GEDocument1.PDF

9 – 6 1 4 – 0 3 2
R E V : M A R C H 9 , 2 0 1 5

Professors Karim R. Lakhani and Marco Iansiti and Case Researcher Kerry Herman (Case Research & Writing Group) prepared this case. It was
reviewed and approved before publication by a company designate. Funding for the development of this case was provided by Harvard
Business School and not by the company. Professor Karim Lakhani has previously worked at GE. He has also taught at GE Crotonv ille and has
had several consulting engagements with the company. Professor Marco Iansiti has also consulted to GE. HBS cases are developed solely as the
basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of eff ective or ineffective
management.

Copyright © 2014, 2015 President and Fellows of Harvard College. To copies or request permission to reproduce materials, call 1 -800-545-
7685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu. This publication may not be digitized,
photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.

K A R I M R . L A K H A N I

M A R C O I A N S I T I

K E R R Y H E R M A N

GE and the Industrial Internet

We are the oldest remaining company in the Dow Jones Industrial Average. This is not because we are a
perfect company; it is because we adapt. Through the years, we have remained productive and competitive. We
have globalized the company, while investing massive amounts in technology, products and services. We know
we must change again.

— Jeff Immelt, CEO, General Electric1

General Electric (GE) CEO Jeff Immelt (MBA 1982) sat in his office with Chief Marketing Officer
Beth Comstock and Vice President Bill Ruh, head of the new GE Software business unit. It was early
2014, and they were reviewing the latest report of completed and potential customer contracts for
GE’s new Industrial Internet initiative. Now a little over two years after announcing the initiative and
betting more than $1 billion on its rollout, GE could already directly attribute over $800 million in
sales to the effort. Yet these numbers represented only a tiny portion of GE’s annual revenues—close
to $146 billion for 2013. Immelt and his team worried: Were they doing enough to give the initiative
traction within GE? Could GE, an industrial machines manufacturer, sell outcomes-based services
based on analytics and software?

GE’s Industrial Internet initiative proposed an open, global network that connected machines,
data, and people, and provided data synthesis and analysis allowing for real-time and predictive
solutions to optimize the complex operations of GE’s varied customer base, including predicting
maintenance and repair needs and informing performance and operational decisions. GE’s suite of
Industrial Internet offerings was designed not only to create and sell “smart” software–enabled
machines but also to provide outcomes-based services contingent on improving operational
performance via data collected and analyzed in cooperation with customers. Wall Street and
technology analysts projected that the Industrial Internet (also referred to as the Internet of Things)
would generate tremendous value in both increased revenues and decreased costs. Analysts
estimated the Industrial Internet would create $14.4 trillion in economic value between 2013 and
2022.2,3 By that time, they projected that Industrial Internet–related technology spending would

exceed $514 billion.4 Some GE customers were already seeing benefits from their connected machines.

For example, improving airline engine efficiency by 1% equated to $2 billion in annual savings.5
Integrating dispatcher data with customer data improved a carrier’s package delivery by 10.4%.6
(Exhibits 1a and 1b provide examples of potential savings; Appendix A describes the Industrial
Internet.)

This document is authorized for use only by John Rome ([email protected]). Copying or posting is an infringement of copyright. Please contact [email protected] or
800-988-0886 for additional copies.

614-032 GE and the Industrial Internet

2

Since announcing the initiative, GE had seen intense changes in the past 12 months, including the
build-out of a new software headquarters; the launch of a common technology platform across GE’s
diverse industrial businesses; a thorough assessment across the organization of GE’s software
development expertise and readiness evaluation of its sales talent capable of supporting this new
direction; and new and expanded partnerships with companies such as Intel, Cisco, and Accenture.

GE had signed several promising agreements, including a $300 million contract with a utility
company, a reliability-based service for an oil and gas customer for $20 million, a wind farm deal for
$35 million, a $100 million deal with a U.S. hospital chain, and a possible $1 billion rail deal. A
GE/Accenture joint venture, Taleris, which provided intelligent operations for aircraft and cargo
carriers, had also just announced its first signed customer, Etihad. These agreements offered a range
of benefits, including a monitored rate of flow on oil rigs, optimized wind turbines to adapt to
changes in weather, optimized patient intake at hospitals, and predicted replacement of air
conditioners in a fleet of aircraft in to avoid downtime. Each deal was highly idiosyncratic,
relied on deep familiarity and expertise with the specific customer and its sector, and required GE to
be innovative and customize how it partnered with the customer and sold the software-enabled,
outcomes-contingent offering. Many of these deals also required customers to allow significant access
to internal operational data and some kind of value/revenue/profit-sharing arrangement, in contrast
to GE’s more traditional contract service agreements.

