Internet of Things
Internet of Things
Internet of Things
The Internet of Things (IoT) has generated serve as an important new source of revenue. Given
excitement for a few years now, with start-ups and the size of the potential opportunity, McKinsey
established businesses placing bets on the industry’s recently collaborated with the Global Semiconductor
growth.1 Some of the earliest investments have Alliance (GSA) to investigate the Internet of Things
begun to pay off, with smart thermostats, wearable more closely, with a focus on risks that could derail
fitness devices, and other innovations becoming progress. In addition to assembling a fact base, we
mainstream. With new IoT products under surveyed and interviewed senior executives from
development or recently launched—ranging from the semiconductor sector and adjacent industries, as
medical-monitoring systems to sensors for cars— described in the sidebar, “Our methodology.”
some analysts believe that the Internet of Things is
poised for even greater gains. Our research suggests that the Internet of Things
does indeed represent a major opportunity for
Semiconductor companies, perhaps even more than semiconductor companies—one that they should
other industry players, might benefit from the IoT’s begin pursuing now, while the sector is still
expansion. With growth rates for the smartphone developing. We also found, however, that the timing
market leveling off, the Internet of Things could and magnitude of the IoT’s growth may depend
1
A new source of growth
2
companies had mixed feelings about the IoT’s smart cities, with applications to assist with
potential, with 48 percent stating that it would be one traffic control and other tasks within the
of the top three growth drivers for the semiconductor public sector
industry and only 17 percent ranking it first.
The challenge ahead
Despite the size of the IoT opportunity, some Like many other high-tech innovations, the Internet
semiconductor companies have hesitated to make of Things is garnering intense interest in the press,
significant investments in this sector. The greatest with reports of connected cars and smart watches
issue is that products within the Internet of Things making headlines. Although we do not want to
tend to appeal to a niche market and generate diminish the IoT’s potential, our research suggests
relatively low sales volumes. With individual that the following six issues could derail its growth:
products delivering a relatively low return on
investment, some semiconductor companies have inadequate security and privacy protections for
limited their R&D expenditures for IoT-specific user data
chips, preferring instead to adapt existing products.
For instance, wireless system-on-chip players may difficulty building customer demand in the
offer repurposed wireless processors and chip sets absence of a single “killer application”
for the Internet of Things, while microcontroller
players often bundle lower-end processors and a lack of consistent standards
connectivity-chip sets to compete for the same
opportunity. the proliferation of niche products, resulting
in a fragmented market and an unprofitable
As the IoT market matures and increases in scale, environment for creating application-
semiconductor companies may decide to pursue specific chips
new approaches more aggressively. Before moving
ahead, however, they should first determine which the need to extract more value from each
verticals and applications are growing strongly and application by providing comprehensive
assess when their markets will be large enough to solutions, rather than focusing solely on silicon
justify significant investment. While semiconductor
companies could potentially capture growth in many technological limitations that affect the
IoT verticals, six of the most promising markets— IoT’s functionality
those where we chose to focus our research—include
the following: These problems are not insurmountable, particularly
if semiconductor companies are willing to take an
wearable devices such as fitness accessories active role in solving them.
3
key issue while your application or product has not knowledge of hardware security to application
reached scale, but once you are at scale and maybe designers or network-equipment manufacturers,
have a first incident, it becomes a most important since this information would assist with the design
problem.” Our survey results echoed the interview of secure software.
findings, with respondents ranking security as the
top challenge to the IoT’s success. Recent hacks to Customer demand: Developing the end market
online car systems also highlight the importance Many of our interviewees envisioned a future in
of addressing security challenges for connected which IoT applications are more common than
devices, vehicles, and buildings. cell phones are today. Others were more cautious,
however, with one noting, “No one really knows
Ensuring security will not be easy, however, given when the volume will show up; this is a clear
the numerous applications and verticals within the challenge. . . . If you cannot show a $1 billion
Internet of Things, each with its own quirks and opportunity, then it’s hard to get attention.”
requirements. For instance, fitness wearables might
only require relatively basic security measures that In other technology sectors, a single groundbreak-
ensure consumer privacy, such as software-based ing application or use case—a so-called killer
solutions. But IoT applications that control more app—has often spurred explosive demand. Such
critical functions, including medical electronics and was the case in 2007, when the introduction of
industrial automation, need much higher security, the iPhone triggered significant growth in the
including hardware-based solutions. smartphone market. While the Internet of Things
could potentially follow this path, most of our
Most executives we interviewed believed that interviewees felt that growth would stem from a
the technology needed to secure the Internet of string of attractive but small opportunities that use a
Things was already available. They were concerned, common platform, rather than a single killer app.
