“Prediction is Very Difficult” and for 5G, Few Have the Bona Fides
Neils Bohr was right, forecasting is hard. But through the analytical process, one learns a lot about how the market could grow and what are the primers for growth. The secret to a reliable forecast is to build the most coherent model possible based on the insights you have, and then review your assumptions frequently. Many insightful analysts will dodge forecasting because it is so hard. ABI Research makes it a practice to apply the forecasting due diligence that gives you confidence for building your future.
At MWC-2014, all the infrastructure vendors had a 5G story, built around defensive hedging of “we can’t say for sure, but…” group-think. However, they were all diligently showing their 5G research works in progress. Hard it is, says Bohr, but prediction and forecasts, like strategy, are working hypotheses of what the future may become. When new information is available, likely from investment that retires risk, we act accordingly. The infrastructure vendors are hard at work retiring uncertainty and risk.
There is a lot of hype and expectation around 5G. One could almost be mistaken to think 5G is just around the corner and going to take the market by storm. ABI Research wanted to put the 5G market in context. How will it be? Once you factor in which countries are likely to be first movers, as well as the readiness of their “innovator” and “early adopter” subscriber segments to respond to the 5G opportunity? We at ABI look at 5G and forecast 100M subscribers by 2025. A 100 million is no small number but it tells you that in a world of nearly 9 billion mobile subscriptions, “5G subscribers” will be about 1% of the total subscriber base. In other words, 5G will be on the flat part of the Bass model S-curve. Now what does that mean for the mobile ecosystem?
A number 10 years out is not tactical, but how is it strategic? It is strategic because companies make investment decisions. What should they do now? When and where should they invest? As for 5G, the time now is to nail down the fundamental technologies and innovations, and more importantly, a patent portfolio. The size and timing of the market tells you that if your firm missed 4G, then it probably is not 5G that will save the day. Many firms that missed the 3G window are no longer around, even though they tried to leap-frog to 4G.
There is already a growing assembly of hardware vendors, software developers, system integrators, R&D institutions, patent lawyers, regulators and telcos taking a keen interest in 5G. They are already asking themselves:
- How big is this market?
- When will it happen?
- What is the use case?
- What’s the killer app?
If the answer to these questions is, “I don’t have any insight” then what is an executive to do? Probably look for another analyst. Different analysis methods are needed to deal with the long lead times and uncertainties of the 5G future. This is why we adopt tools like Real Options Valuation, where we take our best view, invest a little, retire some risk, and revaluate. So it is with our forecasts and the industry writ large.
But what do we know and can infer about 5G? What does it take and when can we retire uncertainty and risk?
For starters, there is a lot that we do know about 5G. It is not like transitioning from steamships to airline travel. We know 2G, 3G and 4G subscriber uptake and forecasts. We know 4G, which is based on OFDMA, has technical characteristics that give it greater bandwidth, more capacity and better performance than 3G, which is based on CDMA, and we can infer what attributes and benefits 5G must have to succeed. A “4G subscriber” did not sign up for “4G” because of its technical characteristics but because of the utility of the service (internet access on the go, a high speed connection), the price at which it can be delivered to them, and whether their operator is offering the 4G service. Neither will a “5G” subscriber sign up because of a millimeter-wave air interface.
We know the key objectives 5G is meant to deliver:
- 10-100x typical end-user data rates
- 1000x higher mobile data volumes
- 5x lower latency
- 10-100x higher number of connected devices
- 10x longer battery life for low-power devices
Even if all of these objectives are not met concurrently or immediately, we know that certain combinations of these objectives will target certain applications and devices.
We are certain that 5G needs the “network of tomorrow” but the converse is not assured, that is, the “network of tomorrow” will need 5G. In other words, all those things that will happen upstream of the air interface will happen regardless of 5G.
On the air interface side – the part that defines generations of mobile wireless technology, we need only simple logic to see the direction 5G is taking. We know what the likely key technologies will be because of industry and university research. We know that part of the huge gain in data rate is due to the shift from wideband channels used by LTE-A to ultra-wideband channels for 5G. We know that those ultra-wideband channels are available in centimeter and millimeter wave spectrum. We know what RF technologies, such as massive MIMO and 3D beamforming, are required to make that spectrum useable when historically it was considered impossible to use. Not only will these technologies make the spectrum usable, but combined with high-frequency spectrum will allow for accurate spatial division around each base station, which taps yet another dimension of spectrum more efficiently than using cell division alone. One can paint a picture about the direction the industry is moving and core technologies are becoming clear. Some important parts, such as the waveform and modulation used are still not clear, but the possibilities are not a secret. The details of the standard will come for interoperability, but those are the finer points built around the core technologies that are already taking shape.
Can we put some dimensions on 5G? Yes, we believe we can put in place reasonable assumptions about 5G. We do have a sense of what percentage of operators will adopt 5G, and since 5G will deliver even better services, and probably even widen the addressable market (handsets, M2M modules, IOT), even if we just simply used baseline 4G (or even 3G) penetration assumptions, we can generate a “5G subs forecast”. Will we refine the forecast over time? Absolutely, as new information comes in and uncertainty retired, we deal with it.