The design of the network of radios and their connections to network nodes that run that radio network and connect it to the internet is profoundly ambitious, in multiple dimensions.
The performance goals of the 5G network are difficult to attain and sustain for several reasons among which:
5G infrastructure will be very expensive and complex to build
Some of the technical wizardry of 5G will be hard to implement reliably in real-world networks
A 5G network will cost significantly more to run and maintain than a 4G/LTE network
The infrastructure of the 5G radio network, when it operates in millimeter wave bands connecting to phones, requires a vast number of radios, referred to as "small cells."
The advantage of a large number of small cellular radios is that some difficult scenarios can be addressed. 5G networks in airports, stadiums, and conference halls will be able to offer enough capacity for people in large crowds to get mobile wireless connections better than they do now. These are the places where millimeter wave bands, and lots of small cells, enhance the ability to solve network engineering problems.
The question is: Are there enough plausible scenarios to make wide deployment of small cells viable? As press demos of 5G have noticed, 5G millimeter wave connections work best when standing still near a 5G cell site, with a clear line of sight between you and the cell site. Almost anything, including your own body and those of other people, can block a millimeter wave signal.
If small cell millimeter wave technology is to deliver its performance potential, it will need a very fast and very dense network moving data between those small cells and high capacity infrastructure routers connecting those sites to the internet. This is called the standalone network (SA).
The fully realized vision of a 5G SA network that has extremely low end-to-end latency, and can do exotic things like "network slicing," that sets aside a "slice" of a network for public safety purposes, is challenging to build.
A standalone network requires an incredibly expensive reconstruction of the mobile network for fiber optic backhaul and high performance network nodes, plus complex control software for features like beamforming that are key to improving usability for millimeter wave small cells.
5G network operators can opt to build 5G networks gradually. 5G telecom industry standards provide more than ten different approaches to designing and building a 5G network, divided into two broad categories: Standalone (SA), where a pure purpose-built 5G network connects 5G devices, and Non-standalone (NSA) networks where the 5G New Radio is used with mostly 4G/LTE network infrastructure, alongside 4G/LTE radios, and which does not enable the very high data rates possible with millimeter wave bands. The availability of gradual paths to 5G means that, at first, on most networks, in most places, the 5G experience will not feel different from 4G.
There will be islands of 5G nirvana, if only because most network operators want to showcase the highest performance capabilities of 5G. they are motivated to test whether a complete 5G implementation adds up to a qualitative difference in the mobile user experience and in the kinds of products they can sell. Nevertheless, most of the planet will be 4G/LTE for a long time to come.
Eventually, though, the 5G standalone architecture may start to pay off: The 5G SA network architecture is inherently simpler than LTE or the 5G NSA architecture. Small cell hardware will become smaller and less expensive. Compact form factors, wireless backhaul and low power requirements have the potential to transform physical requirements for small cells. Those improvements, plus social acceptance of 5G, may eventually enable wider deployment, at supportable cost, of 5G SA networks in cities. The time frame for these potential benefits to be realized is several years, even though SA networks can take advantage of infrastructure investment in LTE Enhanced Packet Core (EPC).
The risk facing 5G network operators taking a more aggressive approach to building out a 5G network is that the new sources of revenue that 5G is supposed to enable, like public safety agencies renting a network slice, 5G in factories, telemedicine, and IoT device makers using 5G extensively, don't actually materialize.
In the worst cases, Building out the 5G network has the potential to be a trainwreck. For example, using millimeter wave bands for cars on the go is challenging because even car windows can block millimeter wave signals. That's right: 5G implies a mobile network that has trouble with signals penetrating cars. Solutions like antennas embedded in car windows have been proposed.
The capital and operating costs of 5G have driven both equipment makers and network operators to speculate on sometimes fanciful uses of 5G, and to reach for markets that may be mirages.
In reality, the likely answer for reducing risk in 5G is gradualism in evolving the network toward 5G.