When Do New Technologies

Become Economically Feasible?

The Case of Electronic Products

A/Prof Jeffrey Funk

Division of Engineering and Technology Management

National University of Singapore

For information on other technologies, see http://www.slideshare.net/Funk98/presentations

Some Theory: Modeling

Improvements in Products

Learning curve

Costs fall as cumulative production rises

Assumes most improvements occur on factory floor

Experience curve

Costs fall and performance rises as cumulative production increases

Assumes that increasing cumulative production encourages product

and process R&D, thus leading to improvements

But which improvements are most prevalent?

Learning in factories, product and process R&D?

We analyzed technologies that experienced rapid rates of

improvement before commercial production began in order to

eliminate factory floor learning

Rapid Improvements with No Commercial

Production: How do the Improvements Occur?

Analyzed how improvements in cost and performance

occurred for 13 technologies that

experienced rapid improvements (>10%/year)

without commercial production

13 technologies

Organic transistors, solar cells, LEDs; Quantum dot displays and

solar cells; Superconducting Josephson Junctions, Quantum

computers; Superconducting cables, Carbon nano-tube

transistors; Non-volatile memory: PRAM, MRAM, FeRAM, ReRAM

Most common methods of improvement

New materials (and new processes for them)

Reductions in scale of featuresSource: Funk J and Magee C 2015. Rapid Improvements with No Commercial Production: how do the improvements occur? Research Policy

Are there other Technologies

that Contradict the Learning

and Experience Curves?

Rapid improvements occur but…

Improvements in assembly or

product and process R&D for the

final product cannot be the

explanations for the rapid

improvements

What Happens When Cost of Standard

Components are far Higher than Cost

of Assembly Operations?

Low contribution of assembly to total costs means that learning in assembly or process R&D for assembly cannot be key sources of improvements

Standard components are used

by multiple suppliers of end products and/or

in different types of products from the same (or different) end product supplier

Large contribution of standard components

Means that opportunities for product R&D are limited

Improvements in the end products come mostly from improvements in the standard components

Nine Electronic Products Were Analyzed

Smart Phones

Tablet computers

Laptop

computers

eBook

Readers

Game Consoles

MP3 Players

Large screen TVs

Internet TVs

Google

Glass

Example of Cost Data Collected

Type of

Product

Final Assembly Standard Components1

Number of

Data Points

Average

(%)

Number of

Data Points

Lower Estimate for

Average2 (%)

Smart Phones 28 4.2% 26, 28 76%, 79%

Tablet

Computers

33 3.1% 33, 33 81%, 84%

eBook Readers 6 4.9% 6, 9 88%, 88%

Game Consoles 2 2.4% 2, 2 64%, 70%

MP3 Players 2 3.4% 2, 9 78%, 80%

Large Screen

Televisions

2 2.4% 2, 2 82%, 84%

Internet TVs 2 5.7% 2, 2 57%, 61%

Google Glass 1 2.7% 1, 1 62%, 64%

Cost Breakdown for Electronic Products

1 Values as a percent of total and material costs

2 Excludes mechanical components, printed circuit boards, and passive components

Preliminary Conclusions

Cost of assembly is very low

Improvements in assembly operations or process R&D

for assembly cannot be important sources of cost

reductions

Contribution of standard components is very high

Design space and opportunities for product R&D are

limited

Demand for the standard components are not driven

by a single end product

What is Driving the Improvements?

Let’s look in more detail at the standard components

in these products

Type of

Product

# of

Data

Point

Memory Micro-

Proc-

essor

Display Cam-

era

Connect-

ivity,

Sensors

Bat-

tery

Power

Mgmt

Phones 23 15% 22% 22% 8.2% 7.9% 2.3% 3.8%

Tablets 33 17% 6.6% 38% 2.9% 6.3% 7.3% 2.5%

eBook

Readers

9 10% 8.1% 42% 0.30% 8.3% 8.3% Not

available

Game

Console

2 38% 39% none none Not

available

none 5.8%

MP3

Players

9 53% 9% 6% none Not

available

4% 3.5%

TVs 2 7% 4.0% 76% none Not avail. none 3.0%

Internet

TVs

2 16% 31% none none 10.5% none 3.5%

Google

Glass

1 17% 18% 3.8% 7.2% 14% 1.5% 4.5%

Contribution of “Standard Components” to Costs

of Selected Electronic Products

Summary of Previous Slide Processors and Memory (including DRAM, SRAM, flash, hard

disks, CDs) represent high percentage of costs

Game consoles (77%), MP3 players (53%)