however, with the piecemeal nature of most
security products and wanted to ensure that players Some of the most innovative IoT applications—and
protected the entire IoT stack—cloud, servers, and those most likely to stimulate customer demand—
devices—rather than focus on only one of these areas. could come from start-ups. Businesses outside the
As one executive said, “Overall security is only as technology sector, such as retailers, insurers, and
good as its weakest point.” oil and gas players, might also develop interesting
products that appeal to a wide customer base,
Semiconductor companies can assist with end- although some of our interviewees felt that these
to-end solutions by providing on-chip security, companies would face tough odds. Semiconductor
partitioning processor functions on chip, or players could help indirectly stimulate demand
supplying comprehensive hardware and software for IoT devices if they adopt new strategies to
services, including authentication, data encryption, help these players thrive. For instance, start-ups
and access management. Those that specialize in and nontechnical businesses often have limited
security might be able to use their own products experience with semiconductors, so they might
to provide comprehensive solutions, but others appreciate simple solutions and more hands-on
will need to undertake M&A or form partnerships support, including guidance from dedicated field
with players further up in the stack to gain broader engineers who assist with board-level design
expertise in software or the cloud. For instance, and solution integration (from silicon through
semiconductor companies could lend their applications in the cloud). IoT customers might
4
also prefer one-stop solutions—complete platforms To see how a lack of uniform standards can
with all relevant elements that an IoT device complicate product development and industry
needs, including connectivity, sensors, memory, growth, consider connectivity issues. There are
microprocessors, and software. For some small competing, incompatible connectivity standards for
businesses with limited funds, such platforms may devices with a low range and medium-to-low data
be the only economically feasible option. rate—for instance, Bluetooth, LTE Category 0, and
ZigBee . With so many options, product designers
IoT standards: The need for consistency may be reluctant to create new devices, since they do
Some layers of the IoT technology stack have no not know if they will comply with future standards.
standards, and others have numerous competing Similarly, end users may be reluctant to buy devices
Semiconductors 2015
standards with no obvious winner. In our survey and that may not be interoperable with existing or future
Internet of Things
interviews, most respondents cited this situation as products of the same type (Exhibit 1).
Exhibit 1 of 2
a major concern, with one executive stating, “What
is critical is which standards will win and when this Although multiple organizations, including interest
will happen.” groups and industry consortia, are attempting to
Exhibit 1 Standards for the Internet of Things (IoT) are not mature in many categories,
including connectivity.
Widely adopted New standard Established, adoption ongoing
Data rate, log scale Power consumption, indicative 1 Many competing standards for
low-range, medium-low data
1 Gbps rate hinder growth for many
High
Wi-Fi
IoT applications
• Interoperability missing
100 Mbps 4G LTE • Consortia wars might be
emerging
Bluetooth • Additional incompatibilities in
10 Mbps higher communication layers,
1
eg, 6LoWPAN vs ZigBee
2
1 Mbps 2 Standard white space for
LTE Cat. 01 low-data-rate, low-power,
high-range applications such
100 Kbps ZigBee as smart grid
802.11ah • Wi-Fi and LTE have high
power consumption
10 Kbps • Alternatives with low power
5
establish standards, it is impossible to predict which For instance, many low-cost applications have
ones will prevail in each IoT vertical. Faced with this common requirements for short-range, medium-
uncertainty, semiconductor players should pursue data-rate connectivity and limited data processing.
a hedging strategy, focusing on selected standards If semiconductor companies create a common
that are likely to gain widespread acceptance while platform for applications that fit this archetype, they
simultaneously planning for alternative scenarios. will simultaneously increase demand and reduce
In all cases, semiconductor players should actively R&D spend. One downside of a platform approach is
engage with industry associations or other groups that the chips may not deliver optimal performance
that are trying to develop IoT standards, with the for every application they cover.
goal of supporting the best ones.
