the blog.

Labor costs can be your enemy. The announcement said that Google Fiber had to slash their work force from 1,000 to 500. They face the same dilemma as other broadband start-ups – there is a lot of work to do in launching new markets compared to the number of people needed to operate them once they are mature.

For many years I have used a general rule of thumb for the number of employees that a telecom company ought to have. For example, a medium sized carrier with 20,000 to 50,000 customers ought to have roughly 1 employee for every 350 customers. As companies get larger and more efficient that ratio increases and companies up to about 250,000 ought to have around 1 employee for every 400 customers. Bigger companies should get even more efficient, but seems to be a natural cap at some size reached by very large companies.

It’s impossible to judge if Google Fiber has too many employees without knowing how many customers they have today. I also have no idea if the company uses contractors in addition to full-time employees because those would count in this calculation. But if the company has 400,000 customers today and doesn’t use contractors then the 1,000 employee count wouldn’t be too bad. But if they have fewer customers and also use contractors they are currently overstaffed.

One issue that can justify a larger staff than normal is the need to have a staff that is working ahead on future markets. Those employees would not be counted when looking at the size of the needed staff.

Also, the ratios I cited are more than a decade old and if I put some thought to them I would probably revise them higher. There are numerous improvements in customer service tools and other efficiencies that can reduce the number of times that a customer has to talk to or see a live employee. With the upcoming AI revolution one would imagine that a lot of customer service is going to be handled by bots and lower labor costs.

The bottom line of all of these issues is that Google Fiber hit the same wall hit by every other overbuilder. There comes a time when you have to show profitability or the money dries up. And it’s really hard to show profitability when you are still growing rapidly.

What’s the Right Price for a Gigabit?

Speed_Street_SignI often get asked how to price gigabit service by clients that are rolling it out for the first time. For an ISP already in the broadband business, layering in a super-fast Internet product on top of an existing product line can be a real challenge.

Google certainly lowered the bar for the whole industry when they priced a gigabit at $70. And that is the real price since Google doesn’t charge extra for the modem. I think the Google announcement recalibrated the public’s expectations and anybody else that offers a gigabit product is going to be compared to that price.

There are a few other large companies marketing a gigabit product in multiple markets. CenturyLink has a gigabit connection for $79.95 per month. But it’s hard to know if that is really the price since it is bundled with CenturyLink’s Prism TV. The cheapest Prism TV product offered on the web costs $39.99 per month and includes 150 channels of programming and also comes with an additional settop box fee of $9.99 per month – the highest box fee I’ve seen. I don’t know exactly what kind of bundle discount is available, but on the web I’ve seen customers claiming that the cheapest price for the gigabit bundle is around $125 per month. That’s a far cry from Google’s straight $70. And for customers who want to use a gigabit to cut the cord a forced bundles feel a bit like blackmail.

Verizon FiOS has not yet given in to the pressure to offer a gigabit product. In looking at their web site their fastest product is still a symmetrical 500 Mbps connection at $270 per month plus an added fee for a modem, and with a required 2-year commitment. A 1-year commitment is $280 per month.

Comcast will soon offer a gigabit in more markets than anybody else. In Atlanta where Comcast is competing against Google Fiber a gigabit is $70 per month with a 3-year contract, including an early termination fee (meaning that if you leave you pay for the remaining months). This package also requires an additional modem charge. Without a contract the price for the gigabit is $140. It’s unclear if Comcast is offering the same lower-price deal in other markets with newly upgraded DOCSIS 3.1 like Chicago. The word on the Internet is that customers are unable to sign-up for the lower-price option in these markets, but the company says it’s available. I’m sure the availability  will soon become clear.

One thing that happens to any company that offers a gigabit is that the prices for slower speeds are slashed. If a gigabit is $70 – $80 then slower products must become correspondingly less expensive. Google offers a 100 Mbps product for $50 and each of the other companies listed above has a range of slower bandwidth products.

The first question I always ask an ISP is if they are offering gigabit speed for the public relations value or they really want to sell a lot of it. There are plenty of ISPs that have gone for the first option and have priced a gigabit north of $100 per month.  But for somebody that hopes to sell the product, the dilemma is that they know that the majority of their customers will buy the least expensive product that provides a comfortable speed. The rule of thumb in the industry is that, in most markets, at least 80% of customers will buy the low or moderate priced options. But if the choice is between a gigabit product and a 100 Mbps product, the percentage buying the slower product is likely to be a lot higher.

