Yesterday we touched on the subject of 5G small cells. We noted that none of the carriers, with the kinda, sorta exception of Verizon, are in a big hurry to deploy those small cells. We have detailed the obstacles remaining for 5G, and in addition to the (many) technical challenges, there are two key non-technical hurdles gating adoption, both of which relate to the positioning of small cells. The first is just finding enough locations to place the cells, the second is deploying all those cells will require significant capex dollars. And while small cells are going to take a while to reach market, there is a new crop of companies looking to supply into that vast pool of dollars. The trouble is that no one is quite clear what those boxes will look like and how much they will cost.

Traditional, macro base stations are bulky and costly. Roughly the size of a refrigerator, they can cost several hundred thousand dollars for top of the line gear. By contrast small cells look a lot more like a really bulky Wi-Fi router or a cable TV set top box. What are the economics of these boxes?

There are two ways to approach the problem. The big gear makers want these to be shrunk down versions of telco-grade equipment, these could cost $5,000 plus or minus, and come with all the bells of whistles and reliability telcos expect. Alternatively, this gear could instead be upgraded versions of consumer-grade equipment. They would need a layer of durability for outdoor sitings, but should still cost no more than $1,000 plus or minus. Most people we speak with assume that the industry will end up closer to the bottom end of the range.

Let’s do some math. $1,000 times 2 million sites is $2 billion in equipment for the US alone, then tack on an equal amount for labor and installation and the bill comes in at $4 billion. Not tiny, but Verizon’s capex budget in 2020 was $18 billion. If we end up with $5,000 boxes, the bill comes in at $20 billion, which is starting to look like real money.

This is not just a question of the total bill, the nature of the requirements will also determine who will supply the small cells. If we go down the telco-grade path, the market is likely to be dominated by the traditional suppliers like Nokia and Ericsson. They are not good at dealing with these kinds of volumes (and have plenty of other problems), so there is room for a start-up or two to emerge as suppliers for them. By contrast, if we take the consumer-grade path, the industry will need someone with a radically lower-cost approach to building gear. This leaves more room for start-ups, but it will also open the door to some large players accustomed to doing large volumes of electronics.

We know of at least five start-ups taking various approaches to this market. And when we speak with them, the conversation always boils down to – What will Qualcomm do? Qualcomm has a mixed track record in the infrastructure market. Their pattern has been to produce silicon for base stations (called CSMs, in contrast to handset chips called MSMs) early in the adoption cycle of each ‘G’ of the wireless standards. This seeds the market and accelerates adoption of the new standard, but then after a period the company stops producing CSMs. Qualcomm specializes in high volume chips, and the infrastructure market has never been big enough to generate sustained interest from the company. Could small cells be different?

Small cells are something of an in-between market, middle volume. True, it will eventually be millions of units a year, but that is stall ‘small’ by Qualcomm standards which produces several hundred million chips a quarter.

Another important distinction for this market will be who pays for the small cell. The industry is littered with failed attempts to get building owners to pay for telecom gear for their premises, aside from a few markets where regulatory conditions make this viable. So it is seems unlikely that building owners will pay up this time around.

Take the example of sports stadiums an area where we have done A LOT of work lately. There require high density, high capacity systems. The stadium owners also come under immense pressure to upgrade their systems because they do not want the reputation of being in a cell coverage “No Service” zone right in the middle of a city. Nonetheless, selling into a stadium is a difficult process. The most common solution is for operators to support these installations in some way.

But in this example there may be a hint of a new path, maybe with small cells, the market is ready for something different. One of the big features of 5G is the ability to create “private networks”, these are envisioned as networks deployed by enterprises for their own use, where they buy the equipment and pay the telcos to use some portion of their spectrum. This is a whole other complicated topic which we will address in some future post. Enterprises are probably not going to want to use gear from Nokia and Ericsson. Both of those companies have tried to sell into the enterprise over the years, and never gained much traction. So no one has a lock on this market. Now to be clear, private networks are more than just small cells, but if we think about warehouses and offices and factory floors, it should be clear that small cells will be an important element in these deployments, and thus an opening for non-traditional suppliers to enter the market.

We could even think about taking this a step further. The telcos are going to struggle to provide mmWave and other small cell coverage. They are going to have think through new models for deploying these systems, and with some creativity – say revenue share with property owners who buy their own gear – they will likely end up with some new business models.

In the end, we think this leaves open an opportunity for new entrants into telco equipment space – both for makers of boxes and for silicon components. We will likely see a lot of the market going to traditional vendors, but there is likely to be enough of the market left over for others to have a chance of building real companies supplying small cells.