Extracting value beyond silicon: Opportunities in
How to address a fragmented marketplace: software and more
Creating common platforms Semiconductor companies have a well-deserved
IoT devices have widely varying requirements for reputation and track record for technological
power, data-processing speed, form factor, price, innovation, with some of their inventions
and other dimensions. Smart water meters, for spurring advances in personal computing, mobile
instance, need to run for months, if not years, telecommunications, and elsewhere. But they are
independent of power supply. They also require also known—fairly or unfairly—for failing to extract
high-range connectivity, but data rates can be under full value from their innovations, with other high-
one kilobit per second. By contrast, IoT devices used tech players, such as software firms, profiting most
for industrial automation typically require a direct from device enhancements. Our analysis suggests
connection to a power supply and high data rates, that semiconductor companies might face a similar
but their connectivity range is lower than that for dilemma with the Internet of Things. One executive
smart meters. we interviewed noted, “Value extraction has always
been a particular challenge for semiconductor
These variations in device specifications become players, and it becomes particularly challenging
significant when considering R&D costs for a in the Internet of Things, as even more players
single chip. Assuming typical integrated-circuit participate in the stack and business models are still
design costs and product lifetimes, semiconductor immature.” Other interviewees stated that big data
companies will need to ship 20 million to 70 million and cloud companies were positioned to capture
chips annually to break even.2 Only a few segments, far more value from the Internet of Things than
such as wearables, are large enough to require semiconductor businesses.
so many chips, making it impractical to create
customized solutions for individual applications. But To tackle this problem, some semiconductor
rather than abandon the IoT market, semiconductor companies have already begun to create complete
companies should investigate an approach that solutions that cover multiple layers of the technology
involves classifying devices into archetypes based stack, especially since nontraditional customers—
on their specifications and then creating a single start-ups and businesses outside the technology
platform to cover each one. sphere—prefer this approach. It is still too early to
identify winning strategies, but the advantage may
Products from multiple verticals can fall under one go to companies that pursue the following three
archetype as long as they have similar specifications. opportunities (Exhibit 2 ).
6
Software. Semiconductor companies have provided chip-security solutions, but they may
been complementing integrated circuits with find additional opportunities in other layers,
supporting software for many years, but this particularly if they can offer software products.
trend will become even more important as the
Internet of Things grows. Many semiconductor Systems integration. One interview subject stated,
companies have recently sought to build their “Today most players have a partial but not full
software skills through M&A or partnerships, solution for integrated systems, so who is the
while others have2015
Semiconductors focused on improving their integrator?” Semiconductor companies could
IOTin-house capabilities. fill this role, especially if they provide system- or
Exhibit 2 of 2 application-level software supporting integrated
Security. As noted earlier, the Internet of Things circuits, although some interviewees noted that
requires end-to-end security across the stack. this might be too much of a departure from their
Semiconductor players have traditionally core competencies.
Exhibit 2 Semiconductor companies must seek value beyond silicon and software.
Semiconductor players’ approaches to a business model for the Internet of Things (IoT)
End-user
applications
Applications
Cloud
services
Servers/
infrastructure
System Devices
integration
Software/
algorithms
Processing
chips
Connectivity Core semiconductor Core semiconductor Core semiconductor
Silicon business business business
chips
(Embedded)
sensors
7
Technological issues: Finding opportunities improvement. Innovation in power and battery
within the challenge life will likely come from various sources, such as
In our survey, two-thirds of respondents stated that on-device power management, further advances
technological issues present little to no challenge in storage, over-the-air and wireless charging, and
to the success of the Internet of Things.3 The energy harvesting.
remaining respondents were split evenly between
those who thought that technological issues Although semiconductor companies that offer
were above average in importance and those who leading-edge technological advances will find
considered them a major challenge. Executives may themselves in high demand, not every player has
hold varying opinions because technological issues to focus on innovation. Those companies that
differ by vertical and application. For instance, the offer more dated solutions will still have a role in
interviewees agreed that wearable technology needs the Internet of Things, since many applications—
improvement. “With wearables, there is a constant particularly the sensors they contain—will continue to
issue of charging,” one executive stated. “We need to rely on existing (albeit highly specialized) technology.
make devices last for a trip.” By contrast, technology
for smart-home applications is well advanced, but Implications for semiconductor players
there are few standards governing interoperability, Semiconductor companies that want to capture the
which has limited their adoption. IoT’s enormous growth potential might be tempted
to move ahead quickly, without changing their
When we asked our interviewees about the most existing operating model, but this could be a mistake.
crucial technological innovations for the Internet of The Internet of Things is unlike any high-tech
Things, most focused on lower power consumption segment that they have previously served, and their
and battery life. A step change in the time between traditional strategies may not succeed with the new
charges will increase demand for existing devices customer base. With so much at stake, semiconductor
while also enabling product designers to create companies need to reevaluate all aspects of their
new applications. Wearable computers, distributed businesses and potentially make some significant
sensors for agricultural applications, and retail changes. From a strategic perspective, three tactics
beacons are just a few of the applications that require will be particularly important.