The issue that small ISPs face when recalibrating their speeds is that they end up increasing speeds for most existing customers. If they migrate from a scale today where 50 Mbps or 100 Mbps is the fastest product up to a new scale topped by a gigabit, then they have to increase speeds across the board to accommodate the new gigabit product.

This is a hard mental block to get over for many small ISPs. If a company offers a range today of products from 6 Mbps to 75 Mbps it’s mentally a challenge to reset their slowest speed to 50 Mbps or faster. They often tell me that in doing so they feels like they are giving away something for free. If a company has been an ISP since the dial-up days they often have a number of customers that have been grandfathered with slow, but inexpensive broadband. It’s a real dilemma when rebalancing speeds and rates to know what to do with households that are happy with a very cheap connection at 1 Mbps or 2 Mbps product.

For the last ten years I have advised clients to raise speeds. ISPs that have raised speeds tell me that they generally only see a tiny bump in extra traffic volume after doing so. And I’ve always seen that customers appreciate getting faster speeds for the same price. Since it doesn’t cost much to raise speeds it’s one of the cheapest forms of marketing you can do, and it’s something positive that customers will remember.

I think most ISPs realize that the kick-up to gigabit speeds is going to be a change that lasts for a long time. There are not many customers in a residential market that need or can use gigabit speeds. What Google did was to leap many times over the natural evolution of speeds in the market, and I think this is what makes my clients uneasy. They were on a path to have a structure more like Verizon with a dozen products between slow and fast. But the market push for gigabit speeds has reduced the number of options they are able to offer.

Investing in Fiber

fios vanThere is a recent short article in Forbes titled, To Evade the Wheeler Tax, Capital is Fleeing Digital Infrastructure by Hal Singer. The premise of the article is that the FCC’s move to regulate broadband under Title II has somehow driven the large ISPs to stop investing in broadband. He cites the fact that Verizon has used their excess cash to buy content and software rather than invest it in infrastructure. Just in the last year Verizon has bought AOL for $4.4 billion, Fleetmatics (connector of smart devices) for $2.4 billion and recently announced the purchase of Yahoo for $4.8 billion.

But Singer couldn’t be more wrong. It’s obvious that Singer has a bias against broadband regulation by the title of his article, and certainly Forbes is in generally favor of the unregulated marketplace. But for his premise to be true, Verizon would have to be stopping its expenditures for broadband and instead be choosing these new paths. And that has not happened.

Verizon stopped building FiOS fiber well over a decade ago. It’s been clear for a long time now that Verizon doesn’t think their future is in landlines. For a very long time they have considered themselves as a wireless company. One only has to look at their annual report to understand that somebody who didn’t know the company might barely realize they are in the wireline business. Those reports talk almost entirely about cellular, which makes sense since the wireless business dwarfs the broadband business.

Singer is right that Verizon has been ditching landline properties. But these were mostly in areas away from Verizon’s northeast core and included both copper and fiber assets. But it’s also clear that Verizon hasn’t given up broadband in the northeast. They are currently in the process of buying XO for $1.5 billion, one of the larger fiber-based CLECs that’s centered in the northeast. Verizon also recently announced they were going to bring broadband to Boston, and it now looks like this will become a test bed for using millimeter wave radios as a fiber-to-the curb deployment, rather than building traditional FiOS networks. My guess is that Verizon sees wireless local loops for broadband as their next big use of their cellular spectrum and existing fiber assets, and if Boston proves the new technology then Verizon will probably begin making huge investments again in broadband.

The problem with articles like Singer’s is that rich people that make big investments read Forbes and might decide that investing in broadband is a bad idea. Before they do that I hope that they look at broadband investments with the right perspective. Investing in broadband is an investment in infrastructure. And that means that such the investment is going to earn infrastructure-like returns. Anybody that builds broadband networks is likely looking at long-term returns of 10% to 20%. The returns can be a little higher for cherry-picking only the best neighborhoods. And the returns will probably be higher if wireless local loops can save on capital expenditures.

But infrastructure returns are not venture capital returns. Investments today in software and content are seeking returns of at least 30%. But such investments are a lot riskier than investing in broadband – and thus the relative returns.

