When do Low End Innovations Become Disruptive Innovations?

When do Low-End Innovations become Disruptive Innovations?

Jeffrey FunkAssociate Professor

Division of Engineering & Technology ManagementNational University of Singapore

Mail: [email protected]

These slides summarize ideas that are described in What Drives Exponential Improvements, California ManagementReview, Spring 2013. Technology Change and the Rise of New Industries, Stanford University Press, and “What Drives Improvements in Performance and Cost,” also on my slide share account

Why is this Issue Important? (1)

• Some people assume that low-end innovations naturally become disruptive innovations

• This assumption grows from problems with conventional wisdom on technology change (more later)– costs fall as cumulative production increases and as firms

introduce equipment and organize it into flow lines (Utterback, 1994)

– Christensen: emergence of a niche product leads to investment and thus improvements in both performance and costs

• This assumption becomes more common when – disruptive innovations are called “low-end innovations”– radical innovations are defined as “high-end innovations”


• Calling a disruptive innovation a low-end innovation– Focuses students’ attention on the characteristics of initial product

and market and not on the• degree of similarity between markets (Adner, SMJ, 2002)• technology’s potential for improvements

– For example, some argue the keys to creating disruptive innovations are doing such things as

• miniaturizing products and• reducing the number of features

• Calling a radical innovation a high-end innovation– Makes it hard for students to understand the importance of the

concept that forms the basis for new technologies (Henderson and Clark, 1990) and thus the

• technology’s potential for improvements

– For example, low-end phones do not have potential for diffusion unless they are based on a new concept that provides potential for eventual superiority in multiple segments

Transitive Property: If a=b and b=c, then a=c

a:low-endinnovation

c: the following phenomenon

b:disruptiveinnovation

If we define disruptive innovations (b) as low-end innovations (a) and the phenomenon of disruptive innovations is represented by the above figure (c), then low-end innovations automatically replace the mainstream technology as shown in (c)


• Because my students had learned these things ……,– 1/3 of my students made the following

argument: “technology A is inferior to the existing technology and thus is a disruptive innovation and will replace the existing technology”

– and seemed to believe that no further analysis needs to be done

One Final Example: Case of “Double Disruption

• Is a combination two-low end innovations even better than one?

• For example, should producers of solar cells focus on – ones with poor efficiency (e.g., ones based on

photosensitive dyes) because they are inferior to high-efficiency ones

– users of low-end electrical devices such as electric bicycles (they have lower performance than do automobiles)?

– By selling low-end solar cells to users of electric bicycles, producers of solar cells will somehow benefit from a “double disruption” when the electric bicycles naturally replace automobiles and the low efficiency cells naturally replace the high efficiency ones

Don’t be Fooled by Hype

• Many people offer simple models that sound good

• But these simple models don’t accurately represent the phenomenon

• There is a better way: Let’s understand when low-end innovations might become disruptive innovations by understanding what drives improvements in cost and performance

Outline• Conventional Wisdom on Drivers of Improvements,

i.e., Technological Change• What drives improvements in cost and

performance?

– Creating materials to exploit physical phenomena– Geometrical scaling– Some technologies directly experience improvements

through the two mechanisms while others indirectly experience them through improvements in specific “components”

• Implications for Disruptive Innovations• Summary

Conventional Wisdom on Drivers of Improvements

• Costs fall as cumulative production grows in learning or experience curve as automated manufacturing equipment is – introduced and organized into flow lines

• Implications: stimulating demand will lead to cost reductions. This is one reason why many governments subsidize the introduction of clean energy more than they subsidize R&D spending

• Clayton Christensen’s theory of disruptive innovation also implies that increases in demand will naturally lead to reductions in cost and improvements in performance

Problems with Learning Curve• Can’t use learning curve until production has begun• Learning curve assumes all components are unique to new product• Learning curve doesn’t help us understand why some technologies

experience more improvements in cost and performance than do other technologies

• An emphasis on cumulative production – focuses analyses on the production of the final product– implies that learning done outside of a factory is either unimportant or is

being driven by the production of the final product

• But many cost reductions or performance improvements are the result of activities done outside of the factory– advances in technology or science done in laboratories– reductions in scale (e.g., ICs) or increases in scale (e.g., oil tankers)– improvements in complementary technologies such as components whose

demand are being driven by other systems



performance?