Immelt felt the initiative was an opportunity that GE could not afford to ignore. Global Innovation
Executive Director Steve Liguori said, “We have new non-traditional ‘competitors’ starting to
approach our long-standing customers. IBM mostly, but SAP and big data start-ups are telling our
customers they can provide these analytics and services, on GE assets.” Comstock added, “Our
customers are under intense pressure, given the ongoing uncertain economic environment. We’re not
selling as much hardware ourselves.” Immelt and his team had to ask if GE’s customers were ready
for the Industrial Internet. In October, GE had taken an informal poll of customers about their
readiness and adoption of the Industrial Internet and learned that 63% of customers polled said their
machines were connected to networks, but they were not yet using these data, 13% claimed they used

data for competitive advantage, and 63% were not performing any condition-based maintenance.a7

Internally, debates remained heated over which business model the initiative should pursue.
Some argued that GE should develop software capabilities and give them away for free as part of an
intensified focus on capital equipment sales and service contracts. GE Software Chief Marketing
Officer John Magee said, “In the past, our mentality was to build and ship boxes. Any software we
had was often given away as part of a hardware sale.” A second camp saw opportunity in the
software capabilities themselves and argued that GE should license these offerings as a separate
product. Finally, a third camp argued that GE should embrace the initiative and pursue software and
analytics investments that enabled new, outcomes-based service offerings that would mean rich,
deep integration with GE customers and their data.

Pursuing the third option meant a host of changes. Magee said, “This initiative creates brand-new
business models for us. Software as a service represents a whole new beast for GE.” GE had to
identify and develop new opportunities, source and hire developer and sales expertise and talent,
build the offerings, and price and coordinate sales of each offering through each division’s sales
mechanisms. Liguori added, “At GE, 99% of our sales force sells ‘big iron.’ They are used to selling
capital goods versus gain share or revenue-share arrangements with our customers. They are used to

a Condition-based maintenance referred to maintenance undertaken when needed, due to indicators that alert to either the
deterioration or potential failure of equipment or system.

This document is authorized for use only by John Rome ([email protected]). Copying or posting is an infringement of copyright. Please contact [email protected] or
800-988-0886 for additional copies.

GE and the Industrial Internet 614-032

3

talking to operations managers who run the plants that our equipment is in. Now we need to expand
our message to the entire C-suite, showing how we can help run all their assets and, ultimately, their
enterprise better.” The Industrial Internet called for a different approach. But what levers could the
team pull to accelerate the initiative? Immelt was well known for his metrics-driven management;
which metrics had the most impact on GE’s ability to speed up the initiative? Immelt turned to
Comstock and Ruh and asked, “Are we moving fast enough? Can we be faster in this or not?”

Company History8

Thomas Edison founded General Electric in 1892 after the merger of Thomson-Houston and
Edison General Electric. Edison set the company on a long trajectory of innovation. He amassed over
2,000 patents with inventions such as the phonograph, the radio, and the incandescent light bulb,
along with the capability of generating and transmitting the electricity to electrify them.

Through the 20th century, GE manufactured products and provided services across a broad range
of industries, including aircraft engines, locomotives and other transportation equipment, kitchen
and laundry appliances, lighting, electric distribution and control equipment, generators and
turbines, and medical imaging equipment. By 1980, GE earned $25 billion in yearly revenue from
plastics, consumer electronics, nuclear reactors, and jet engines.9 Through the 1980s and 1990s, GE
shut down underperforming businesses and diversified, investing more heavily in financial services
and entertainment. As it diversified, GE looked less like a traditional industrial company, with profits
coming increasingly from its financial services offerings, which included commercial finance,
commercial aircraft leasing, real estate, and energy financial services. By the late 1990s, analysts
identified the increased sales of service contracts as a contributor to GE’s increased profits.10 From

year to year, under CEO Jack Welch, GE looked for ways to “go downstream,”11 pursuing a
comprehensive services business model that offered a range of ancillary services to its industrial
products, tying the customer more closely to GE.

