Running the Numbers: 700-MHzThursday, August 23, 2007
There is a total of 56 megahertz of spectrum available in the 700-MHz auction. The largest block (22 MHz) is divided into twelve geographic segments, two blocks of 12 MHz each will have 176 and 734 licenses for each segment and the final 10 MHz is for a single nationwide license to be shared with the Public Safety community. (An additional 6 MHz of unpaired spectrum is also available bringing the total to 62 MHz.)
Google and others are talking about bidding for this spectrum and the idea of building a nationwide broadband network that could be used by anyone for anything using any device. In several articles, I have called the 22 MHz of spectrum available a "starter kit" for a nationwide network. Why? Because even with the technologies on the horizon (LTE, UMB and WiMAX-M), 22 MHz is not enough to build a network capable of providing services to a large enough portion of the U.S. market to pay for itself, especially if the winner is going to wholesale spectrum.
The following analysis of how much spectrum is already in service and how many customers are being served by each network backs up my contention that in order to be a serious new entrant with a new network, using the very latest technology, the bidder would have to bid on and win virtually all of the spectrum that will be on the block come January.
Let's take a look at what is out there already. In the 800-MHz band (the first band to be licensed for cellular services in 1981), there is a total of 50 MHz of spectrum broken into two bands of 25 MHz each (which was certainly enough to begin with for the two networks per city). In the PCS portion of the 1900-MHz band, there is a total of 120 MHz of spectrum. Originally, there were six segments to this spectrum, three with 30 MHz and three with 10 MHz, but over time there has been some splitting of these original segments.
The two largest network operators in the United States, AT&T Wireless and Verizon Wireless, both own spectrum in both bands. Sprint and T-Mobile have spectrum only in the 1900-MHz band, although Sprint will also have new spectrum in the 850-MHz band once the Nextel system is dismantled and converted to some other technology.
Prior to the auctions for the AWS spectrum, this meant that the total available spectrum in each city was 170 MHz spread over the four nationwide networks and other operators including second and third-tier operators. This may sound like a lot of spectrum but, in reality, each of the network operators has congestion issues in some part of the nation. The main reason Cingular bought AT&T was for AT&T's rich spectrum holdings.
Taking a look at a few of the larger markets, we find that in New York, for example, with a population of almost 19 million, AT&T has the rights to 55 MHz of spectrum, Verizon 65 MHz, Sprint/Nextel 43 MHz and T-Mobile 20 MHz. In Los Angeles, AT&T Wireless has 65 MHz, Verizon 45 MHz, Sprint/Nextel 43 MHz and T-Mobile 20 MHz. In Denver, with a population base of 2 million plus, AT&T has 45 MHz, Verizon 45 MHz, Sprint/Nextel 43 MHz and T-Mobile 40 MHz with some smaller operators having 10 MHz. (To keep those who like to check the math happy, included in the Sprint numbers is the 13 MHz of spectrum that Nextel averages in most markets.)
Adding AWS Spectrum
As a result of the AWS auctions this year, some of the incumbents added to their spectrum holdings in various parts of the nation. T-Mobile bought enough AWS spectrum to be able to roll out UMTS/HSDPA in most areas, Verizon and AT&T each added some where they needed more capacity and Leap and MetroPCS bought some in areas where they wanted to expand. In addition, the cable companies acquired approximately 20 MHz per market. So after the AWS auction, the same cities mentioned above look like this:
New York: AT&T 55 MHz, Verizon 85 MHz, Sprint (cable companies) 63 MHz, T-Mobile 50 MHz and Metro PCS 20 MHz. Los Angeles: AT&T 95 MHz, Verizon 45 MHz (no addition), Sprint 63 MHz, T-Mobile 50 MHz and MetroPCS 20 MHz. Finally, in Denver, the numbers now look like this: AT&T 65 MHz, Verizon 45 MHz (no change), Sprint 63 MHz, T-Mobile 60 MHz, Leap 30 MHz and one other bidder that won 10 MHz.
How much spectrum does a network operator need in any given market? That depends on how many customers it has and the types of services it is offering. But it should be clear that these companies would not show up at an auction if they thought they had enough spectrum for the future in each of the markets.
On average, AT&T has at least 45 MHz of spectrum in most markets and in some it now has 95 MHz. Verzion Wireless has less, but if you averaged it out, it would be about 50 MHz per market and even T-Mobile, the smallest of the big four, has an average of 50 MHz of spectrum in most markets today.
Consider this: In most markets, Nextel had an average of 13 MHz of spectrum and approximately 15 million subscribers with voice and slow-speed data services. In major metropolitan areas such as New York and LA, trying to make a voice call or complete a PTT call during rush hour can be frustrating. In recent tests we conducted for a client, in LA at 4:00 pm, it took more than twenty minutes for a PTT call to go through. Even with the newer technologies in use today or those on the horizon tomorrow, Nextel's spectrum, by itself, could not support the number of subscribers on the AT&T or Verizon networks.
If Google spends $5 billion on spectrum and manages to purchase all twelve of the 22-MHz licenses, and then decides to build a new network using the latest technologies available, the network could easily cost another $5 billion. So it will have invested $10 billion for a network capable of handling how many customers? Keep in mind that the further you are from the center of a cell site, the slower the data speeds will be and that the total bandwidth per cell sector is shared bandwidth. The only way to increase capacity is to add more spectrum at a cell site or build cell sites that are closer together.
I realize that Google has a lot of money to spend and really doesn't care about the amount of revenue a network could generate―it is after the number of eyeballs that owning the network would drive to its site so it could collect more revenue from its advertisers. Perhaps a mere $10 billion is not a concern to Google, but if it runs out of bandwidth, loads up the network and average data speeds slow down, it will find that the eyeballs it has paid for won't hang around.
I am not trying to discourage Google or anyone else from stepping up to the table and putting their money down, I want to encourage them to run some numbers, understand that wireless bandwidth cannot be augmented by adding more T1s or fiber, and that at some point there has to be a decent return on investment for the money they spend.
If you take this $10 billion investment and then wholesale some of the capacity, the margins get pretty thin. And if Google does wholesale some of its capacity, it won't make back its money from advertising revenue.
This week there were more articles about the Internet and how it could be slowing down because of the new demands being placed on it. Some were expressing doubt that the Internet could handle all of the new demands for video and peer-to-peer file exchanges. Others, of course, believe the Internet will be able to absorb all of this new traffic without a problem. I only know for certain that the Internet is not a mission-critical network, that it is not a managed network, that there is no quality of service in place and that it is subject to slowdowns, denial of service attacks and other issues.
I don't know if the Internet will collapse on itself, but I do know that given the limited amount of spectrum available in the 700-MHz band, even if a company were successful in procuring all of it, it is not enough spectrum to handle millions of Internet users who want to have the Internet in their pocket (if there are that many).
There are two more issues that have to be addressed. The first is how many competitors and networks can 300 million people support? Four? Five? Six? Seven or more? The second is if the desktop Internet on a pocketable device will attract 100 million customers.
I will close with one final point: Today's network operators generate less than 20% of their monthly revenue from data, the rest is from voice. Will the data numbers increase? Yes, they will. Will data income overtake voice income? I don't think so. Even with the newest of technologies, there is a tradeoff between the number of low-bandwidth voice calls and the number of movie downloads that can be processed in a given cell sector, and at the moment the winner is voice.