– Creating materials to exploit physical phenomena– Geometrical scaling– Some technologies directly experience improvements



What drives improvements? • Some technologies have the potential for larger

improvements in cost and performance than do other technologies. Improvements come from– Creating materials to exploit physical phenomena– Geometrical scaling

• When a technology has a strong impact on performance and cost of higher-level system and the technology experiences large (e.g., exponential) improvements in performance and cost, such a technology can – drive large improvements in cost and performance of system– and lead to or facilitate discontinuities in systems

• For students– When might a technology offer a superior value proposition and for

what customers?

Luminosity per watt (lm/W) of lights and displays

OrganicTransistors

TechnologyDomain

Sub-Technology

Dimensions of measure

Different Classes of Materials

Energy Trans-formation

Lighting Light intensity per unit cost

Candle wax, gas, carbon and tungsten filaments, semiconductor and organic materials for LEDs

LEDs Luminosity per Watt

Group III-V, IV-IV, and II-VI semiconductorsOrganic LEDs Small molecules, polymers, phosphorescent materials Solar Cells Power output

per unit costSilicon, Gallium Arsenide, Cadmium Telluride, Cadmium Indium Gallium Selenide, Dye-Sensitized, Organic

Energy storage

Batteries Energy stored per unit volume, mass or cost

Lead acid, Nickel Cadmium, Nickel Metal Hydride, Lithium Polymer, Lithium-ion

Capacitors Carbons, polymers, metal oxides, ruthenium oxide, ionic liquids

Flywheels Stone, steel, glass, carbon fibersInformation Trans-formation

Organic Transistors

Mobility (cm2/ Volt-seconds)

Polythiophenes, thiophene oligomers, polymers, hthalocyanines, heteroacenes, tetrathiafulvalenes, perylene diimides naphthalene diimides, acenes, C60

Living Organisms

Biological transformation

U.S. corn output per area

Open pollinated, double cross, single cross, biotech GMO

Materials Load Bearing Strength to weight ratio

Iron, Steel, Composites, Carbon Fibers

Magnetic Strength Steel/Alnico Alloys, Fine particles, Rare earthsCoercivity Steel/Alnico Alloys, SmCo, PtCo, MaBi, Ferrites,

Different Classes of Materials were found for Many Technologies

New Processes are Often Key Part of Creating New Materials

• New materials for electronics usually involve new processes– Semiconductor ICs, MEMS, bio-electronic ICs,

nanotechnology, lighting, displays, batteries

• Radical new processes have also played a role in more traditional industries– Bessemer process, basic oxygen furnace, and

continuous casting for steel– Haber-Bosch process for ammonia

– Float glass process

– Hall–Héroult process for aluminum

Incremental Improvements to these processes are also important

• Learning curve emphasizes small changes to the processes, which do play a role in achieving improvements

• But small changes to the processes can’t explain exponential improvements in performance

• Without new materials and most importantly new classes of materials, exponential improvements would not be achieved



performance?

– Creating materials to exploit physical phenomena– Geometrical scaling: both smaller and larger scale– Some technologies directly experience improvements



Geometric Scaling (1)• Definition

– relationship between the technology’s core concepts (Dosi, 1982; Henderson and Clark, 1990), physical laws and dimensions (scale), and effectiveness

– “scale effects are permanently embedded in geometry and physical nature of the world in which we live” (Lipsey et al, 2005)

• Studied by some engineers, but only within their discipline– chemical engineers: chemical plants (many references)– mechanical engineers: engines, tankers, aircraft (fewer references) – electrical engineers: integrated circuits, magnetic and optical

storage (many)• But few references (and even fewer analyses) by management or

economic scholars (Nelson and Winter, 1982; Sahal, 1985; Rosenberg, 1994; Freeman and Soete, 1997; Lipsey et al, 2005; Winter, 2008)

Geometric Scaling (2)

• For technologies that benefit from smaller scale– the benefits can be particularly large, since

• costs of material, equipment, factory, and transportation typically fall over long term as size is reduced

• but performance of only some technologies such as ICs and magnetic storage experience increases in performance as size is reduced.

