Mallinson: eMBB will prevail on path from 4G to 5G while Intel pins its hopes on cloud and IoT

Keith Mallinson

Intel has foundered and is largely withdrawing from smartphones, tablets and 4G: instead, it is cutting its losses somewhat while still gambling on what is hoped to be the next big thing with IoT in 5G. This is a strategic reboot, along with a major corporate restructuring involving 12,000 layoffs, that sidesteps unfulfilled expectations for profitable mobile business operations in the short or medium term. While capitalizing on its leading position with growth in cloud infrastructure, Intel is putting its faith—and testing that of its stockholders—on unproven prospective markets and in Intel's ability to tap them substantially in due course. Meanwhile, Qualcomm, MediaTek and others including their numerous device vendor customers are capitalizing on the interim opportunities along the path to 5G in enhanced mobile broadband (eMBB) that are substantial and well proven both technically and commercially.

As indicated in an article of I have recently written for 3GPP, the path from 4G to 5G will be as much an evolution as it is a revolution. Companies that are contributing technologies to the improvement of LTE in 3GPP Releases 13, 14 and 15, and are capturing substantial near-term product demand growth for chips, devices and network equipment in eMBB including smartphones in particular. They will be best positioned around 2020 for full-blown IMT-2020-compliant 5G in Release 16, including the broad array of use cases and devices that this and subsequent releases will support, some of which might also eventually become "killer applications."

Head in the clouds

In a recent blog entitled "Our Strategy and The Future of Intel" the company's CEO Brian Krzanich wrote "[o]ur strategy itself is about transforming Intel from a PC company to a company that powers the cloud and billions of smart, connected computing devices." Some of this makes great sense but much of it is tricky. The cloud is firmly establishing itself as predominant at the core of our ICT networks and PCs are in decline. With respect to the network end points Krzanich mentioned PCs four times, devices six times, "things" eighteen times; but there was no mention of smartphones or tablets.

Nevertheless, it is still personal connectivity and computing with smartphones and to a lesser extent tablets that prevail and will continue to do so for many years. There are 7.8 billion cellular connections worldwide, but so far only 1.2 billion of these include LTE, nearly half of them have only 2G connectivity and most mobile phones in use are still feature phones rather than smartphones. Upgrading these is where most of the commercial value will be in the next four years prior to the first commercial 5G launches and for several years thereafter in chip and device revenues, if not in unit sales. Whereas the business case for multibillion subscriber growth in smartphones and LTE is well established, the prospect of hoped-for tens of billions of connected things over an entirely new network fabric is not yet proven. Targeting 5G with IoT for market re-entry involves additional risks for Intel in comparison to incumbent competitors.

Intel's shortcomings in smartphones and LTE are manifold. The application and power-consumption performance of the ARM-based processors produced by other suppliers are superior to Intel's x.86-based chips in smartphones. x.86 chips have also been far more costly for OEMs to incorporate, including the other associated components required in device designs, and so Intel has had to heavily subsidize its chips sales. These "contra-revenues" have contributed to many hundreds of millions of dollars in quarterly losses in mobile over several years. Intel has failed to exploit its in-house process technology to competitive advantage and has instead had to subcontract modem chip manufacturing to TMSC, as do smartphone chip market leaders Qualcomm and MediaTek. Intel has been repeatedly late to market in even seeking to provide what the market wants, and then in executing when it belatedly lays plans to do so. For example, Apple's iPhones aside—where Apple pairs its own A-series application processors with modems supplied by Qualcomm since 2011—more than 90 percent of 1.4 billion smartphones sold last year include a system-on-a-chip with the application processor integrated with the baseband modem. Intel was slow to address this major part of the market while commercially shipping only standalone application processors and modems for smartphones until 2015. SoFIA LTE, a low-end integrated application processor and modem built on a foundry 28-nanometer manufacturing process, was supposed to launch in the first half of 2015, but it was delayed to the second half of 2015 and then into the first half of 2016 before being axed in Intel's recently-announced corporate restructuring. Intel has a lot of homework to do to ensure it does not repeat its costly failures. Taking a timeout from the market will surely save it money, but it will not enhance its competitive prospects in 5G.

Three legs to 5G include massive IoT, low latency and eMBB

5G developments including standardization are being driven by various visions for what 5G might ultimately become. As indicated in the figure, the wide variety of use cases requires performance improvements over and above what LTE can offer including ultra-high reliability and low latency, and massive connectivity with IoT. However, 5G's third and most solid leg is very much a continuing progression of the improvements in the evolution of LTE with eMBB speed and capacity increases.  LTE is an astounding success largely on the basis of connecting a rapidly-growing base of smartphones with mobile broadband. According to the Global mobile Suppliers Association, by April 2016, 494 LTE networks are commercially launched in 162 countries. The majority of smartphone models on sale support LTE with 5,104 LTE user devices launched on the market by 417 manufacturers.

Figure 1: Three dimensions to performance improvements with usage scenarios for 2020 and beyond

Economies of scale and cumulative experience with many chips developed and large volumes produced extends most significantly beyond silicon chip processes and fabrication to include application-specific functional design, operating software and acceptance among carriers worldwide. Leading smartphone chip vendors are exploiting what they learn with the new technologies, refinements and optimizations they implement in the large numbers of chips they continue to develop and sell all along the way to 5G. It is a long shot for Intel to hope for a major competitive edge with a fresh start through market re-entry to address unproven types of customer demand with a new generation of technology. Likely as not, most 5G IoT device requirements will play more to the expanding experience with low-cost devices of leading mobile phone chip producers rather than to Intel with its strength in high-performance and more costly chips for PCs and data centers. 5G will not be so entirely new or different that leading LTE modem suppliers will not retain major competitive advantages over new entrants.

Evolution to 5G with LTE-Advanced-Pro

The path from 4G to 5G will be as much an evolution as it is a revolution. Developments are being underpinned by the proven and increasing demand for smartphones and eMBB. My aforementioned article for 3GPP investigates how very well LTE is doing and is continuing to be improved and adopted; and what its relationship with 5G will be in terms of technology developments and supply as well as demand.

Figure 2: The 3GPP route to 5G

Whether one calls the next step 4.5G as announced by Huawei, pre5G as announced by ZTE, or uses the official 3GPP term LTE-Advanced Pro; standard-compliant LTE technology that builds upon current and completed 3GPP Releases 8 through 14 –along with the first drop of New Radio air interface features with Release 15 in 2018 – will form 5G in its Phase 1 deployments commencing 2020.

Keith Mallinson is a leading industry expert, analyst and consultant. Solving business problems in wireless and mobile communications, he founded consulting firm WiseHarbor in 2007. Find WiseHarbor on Twitter @WiseHarbor.