The fact is that for the last ten years almost nobody has invested in broadband in this country. Most of the new construction since Verizon stopped building FiOS has come from independent telephone companies, municipalities and cooperatives. Today we are seeing more activity with the two biggest players being Google and CenturyLink, along with a dozen or so smaller urban fiber builders. We also now see the cable companies making significant investments to move cable modems to the next generation DOCSIS 3.1.

So the decision by the FCC to regulate ISPs under Title II has changed almost nothing because big telcos like Verizon were not investing in landline broadband before that decision. Certainly that decision might eventually put a cap  on the returns of the largest ISPs like Comcast and Verizon. But mostly the FCC rules are going to stop the large ISPs from ripping off the public with data caps or by raising the rates through the backdoor by inventing imaginary fees. But the FCC rules are not going to change the fundamentals of the marketplace that understands that investment in broadband is infrastructure investing. Companies that make such investments will still make infrastructure-like returns, like has been true during all of my career. The much more fundamental question that Singer ignores is, why aren’t there more companies looking to make 10% to 20% infrastructure returns? Answering that question might require a book rather than a blog or a short article.

Fiber for Everyone?

Fiber CableJust a few days ago I wrote about the two cities that are considering having citizens pay for their fiber networks through utility fees and pledges to support the fiber financing. After writing about Ammon, Idaho I heard back from several people in the industry pointing out that the proposed Ammon utility fee was a pledge intended to support bonds. The fees, which are supposed to be about $16.50 per month for about twenty years, would total nearly $4,000 over that twenty-year period and would be used to secure, and then pay for, the bonds needed to build the system.

That raises an issue that I have raised before: how important is it that everybody in a community get access to broadband? Every community that thinks about finding a fiber solution faces this issue. They can look for an approach that will get fiber to every household or they can settle for something less. This choice is sometimes a philosophical decision, but it often comes down to the difference in cost between the two choices.

Ammon has clearly chosen a solution that will benefit homeowners who are able and willing to pledge a lien on their homes. To be able to make the pledge a resident must own a home that can be pledged, so this eliminates renters. Interestingly it might also make it a challenge for anybody who doesn’t think they’ll be in their home for long. According to the US Census, the average time that families stay in an owned home is 13 years. And fewer than 40% of homeowners stay even 10 years. So anybody that thinks they are going to move out of their home in Ammon in the next few years probably ought not to pledge since they are likely to have to cover the remaining amount of the lien when they sell their home.

I don’t want to sound like I’m coming down negative on Ammon, because they have come up with a creative solution to get fast broadband to at least part of their city. And that is exactly what a whole lot of other cities have done. Ammon is unique because of their creative financing solution, but a whole lot of other cities have settled for broadband to less than everybody.

For instance, almost every city getting Google fiber is going to end up with fiber built to only parts of their city. Only cities willing to step up with a lot of city dollars like Huntsville, Alabama are going to get fiber everywhere. And the vast majority of cities that got Verizon FiOS years ago now thinks they made a mistake since they now have fiber in some neighborhoods and not others. They are now seeing a big difference between neighborhoods with fiber and those without. This difference is likely to grow since both Verizon and AT&T have made noises about tearing down copper in older city neighborhoods. We might end up with more urban households without affordable landline broadband than we have today in rural areas.

Fifteen years ago I worked for several cities that wanted to get Verizon’s attention to get onto the FiOS list. At that time these cities were so ecstatic to get some fiber that they didn’t insist that Verizon eventually build their whole city. But it probably would not have mattered if they had – because there are cities that got that agreement from Verizon but which still don’t have fiber everywhere.

I don’t want to make Google and Verizon sound like bad actors because almost every large fiber overbuilder is doing the same thing in only building to the most profitable parts of cities. The returns from only building to the best neighborhoods are dramatically better than from building everywhere – I’ve created dozens of business plans that quantify the difference. This is also the approach being taken by CenturyLink, Aspire, and half a dozen other fiber overbuilders – they are simply making the best financial decision for their company.

This is a tough philosophical issue for a city. Do they take the high ground and hold out for a solution that gets fiber everywhere or do they take the practical approach and get some fiber built? The risk of holding out for a whole-city solution might mean that nobody gets fiber. But the flip side of this is that building to only parts of a city probably means there will be neighborhoods that will be cut off from fiber for decades to come – talk to any city that has FiOS if you don’t believe that.