• placing more transistors or magnetic or optical storage regions in a certain area can increase speed and functionality and reduce both power consumption and size of final product

• For technologies that benefit from larger scale– output is roughly proportional to one dimension (e.g., length

cubed or volume) more than is the costs (e.g., length squared or area) thus causing output to rise faster than do costs, as the scale of technology is increased

Figure 2. Declining Feature Size

0.001

0.01

0.1

1

10

100

1960 1965 1970 1975 1980 1985 1990 1995 2000

Year

Micr

omete

rs (M

icron

s)

Gate OxideThickness

Junction Depth

Feature length

Source: (O'Neil, 2003)

Areal RecordingDensity of Hard DiskPlatter

Example of Benefits from Larger Scale: Engines

Diameter of cylinder (D)

Cost of cylinderor piston is function of cylinder’s surface area (πDH)

Output of engineis function ofcylinder’s volume (πD2H/4)

Result: output risesfaster than costs asdiameter is increased

Heightofcylinder(H)

http://en.wikipedia.org/wiki/File:Cylinder_2_(PSF).png

1

10

100

1000

10000

1 10 100 1000 10000Output (Scale)

Pri

ce p

er O

utpu

tPrice Per Output (Horsepower)

Marine EngineLargest is 90,000 HP

1

10

100

1000

10000

0.1 1 10 100 1000 10000

Rel

ativ

e P

rice

per

Ou

tpu

tRelative Price Per Output Falls as Scale Increases

Steam Engine (in HP) Maximum scale: 1.3 M HP

Marine EngineLargest is 90,000 HP

Chemical Plant: 1000s of tons of ethyleneper year; much smaller plants built

Commercial aircraftSmallest one had

12 passengers

Oil Tanker:1000s of tonsSmallest was

1807 tons

Output (Scale)

LCD Mfg Equip: Largest panel size is 16 square meters

Aluminum(1000s of amps)

Electric PowerPlants (in MW); much smaller ones built



performance?




Improvements in Computations Per Second (Koomey et al, 2011)

Why do computers experience improvements in processing speed?Are these large (or small) improvements in processing speed?How many other products experience such large improvements?

Components and Systems (1)

• Some components have a large impact on performance of a system

• Components that benefit from scaling can– have a large impact on performance and cost of systems,

even before system is implemented– lead to changes in relative importance of cost and

performance and between various dimensions of performance

– lead to discontinuities in systems

• These components may have a larger impact on performance and cost than – novel combinations of components

Components and Systems (2)• Improvements in engines impacted on

– Locomotives, ships– Automobiles, aircraft

• How about solar cells? • Improvements in ICs impacted on

– computers, servers, routers, telecommunication systems and the Internet

– radios, televisions, recording devices, and other consumer electronics

– mobile phones and other handheld devices– controls for many mechanical products

• Improvements in ICs led to many discontinuities in systems

Components and Systems (3)• Improvements in ICs are still driving the

emergence of new electronic systems such as new forms of– Computers (e.g., tablet computers)– networks of RFID tags, smart dust, and other sensors – Cloud/utility computing– Internet content (e.g., mashups, 3D content, video

conferencing)– human-computer interface (touch, gesture, neural)– Mobile phones– Mobile phone systems (e.g., 4G, 5G, cognitive radio)– Autonomous vehicles– Holographic display systems

Components and Systems (4)

• Similar things are happening with bio-electronics, MEMS, nanotechnology: they are enabling new forms of systems to emerge– point-care diagnostic devices– Other forms of sensors and sensor-based systems– Even new forms of mobile phones

• Better forms of DNA sequencers and synthesizers are being driven by reductions in scale of features. They will impact on higher-level systems

For these and other Technologies

• What is the minimum level of performance in a component (such as an IC) that might enable a new electronic system to offer a superior value proposition in for example,– Gesture and neural-based human-computer interfaces?– Cognitive radio for mobile phone systems?– Autonomous vehicles?

• When the concepts and principles that form the basis for a new system are relatively well known, components are often the bottleneck for new systems– This is the case for many technologies



performance?