GE under Immelt

Immelt took over as CEO of GE in September 2001, just days before the September 11th terrorist
attacks in the U.S. Immelt recalled his first weeks as CEO: “It was super intense. I was not cool at
all.”12 Immelt had joined GE in 1982 in corporate marketing, moving from there into leadership
positions with GE Plastics and then GE Appliances. He was president and CEO of GE Medical
Systems from 1996 to 2000, where he increased its share of service business from 25% to 42%.13

By the early 2000s, GE was operating in a changing world, as Immelt noted, “In a deflationary
world you could get margin by working productivity. Now you need marketing to get a price.”14 As
he pursued a more service- and customer-oriented organization, he brought in Comstock in 2003 as
GE’s first chief marketing officer in 20 years. Immelt and Comstock built an entire new marketing
function for GE, recasting many of the firm’s extensive business development talent as marketers;
developing a new leadership program (Experienced Commercial Leadership Program) to parallel
GE’s traditional offerings in engineering, finance, manufacturing and sales; and appointing a vice

president–level marketing head for each business.15 In 2003, Immelt set up and personally chaired a
Commercial Council to convene sales and marketing leadership across the company to transfer best
practices and accelerate initiatives throughout the organization quickly. To track his teams’ success

This document is authorized for use only by John Rome ([email protected]). Copying or posting is an infringement of copyright. Please contact [email protected] or
800-988-0886 for additional copies.

614-032 GE and the Industrial Internet

4

with customers, he relied on two metrics: a net promoter scoreb and an operating metric best suited to
each business.

During Immelt’s reign, services continued to grow as a portion of revenue. By 2005, contract

service agreements (CSAs) made up about three-quarters of GE’s total backlogc and contributed
about 75% of industrial earnings.16 CSAs were regular maintenance agreements for the total
operation management of an asset. They included preventative maintenance and fixes, essentially
“break-fix” insurance; when an industrial product—a machine—broke, GE fixed it. CSAs provided
flexible service packages designed to meet a customer’s specific requirements, goals, and budget to
ensure optimum performance of GE’s equipment, stable operating costs, and planning for spare and
replacement parts. As Magee explained, “CSAs evolved into turn-key, long-term, recurring revenues
streams for GE,” generating reliable high-margin income that lasted the life of the piece of

equipment, often several decades.17 (See Exhibit 2 for an evolution of GE’s service model over time.)

Immelt also looked to focus GE on its strengths and opportunities. He spun off businesses that did
not fit, including GE Plastics, many of its financial services and insurance businesses, and
NBCUniversal, while he grew some of GE’s traditional businesses through acquisitions. In 2007, GE
acquired an aircraft systems manufacturer and an oil and gas production equipment supplier. GE
continued to invest in the technology needed to gather, process, and use the information generated
by its machines and, by 2010, had over 5,000 software engineers worldwide, with two newly opened

software centers near Detroit, Michigan, and Richmond, Virginia.d Revenue from software reached
$2.5 billion (of $150 billion).18 As the first decade of the millennium came to a close, GE continued to

consolidate and focus on its customers’ core needs.19 (See Exhibit 3a for details on each business unit;
Exhibit 3b provides financials.) In September 2013, GE reported a backlog of $223 billion, equivalent
to more than a year and a half of company revenues. Service contracts continued to make up about

three-quarters of the backlog, and services still contributed about 75% of industrial earnings.20 (See
Exhibit 4 for a comparison of services and products.) In 2013, software and software-enabled services

contributed about $4 billion annually to GE’s total revenues.21 And in August, the company
announced that it would spin off parts of GE Capital in one form or another in 2014, reflecting
Immelt’s continued commitment to strengthen GE’s industrial side and his pledge that by 2015, 65%
of GE profits would come from industrials.22

In 2013, GE’s scale and scope meant that literally billions of its devices and machines were
spinning in operation around the globe. Total assets of global operations on a continuing basis were
$337.6 billion in 2012. GE produced aircraft engines, locomotives and other transportation equipment,
kitchen and laundry appliances, lighting, electric distribution and control equipment, generators,
turbines, medical imaging equipment, mining equipment, oil and gas equipment, along with a host of
commercial finance, insurance, real estate, and energy-leasing products, touching virtually every
corner of the globe.23

b Net promoter scores measured the percentage of people who said they would recommend GE to a friend, minus those who
would not..

c Backlog referred to the value of unfulfilled s.

d The Virginia center, opened in 2011, focused on cybersecurity; in 2009, GE launched its advanced manufacturing-software
center in Michigan, focused on avionics and jet engine production, and internal production improvements. See Kate
Linebaugh, “GE Makes Big Bet on Software Development,” Wall Street Journal, November 17, 2011,
http://online.wsj.com/news/articles/SB10001424052970204517204577042532750345206, accessed January 2014.

This document is authorized for use only by John Rome ([email protected]). Copying or posting is an infringement of copyright. Please contact [email protected] or
800-988-0886 for additional copies.

http://online.wsj.com/news/articles/SB10001424052970204517204577042532750345206

GE and the Industrial Internet 614-032

5

The Industrial Internet: A New Opportunity?