It’s almost impossible to build a reasonable business plan today to somehow fill in fiber where Verizon didn’t build – because they built where the construction costs were the lowest.  So Ammon is not at all unique, and in fact they are joining the majority of the cities in the US that have elected a solution that will result in something less than 100% fiber coverage. My primary reaction to this issue is a personal one – I know how I’d feel if I was in one of the neighborhoods that didn’t get fiber. I think that any city that elects to build less than 100% fiber ought to expect to hear an outcry from the rest of the city for many years to come.

Fiber is Not Always Fiber

Fiber CableI had a conversation today that is the same one I’ve had many times. I was talking to a City that has already built hundreds of miles of fiber to connect municipal buildings. They were shocked to hear that the network they had built had very little value if they wanted to build FTTP for all homes and businesses. To do that would require building additional fiber on just about every route they had built in the past.

In order to understand why this is so, you have to understand that there are several very different kinds of fiber networks in the world – long-haul, distribution and last mile. Each kind of these networks is built in a very different way that doesn’t make them useful for the other two uses.

Consider long-haul fiber. This is the fiber that is built to connect cities and to stretch across the country and the world. The key to having an affordable long-haul fiber route is to carry each leg of the network as far as possible without stopping and having to repeat the signals. So generally long-haul fibers will only stop in major POPs in cities or wherever necessary along the way at repeater sites that are used to boost the signal.

I have a client who is sitting on one of the major east-west Interstates in the country. Every major long-haul carrier in the country is passing very near to him with owned or leased capacity on the long-haul route. My client was shocked when not one carrier would come off of the long-haul route to provide him with bandwidth. But it’s very costly to break into long-haul fibers, and every splice creates a little interference, and so long-haul providers are generally extremely reluctant to provide bandwidth anywhere except at established POPs along the fiber.

Distribution fiber is similar to long-haul fiber, but within local fiber networks. Distribution fibers are built to connect very specific points. The networks that cable companies build to get to their neighborhood nodes are a distribution network. So are networks built by telcos to reach DSL cabinets or networks built by cities to reach traffic signals and networks built by school boards to connect schools. These networks were generally built for the specific purpose of reaching those end points.

Distribution fiber routes share the same issue as the long-haul routes and the companies that own them are generally reluctant to break these routes along the way to connect to somebody that the network was not designed for. Certainly distribution fibers have very little use for providing service to many homes or businesses. There are generally not enough pairs of fiber in distribution routes, but more importantly they are not built with access points along the way.

The major characteristic of last mile fibers (and the one that makes them the most expensive) is the presence of many access points – using handholes, manholes, splice boxes or other ways to connect to the fiber where it’s needed. Providing these access points every few hundred feet adds a lot of cost to building fiber, and it’s very expensive, and often impossible to add access points after the fact to existing distribution fiber.

This is why I actually laughed out loud last year when I read about Comcast’s 2 gigabit product that they would sell to people who were close enough to their fiber. The Comcast fiber network is almost entirely a distribution network that connects from a central hub or ring to electronic boxes in neighborhoods known as nodes. When Comcast said a customer had to be close to their fiber, they didn’t mean that fiber had to be close the home, because distribution fiber passes lots of homes and businesses. They meant that a customer had to live close to a neighborhood node which are the only access points on a cable distribution network. Cable nodes are often placed where they aren’t an eyesore for homeowners and so I laughed because in most places hardly anybody lives close enough to a Comcast node to buy the touted 2 Gig service. In a city of 20,000 like mine I would be surprised if there are more than a dozen or two homes that would qualify to buy the Comcast service.

The same is true for most of the big ISPs. Verizon has built a lot of last mile fiber with FiOS and now CenturyLink is starting to do the same. But some of the other large companies like AT&T are very happy to talk about how close they are with fiber to homes and businesses without mentioning that the vast majority of that fiber is distribution fiber, which is close to worthless for providing fiber to homes and small businesses. This is not to say that companies like Comcast or AT&T don’t have any last mile fiber. They just don’t have very much of it and it’s generally limited to business parks or to greenfield neighborhoods where they’ve installed fiber instead of copper.