Internet TVs (47%), Smart phones (37%)

For smart phones, multiple processors

Internal processing of music, video, apps

Processing of cellular network signals (and WiFi)

Displays represent largest percentage in

Large screen televisions (76%)

Tablet computers (38%)

eBook Readers (42%)

All of these components experience rapid improvements

Processors, memory (Moore’s Law, 40% per year), camera: 30 to 50% per year, displays: 12% per year (see next slide)

Camera Chip Price

Examples

of

Rapidly

Falling

Costs

0.0001

0.01

1

100

1950 1970 1990 2010

Dynamic

Random

Access

Memory

Flash

Memory

h. Millions of Memory Bits/Dollar

vs. Time

Interpretation

A small number of standard components play important role in electronic products and improvements in them

Enable improvements

Determine new types of functions

This is why many people emphasize Moore’s Law

Because it is really changing our world

To investigate the role of these components in more detail, we now consider two types of products

Smart phones

Tablet computers

Measure iPhone iPhone 3G iPhone 4 iPhone 5 iPhone 6

Operating

System

1.0 2.0 4.0 6.0 8.0

Flash Memory 4, 8, 16GB 8 or 16GB 8, 16, 64GB 16, 32, 64GB 16, 64, or 128GB

DRAM 128MB 128MB 512MB 1GB 1GB

Application

Processor

620MHz Samsung 32-bit RISC 1 GHz dual-

core Apple A5

1.3 GHz dual-core

Apple A6

1.4 GHz dual-core

Apple A8

Graphics

Processor

PowerVR MBX Lite 38 (103 MHz) PowerVR

SGX535 (200

MHz)

PowerVR

SGX543MP3 (tri-

core, 266 MHz)

PowerVR GX6450

(quad-core)

Cellular

Processor

GSM/GPRS/

EDGE

Previous plus

UMTS/HSDPA

3.6Mbps

Previous plus

HSUPA

5.76Mbps

Previous plus LTE,

HSPA+, DC-HSDPA,

4.4Mbps

Previous plus LTE-

Advanced, 14.4Mbps

Display

resolution

163 ppi (pixels per inch) 326 ppi 401 ppi

Camera

resolution

Video speed

2 MP (mega-pixels) 5 MP

30 fps, 480p

8 MP

30 fps at 1080p

8 MP

60 fps at 1080p

WiFi 802.11 b/g 802.11 b/g/n 802.11 a/b/g/n 802.11 a/b/g/n/ac

Other Bluetooth 2.0 GPS,

compass,

Bluetooth

2.1,

gyroscope

GPS, compass, Blue-

tooth 4.0, gyroscope,

voice recognition

Previous plus finger-

print scanner, near-

field communication

Evolution of iPhone in Terms of Measures of Performance

Fps: frames per second

480p: progressive scan of 480 vertical lines

Summary of Previous Slide More memory enables more data to be saved

Songs, pictures

Videos, games, apps

Faster processors means

More sophisticated apps, games and cellular networks, the latter enables higher speeds

Higher resolution audio, displays, video, cameras

Faster and newer WiFi and Bluetooth chips

Mean higher data speeds

Better displays means higher resolution video, pictures

New functions come from new components

Compasses, gyroscopes, voice recognition

Finger-print scanners, near-field communication

For the first iPhone

What Levels of Performance and cost

were needed in each Component

before the iPhone was economically

feasible?

Touch Screen and Overall Display

DRAM and Flash Memory

Microprocessors

Can we use such an analysis to better

understand the future?

Let’s look at flash memory

The 4GB iPhone could store

760 songs, 4000 pictures (4 megapixel JPEG),

four hours of video, or 100 apps/games, or

some combination of them

Equal usage

190 songs

1000 pictures

one hour of video

25 apps/games

Was 4GB of flash memory necessary, or would

less have been sufficient?