With the early Internet, we never imagined the implication of a billion people being connected, so when 50
billion machines become connected . . .

— Bill Ruh, Vice President, GE Software

Sizing the Opportunity

The Internet of Things encompassed all connected devices—from consumer to industrial. Cisco
estimated that there were about 9 billion connected devices in 2010 and that the number would grow

to 50 billion by 2020.24 Quantifying the size of the Industrial Internet was challenging, with
microprocessors, sensors, and other software components already embedded in most industrial
assets. One industry player expected the Internet of Things to create $14.4 trillion in economic value
between 2013 and 2022; another expected that $1.9 trillion of economic value would be added in 2020

alone.25 Others pegged the value of Industrial Internet–related activity at about $23 billion in 2012,
with growth projected to hit $1.3 trillion by 2020.26 These analysts projected Industrial Internet–

related technology spending to reach $514 billion.27

By 2011, GE’s assets also had significant embedded software, along with sensors and
microprocessors, running power plants, jet engines, hospitals and medical systems, utility companies,
oil rigs, and rail and other industrial infrastructure worldwide. GE predicted that smart components
would save the oil and gas industry more than $90 billion a year through reduced operating costs and

fuel consumption,28 while improved resource usage and tracking and status of equipment could save
the health-care sector $63 billion from efficiency gains (see Exhibit 5a and 5b).29 GE projected that
efficiency gains as small as 1% could have huge benefits over time when scaled up across the
economic system (refer to Exhibit 1).

The decision to build an Industrial Internet never gave Immelt pause; he explained, “I have a great
deal of confidence in our core hardware. We have the most stuff. It’s hard to replicate. We started
from a real position of relative strength.” It was GE’s huge scale and massive installed base that could
speed its vision of the Industrial Internet to reality. As one GE researcher noted, “We have some of
the biggest industrial data sets, because we’ve been operating this equipment for a long time. . . . We

have the before and after and can test any algorithm and see how it works.”30 The analytic
applications GE could build using these data could qualitatively enhance a customer’s business. In
2009, GE Transportation had released software that railroads used to calculate the weight and length
of a train, topography, speed limits, and other variables in to avoid unnecessary braking,

saving an average class-1 U.S. rail company $150 million annually in fuel savings.e According to one
GE insider, with over 400 data points collected continuously in one wind turbine, for example, or tens
of thousands of data points in one wind farm, “We have an opportunity, as we think about reliability,
availability, and performance, to use that data to come up with better product and service offerings
and get more out of the existing fleet [of wind turbines].”31 Describing the opportunity, one observer
wryly noted:

Not only is GE allowing a zillion smart-machine flowers to bloom, but its vision is to
use the data from these machines, not in isolation, but in aggregate to create value for its
customers and improve the lives of individuals, corporations, governments and
humankind in general. That’s big. Supply chains. Health. Instrumenting infrastructure.
Supporting the build out of massive cities in growth regions. It’s mind-boggling big.32

e This assumed 10% fuel savings, $75,000 per locomotive per year, and 2,000 locomotives for an average railroad company.

This document is authorized for use only by John Rome ([email protected]). Copying or posting is an infringement of copyright. Please contact [email protected] or
800-988-0886 for additional copies.

614-032 GE and the Industrial Internet

6

Collectively, GE’s various business units employed over 8,000 software professionals that helped
generate several billion dollars in software revenue alone. However, there was no overarching
strategy guiding their technical choices and commercial offerings. Thus, each business unit and even
each product leader optimized software choices to the local conditions, resulting in significant
heterogeneity in technical and commercial success. The Industrial Internet initiative required new
types of software technology and required a more coordinated approach. The advent of cloud
computing added more opportunities and threats. As the scale and scope of the Industrial Internet
opportunity across GE’s businesses became clear, two things became evident to headquarters staff:
GE needed a global software center to develop and support emerging software applications
uniformly across the businesses and needed new and innovative approaches to managing customer
relationships, including how to sell and service the new offerings.