All the Disruptive Innovations in Christensen’s 1997 Book Exhibit Geometric Scaling

System Component Geometric

Scaling

Computers ICs In component

Hard Disk Drives Platter In component

Retail Outlets (& their info systems)

ICs (via impact on computers)

In component

Mechanical Excavators

Hydraulic actuators

(has piston, cylinder)

In component

Electric Arc Furnace (Mini-mills)

Furnace In system

Geometric Scaling in Disruptive Innovations from Christensen’s Other Publications

• “The Great Disruption,” article in Foreign Affairs– transistors and ICs in transistor radios and TV– magnetic tape in audio and video recording equipment

• The Innovator’s Prescription– advances in medical science and improvements in information

technology (IT), of which latter depends on scaling in ICs, enable precision medicine to be implemented and this precision medicine enables health care to depend more on low-wage nurses (and patients) than on high-wage doctors

– In addition to better computers, this IT also comes in the form of better imaging, molecular medicine, and biochemistry.

• Disrupting Class– improvements in information technology also enable more

customized learning than do existing methods

When do Low-End Innovations become Disruptive Innovations? (1)

• When the needs in market segments are similar– Ron Adner (2002, Strategic Management Journal) calls

this “preference overlap”– Low-end products are more likely to diffuse across

segments when there is high preference overlap across segments

• When the technology has high potential for improvements, as characterized by rapid improvements through– Creating materials to exploit physical phenomena– Geometrical scaling

When do … Disruptive Innovations? (2)

• When improvements in a component have bigger impact on a low than high-end product

• Low-end transistor radios and televisions replaced high-end ones because – transistors and ICs were initially more appropriate for low-end

radios and televisions– improvements in these transistor and ICs directly improved cost

and performance of initially low-end radios and televisions. • Low-end magnetic disks replaced magnetic cores and

drums because– improvements in magnetic recording density impacted more on

their performance and costs than on those of magnetic cores and drums.

• Improvements in Internet have much larger impact on performance of SaaS, which has started with low-end users, than on performance of packaged software– thus these improvements may cause SaaS to replace packaged

software

When do ….. Disruptive Innovations? (3)

• When improvements in components impact on a dimension of performance for which technology overshoot may occur

• Increases in recording density of hard disk platters – caused hard disk capacity to overshoot needs of most users– this facilitated the replacement of large with smaller hard disks

• Improvements in magnetic recording density of tape– caused high-end systems to overshoot needs of most users– this facilitated their replacement with low-end systems such as

VHS

• Increases in number of transistors per chip caused – mainframe computers to overshoot the needs of most users in

terms of processing speeds– this facilitated their partial replacement with mini- and personal

computers

Example of when they do not• Improvements in ICs, LCDs and other components enable

– improvements in applications and user interface of high-end smart phones and thus drive their diffusion

– Unlikely that low-end mobile phone (unless it is based on completely new concept such as wireless LAN) will form basis of a product that displaces current mobile phones

• Key point is not whether mobile phone is low-end or not, it is whether– mobile phone is based on technology with more potential for

improvements than existing technology• This example reinforces why it is dangerous to call a

disruptive innovation a low-end innovation (and a radical innovation a high-end innovation)– If students hear about high- and low-end and not changes in

concepts and architectures– they will not understand that key question is whether the low-end

innovation involves a change in concept or architecture that will enable it to have large potential for improvements

Summary (1)• Technologies that experience large improvements in

performance and cost are more likely to form the basis for new industries than are other technologies

• The following concepts provide a better understanding of why and how improvements occurred than does the learning curve, i.e., increases in cumulative production – Creating materials to exploit physical phenomena– Geometrical scaling

• We (including students) can use these concepts to think about – when a new technology might offer a superior value proposition– whether that technology is appropriate for low-end or high-end

market– whether a low-end innovation will become a disruptive innovation

Summary (2)

• We need to help students understand when a new technology might offer a superior value proposition in order for them to– create new businesses– understand the limitations of proposed solutions to global problems– come up with better solutions

• For new technologies, students need to assess– Current advantages and disadvantages– Sources and rates of improvement– Do we expect these improvements to accelerate or de-accelerate?

• Let’s give students the necessary tools for them to– design their own future– including find disruptive technologies

Business

When do Low End Innovations Become Disruptive Innovations?