The Average User Downloaded 58 Apps or a Significant

Fraction of Memory Available in 4GB Phone

Sensitivity Analysis of Flash Memory Price

Cost of iPhone 5 varied from $207 to $238 depending

on flash memory capacity

16GB, 32GB, or 64GB

For iPhone 4s, costs range from $196 to $254 for same

range in flash memory

For iPhone 3GS, 16GB of flash memory are $24 thus

suggesting costs for same change in capacity would

range from $179 to $251

In percentage terms, same changes in flash memory

capacity led to increase of 40% in iPhone 3GS and

increase of only 15% in iPhone 5

Interpretation of Previous Slides

Improvements in flash memory were

essential for the iPhone to become

economically feasible

Similar analyses could be done for

microprocessors and displays

And would probably show similar results

Let’s now look at the iPad

Measure iPad iPad2 iPad3 iPad4 iPad Air iPad Air 2

Operating System 5.1.1 iOS 8

System on Chip Apple A4 Apple A5 Apple A5X Apple A6X Apple A7 Apple A8X

Application

Processor

1 GHz ARM

Cortex-A8

1 GHz dual-core ARM

Cortex-A9

1.4 GHz dual-

core Apple

Swift

1.4 GHz dual-

core Apple

Cyclone

1.5 GHz tri-

core

Graphics Processor PowerVR

SGX535

Dual-core

PowerVR

SGX543MP2

Quad-core

PowerVR

SGX543MP4

Quad-core

PowerVR

SGX554MP4

Quad-core

PowerVR

G6430

Octa-core

PowerVR

GXA6850

Flash Memory 16, 32, or 64 GB 16, 32, 64, or 128 GB 16, 64, 128 GB

DRAM 256 MB 512 MB 1 GB 2GB

Display 132 ppi 264 ppi

Camera resolution,

video speed, digital

zoom

None .7 MP, 30fps

5 times

5 MP, 30fps,

5 times

8 MP, 30 fps

3 times

Wireless without

cellular

Wi-Fi 802.11a/b/g/n;

Bluetooth 2.1

Wi-Fi 802.11a/b/g/n; Bluetooth 4.0 802.11a/b/g

/n/ac

Bluetooth 4.0

Wireless w/cellular Above plus 2G EDGE, 3G

HSDPA

Above and left plus LTE

Geolocation

without cellular

WiFi, Apple location database Previous plus

iBeacon

Geolocation with

cellular

Assisted GPS, Apple

databases, cellular network

Previous plus GLONASS (Russian-based GPS) Previous plus

iBeacon

Other Accelerometer,

light sensor,

magnetometer

Previous plus gyroscope Previous plus

barometer

Evolution of iPad in Terms of Measures of Performance

Summary of Previous Slide for iPad

More memory enables more data to be saved

Songs, pictures

Videos, games, apps

Faster processors means

More sophisticated apps, games and cellular networks, the

latter enables higher speeds

Higher resolution audio, displays, video, cameras

Faster and newer WiFi and Bluetooth chips

Mean higher data speeds

Better displays means higher resolution video, pictures

New functions come from new components

Accelerometers, light sensors, magnetometers, compasses,

gyroscopes and barometers

For the first iPad

What Levels of Performance and cost were

needed in each Component before the iPad

was economically feasible?

Touch Screen and Overall Display

Memory

Microprocessors

Can we use such an analysis to better

understand the future?

Let’s look at display, memory and

microprocessors

For the first iPad (2)

Touch Screen and Overall Display

Much larger and thus more expensive than one for iPhone – probably was a bottleneck

But falling cost (12%), price 20%) of displays

Memory and microprocessors

Cost of iPad Air varies from $274 to $331 depending on flash memory capacity (16GB, 32GB, 64GB) and whether cellular processor is included

For earlier iPads

From $316 to $406

From $229 to $346

Percentages for memory and microprocessors drop from 50% in first iPad to 21% in iPad Air

Did the first iPad need a cellular processor?

Interpretation of Previous

Slides

Improvements in displays, memory and

microprocessors were essential for the

iPad to become economically feasible

Let’s now think about the future – for

smart phones, what will be next?

For Smart Phones, What will be Next?

What components are experiencing rapid

improvements?

Can they tell us something about the “next big thing”

Improvements will probably continue in

Microprocessor, memory and other ICs

MEMS, bio-electronic ICs

Displays including flexible ones

Lasers, LEDs, photo-sensors, and other sensors

Speeds of cellular networks and WiFi

New forms of user interfaces (gesture, touch)

Open source software is becoming more available

What will be Next? (1)

New features, perhaps for high-end phones

Health care: phones monitors health (heart rate, brain wave, blood pressure) using sensors

Home automation: use phones to control homes

Better navigation, sharing economy

Engineering assistant: environmental data (temperature, pressure, air and water quality) and also data from satellites

Different phones for different applications?