How to Build an Industrial Internet: GE Software

Based in San Ramon, California, and headed by Ruh, the inception of GE Software was first
announced in November 2011 as a part of GE’s Global Research Center (corporate research and
development laboratories). Ruh explained, “Co-location is everything. Distributed software
development in my view doesn’t work, although we still do some. Our fastest-developing programs
are in one place. New things are easier to create a team around when they are all in one place.” GE
Software was funded by Immelt and GE’s Global Research, and did not have its own profit-and-loss
statement (P&L). Immelt had very specific criteria for GE Software’s leadership; as Ruh recalled,
“They wanted someone with a background in innovative software development; someone who was
service oriented versus product development; and someone able to manage a start-up–like
environment in a very large, complex company.” Ruh had had a long career in software, working
closely with federal agencies early on to build systems that transmitted images from space and
tackling early digitization problems. He had focused on secure integration of large business-to-
business systems and had worked in analytics at the Central Intelligence Agency. He gained expertise
in large systems-based networking at several large companies, eventually landing at Cisco, where he
helped create products around new services and eventually ran service product management for all
of Cisco as the chief architect of advanced services.

By mid-2011, Immelt had attracted Ruh to head GE Software. “They made it clear this was not
about building product,” Ruh said. “To do this right, they said, ‘You have to understand our service
business.’” Ruh started out as the sole employee, with a temporary office in San Ramon. (See Exhibit
6 for map of the region.) He hired designer Greg Petroff early on to head user experience (UX) for GE
Software and tasked him with developing a system that could bring all of GE’s machines “onto one
cloud-connected, contextually aware, super-efficient platform.”33

By January 2013, Ruh had hired 62 employees, and by June, about 150 employees moved into the
new offices.34 By the end of 2013, Ruh expected to have 350 employees, with transfers from other
parts of GE making up less than 2% of his team. Chief Technology Officer Mark Little, senior vice
president, director of GE Global Research, said, “Bill Ruh was one of the few with the domain
expertise and leadership who could come into this big complicated company and move us from the
notion of doing something interesting and put us on a well-defined path. It’s remarkable how far
we’ve come in the few short months he’s been here.”

Ruh and senior leadership focused GE Software’s next task: getting a handle on the scope of GE’s
existing software globally, most of which was embedded and customer-facing, and the developers
behind it. Ruh said, “Every one of our products had a different underpinning platform, architecture,
technology, and set of vendors.” A survey of the 136 existing GE software offerings revealed further

This document is authorized for use only by John Rome ([email protected]). Copying or posting is an infringement of copyright. Please contact [email protected] or
800-988-0886 for additional copies.

GE and the Industrial Internet 614-032

7

complexity. He added, “Only 17 of our software offerings were generating profits. We began to look
at the problem: it was taking us years to build the software, and years to get it out the door. And
customers’ needs were changing too rapidly to keep up. An additional problem was that a large
percentage of the development cost was being spent on aspects of the core platform rather than on
the application itself.”

An example from the Oil and Gas business unit illustrated the challenges. Everything on an oil rig
was monitored locally, on the rig. “That’s where the data is,” said Paul Rogers, GE Software head of
software development. A customer approached GE about developing a tool that could monitor some
underwater switches remotely, via the cloud. The customer requested that Oil and Gas engineers
develop a simple tool to measure when a sensor was on or off, saving a physical visit to each rig to
collect the data. The engineers made a proposal and promised product delivery within 18 months.
Three years later, there was still no deliverable, and the vision for the product had expanded to
incorporate 5,000 “cool” features, Rogers recalled. The group came to Ruh and his new team at GE
Software asking for help. In three months, the team had a low-cost solution. Rogers said, “The project
they were originally working on didn’t even have the customer’s initial vision in it. They thought that
was too boring. There’s a legacy mind-set here that ‘It can’t be valuable if it’s not hard.’”

Developer talent had to be addressed. Ruh explained, “Our software engineers had experience in
one of two ways: they were either mechanical engineers or they were computer scientists. But most of
them had experience with technologies that were last generation. They were very reliant on outside
vendors, sometimes for full development. And when you’re building products and services, you
have to own the intellectual property, or IP.” Further, there was no common language among the
ranks of GE’s existing software experts. Brad Surak, general manager of Industrial Internet Programs
at GE Software, said, “In other organizations, you may not agree, but you speak the same language.
Here we didn’t even speak the same language.” Finally, many at GE were skeptical about moving
services and capabilities to the cloud, which the initiative would rely on extensively. Immelt
immersed himself in the technology behind GE’s platform. His familiarity with the pros and cons of
the cloud “allowed us to move a lot quicker,” said Surak. “Software at GE had a culturally different
feel than the world of applications [apps] and services we were trying to move into. It was a bit more
conservative. We faced a lot of difficulties convincing people that moving some of our capabilities to
the cloud was important. So Immelt’s engagement and support was critical.”

As they moved past this initial assessment, Ruh and his team found that some GE businesses were
more ready to collaborate than others. The businesses were huge and global, …