8
Finding the right niches suited to your competencies Revisiting (and revolutionizing) the corporate
There are likely to be many profitable I0T niches operating model
within the fragmented market, and semiconductor Operating models focusing on hardware and
companies will need to identify the most promising embedded software helped semiconductor
ones that represent a fit with their capabilities. The companies thrive in many high-tech segments, but
use of a platform approach to cover multiple niches they may not be well suited to I0T customers. For
will be important, since R&D costs may otherwise be instance, most companies now include a limited
prohibitive. When companies are selecting the right number of large business units, a focus on direct
niches, one of the most important considerations sales and field-application engineers, and an
is their own expertise. Semiconductor players that emphasis on application-specific R&D programs.
have strong ties to consumer-electronics companies A more appropriate organizational structure for the
and possesses full-system-integration capabilities Internet of Things would emphasize a multimarket
might best focus on wearables and smart-home sales approach and a greater reliance on channel
devices, developing silicon, software and algorithms, partners, such as distributors, as part of the go-to-
and device-level designs. They could also potentially market strategy. This arrangement is well suited
provide server-side software, connectivity gateways, to the I0T’s fragmented market, which contains
and associated infrastructure. By contrast, a very different companies, including many small
company with specific expertise with high-reliability businesses, with unique needs. Other possible areas
integrated circuits and security might be well suited for improvement include the following:
to provide full IoT solutions for medical applications.
R&D. The move from customized chips to a
Developing a solid strategy to seek value platform approach should occur as soon as
beyond silicon possible, but this does not always entail massive
Semiconductor companies are mostly well aware internal changes. Instead, companies may be
that their chips represent only one small part of the able to license another player’s intellectual
IoT value chain, so they are exploring opportunities property to build a platform—for instance, for
in software, the cloud, and other services. But they image processing—thereby gaining access
may also need to consider more radical approaches to new technologies without increasing
to improve the value captured, including a shift to development costs.
new business models. For instance, a move to usage-
based pricing would allow semiconductor players to Investments. Rather than making a limited
capture revenue for the entire lifetime of a device or number of large portfolio bets under the
service, not just at the time of chip purchase. (This direction of a business-unit lead, companies
might only be possible if a semiconductor company should investigate numerous applications
is willing to provide the full system or partner in diverse markets. This approach will help
with a system-level player.) To mitigate risks and companies avoid the common mistake of
avoid moving too far from their core competencies, allocating most funds to core products, rather
however, new solutions should be carefully evaluated. than using them to develop new applications.
The fact that the IoT has many niches will be helpful
during the evaluation process, since companies can Change management. If management wants
test solutions in one of them and make necessary employees to cultivate new capabilities or
adjustments before undertaking a broader rollout. develop innovative products, they may need
9
to revise their key performance indicators. For 1 For the purposes of this report, we defined the Internet of Things
example, companies should provide incentives as a network containing all “smart” devices with some sort of
sensing mechanism that can communicate via the Internet with
that encourage R&D to develop chip platforms
other smart devices or the cloud without human interaction.
that are appropriate for several verticals, such Most executives in our survey preferred this definition. Our
as connected cars and industrial automation, findings would also be relevant to the Internet of Things if it were
more broadly defined as a network consisting of any object that
rather than optimize integrated circuits for a
can be accessed through the Internet, including PCs, tablets,
single vertical. Likewise, leaders that want to and smartphones.
focus on mergers or other outside alliances must 2 Design costs are assumed to be $12 million for low-complexity
help companies recognize their importance by integrated circuits and $40 million for high-complexity
integrated circuits, based on benchmarking data from
encouraging such partnerships more aggressively.
Numetrics (a McKinsey Solution). Product lifetime is assumed
to be five years in a market with ten competitors, 40 percent
incremental margin, and $3 average selling price.
3 Survey respondents were asked to distribute 100 points across
10