One phone does everything?

Multiple segments each with multiple applications for phone?

Specific phones must be defined for specific users

What will be Next? (2)New forms of computers

Wearable computing?

Smart watches?

Wrist displays?

Google glasses or something similar?

Type of

Product

# of

Data

Point

Memory Micro-

Proc-

essor

Display Cam-

era

Connect-

ivity,

Sensors

Bat-

tery

Power

Mgmt

Phones 23 15% 22% 22% 8.2% 7.9% 2.3% 3.8%

Includes

WiFi

Contribution of “Standard Components” to Costs

of Selected Electronic Products

Can any of these components be eliminated?

Are there improvements in components and/or

technological trends that can help us think

about components to eliminate?

What will be Next? (3) Can microprocessors and memory be eliminated to

create low-end phones that bypass network providers (SingTel, StarHub)

Lower cost phones

Lower cost services

Cellular processors are eliminated as WiFi becomes more available?

If WiFi is main connection and it works good enough

Can we reduce memory capacity?

Can we reduced performance of application processor?

Lower resolution cameras, displays, and other components will also reduce costs

How might open source software enable lower costs?

Recent Article in New York Times

Cellphone Start-Ups Use Wi-Fi First to Handle Calls

and Take On Rivals

Two start-ups are trying to lower cellphone costs by

relying on Wi-Fi routers, and now some of the

bigger companies are looking to follow their lead.http://nyti.ms/1AFMiFW

Great Source on WiFi Diffusion:

Number of world WiFi access points is 56 million

http://www.ipass.com/wifi-growth-map/

Can Google’s Project Ara provide Users with Better

Choices about Inexpensive Phones that have less features?

Modular phone

that enables

Users to Choose

Specific Modules

Will Apple be

Disrupted?

Apple has

highest prices

Does it Deserve

High Prices?

Will Low-End

WiFi Phones

Impact

Apple or other

phone

suppliers?

How About the Other 8 Products? How might improvements in electronic components

change the other products that were briefly mentioned?

Tablet computers, eBook Readers

Game consoles, MP3 Players

Televisions, Internet TVs, Google glasses

All of these products are being improved with new

electronic components

WiFi may eliminate cellular processors in tablet computers

and reduce memory in game consoles

What will these products become?

Will they become more important in our lives?

Or will they disappear as their functions are absorbed

by other products?

How About Still Other Products (2)

Improvements in computers: logistics, free routing if aircraft,

pre-fab housing, computer assisted-doctors, Big Data, online

universities

Improvements in ICs, MEMS, GPS, other electronic components:

Smart homes, Internet of Things, Drones, Autonomous Vehicles,

Sharing economy including shared bicycles, GPS for buses

Improvements in displays and ICs; rolled displays, wrist displays,

wearable devices

Improvements in microprocessors and power electronics: cheaper

electric vehicle chargers that enable more frequent recharging

and thus reduce the need for high energy storage batteries

Many of these products can contribute towards sustainability

These things and more are discussed in MT5009, Analyzing Hi-Tech

Opportunities (for more info, see http://www.slideshare.net/Funk98/presentations)

Comments on Learning and Invention

When standard components contribute much more to costs than do assembly operations and they experience rapid improvements

Factory floor learning and process R&D for assembly are not important sources of improvements

Product design opportunities are also limited since most of the components are standard ones

Demand for the standard components is driven by multiple products and thus the volumes for a new product have little impact on the standard components

Instead, learning revolves around the standard components and how to use them to introduce better products

Comments on Learning and Invention (2)

Contrast this with automobiles

Discussions of automobiles emphasize speed,

acceleration, fuel economy, smoothness,

quietness, and interior and exterior aesthetics,

measures

These depend more on overall product and

process design than any one component

Even speed, acceleration, and fuel economy

depend on many design factors

and not just the engine design since they involve

the aerodynamics and weight of the overall

automobile

Comments on Learning and Invention (3)

When standard components contribute much

more to costs than do assembly operations

and they experience rapid improvements

Monitoring these components is essential for

“inventing” and developing new products

What components are in the new product?

What levels of performance and cost are

needed in this product and in its components

before the product will become economically

feasible?

Thank you!