These slides show how the most successful startups of today (Unicorns) are not doing as well as the most successful of 20 to 50 years ago. Today's startups are doing worse in terms of time to profitability and time to top 100 market capitalization status. Only one Unicorn founded since 2000 has achieved top 100 market capitalization status while six, nine, and eight from the 70s, 80s, and 90s did so. It is also unlikely that few or any of today's Unicorns will achieve this status because their market capitalizations are too low, share prices increases since IPO are too small, and profits remain elusive. Only 14 of 45 had share price increases greater than the Nasdaq and only 6 of 45 had profits in 2019. The reasons for the worse performance of today's Unicorns than those of 20 to 50 years ago include no breakthrough technologies, hyper-growth strategies, and the targeting of regulated industries. The slides conclude with speculations on why few breakthrough technologies, including science-based technologies from universities are emerging. We need to think back to the division of labor that existed a half a century ago.
The Troubled Future of Startups and Innovation: Webinar for London Futurists
1. The Troubled Future of
Startups and Innovation
Jeffrey Funk
Retired Associate Professor
Webinar, London Futurist, July 18, 2020
“the 2010s were the worst decade for
productivity growth since the early 19th century”
Quote from an April 2020 Financial Times article
2. Startup Foun
ded
Year for
Profits
Years to Top
100 Mkt Cap
Microsoft 1975 1 12
Apple 1976 4 28
Genentech 1976 8 27
Oracle 1977 3 19
Home Dep 1978 3 17
EMC 1979 6 17
Amgen 1980 9 19
Adobe 1982 1 35
Sun 1982 6 15
Cisco 1984 5 11
Dell 1984 6 13
Compaq 1984 4 13
Startup Foun
ded
Year
Profits
Top
100
Qualcomm 1985 10 14
Celgene 1986 17 28
Gilead Sci 1987 15 21
Nvidia 1993 6 24
Amazon 1994 10 16
Yahoo! 1994 4 5
Ebay 1995 4 10
Netflix 1997 5 21
Google 1998 5 8
PayPal 1998 4 21
Salesforce 1999 4 19
Facebook 2004 6 10
Years to Profits, Top 100 Market Cap for Valuable Startups of Last 50 Years
Only 1
founded
since
2000
versus
6 in 70s
9 in 80s
8 in 90s
3. Lack of Venture Capital Funding Isn’t Problem
VC funding
recovered a few
years after dotcom
bubble burst
Began to grow in
2010 reaching
record 5-year high
(2015 – 2019)
Many new Googles
and Amazons should
have already
succeeded
4. Ex-Unicorn
(14 of 45)
Year
Founded
Market Capitalization ($B) Share Price
Change
Nasdaq
Change2019 March 9, 2020
Uber 2010 60 38.9 -46% - 9%
Square 2009 24 23.3 +316% +41%
Zoom 2011 20 30.2 +77% - 10%
Twilio 2008 17 10.9 +198% +46%
Lyft 2012 17 7.3 -35% - 8%
Snapchat 2011 17 14 -61% +23%
Crowdstrike 2011 15 8.0 -35% - 7%
Slack 2009 14 11.8 -42% - 10%
Pinterest 2009 14 7.6 -45% - 10%
Roku 2002 12 9.5 +197% + 4%
Wayfair 2002 12 3.1 +1.5% +48%
Okta 2009 11 13 +351% +24%
DocuSign 2003 10 12.3 +73% +19%
Dropbox 2007 9 6.6 -45% + 3%
Only 14
of 45 ex-
Unicorns
had
share
price
changes
greater
than
Nasdaq
$98B
Needed
to be
in top
100
in 2019
6. Among all startups
at IPO time
Percent profitable fell
from 80% in early 1980s
to 20% in late 2010s
Despite median age
(founding to IPO)
almost doubling
https://www.businessinsider.com/uber-lyft-ipo-
trends-money-losing-unicorns-could-cause-stock-
market-issues-2019-5?IR=T
Median Age
% Profitability
% Profitability
7. Amazon had profits by Year
10, neither Uber nor Tesla
did. Amazon’s cumulative
losses didn’t reach $3B while
Uber’s exceeded $20B and
Tesla’s $6B. Latter two losses
still growing
Tesla’s Losses
(Year 11 to 17)
Amazon’s Net Profits
https://qz.com/1196256/it-took-amazon-amzn-14-years-to-make-as-
much-net-profit-as-it-did-in-the-fourth-quarter-of-2017/
https://promarket.org/the-uber-bubble-why-is-a-
company-that-lost-20-billion-claimed-to-be-successful/
https://www.statista.com/
statistics/272130/net-loss-
of-tesla/
Tesla and Uber have Lost Much
More Money than Amazon
8. Will Ex-Unicorns Reach Top 100 Market Cap Status?
Two are 1/5 of the way to $98 market cap with >$20B
Both have share price increases greater than Nasdaq increases and
they had profits in 2019 (Zoom and Square)
Ten are 1/10 of the way, with >$10B market cap
But only 3 had share price increases > Nasdaq increases
And none had profits in 2019
Will Zoom make it to top 100 market cap, or Tesla or Uber?
By the way, only fintech is profitable, and what will happen to
Unicorns that have yet to do IPOs (479, $1.4 trillion valuation)
9. Why Are Unicorns Doing Worse than past ones?
One hypothesis: new startups acquired by large incumbents
before achieving top 100 market cap status
All founded since 2000: Youtube, Instagram, GitHub, Linkedin and
WhatsApp.
But all successful startups made acquisitions. Microsoft obtained Power
Point, through acquisition
A bigger problem is acquisition argument assumes new startups
must challenge strong incumbents
Successful startups avoided strong incumbents by commercializing new
technologies not within interests of strong incumbents.
Silicon Valley evolved from semiconductor companies to disk drives,
networking equipment, PCs, workstations, software products and then
Internet in 1990s
10. Problem is No Breakthrough Technologies
Ride sharing and food delivery use same vehicles, drivers, and roads
as did previous taxi services
Online sales of juicers, mattresses, and exercise bikes are sold in same
way Amazon currently sells almost everything
New business software enables more cloud-based work, not a huge
advantage during normal times
Fintech startups use algorithms to find low-risk borrowers or
insurance subscribers, but advantages are still small
Online education may deliver content differently, but it is the same
content
In all these cases, the technology is not revolutionary.
11. Regulated Industries and Hyper-Growth Strategy
Harder to succeed in regulated Industries
Taxi services regulated because of congestion, which plagues ride
sharing and challenges scooters and bicycle rentals
Fintech challenging traditional banking companies
Education startups fighting highly regulated industry and huge clash
between public and private schools
Hyper-Growth Strategy prevents experimentation
Startups have subsidized users in effort to grow, thus bypassing
experimentation
Ride sharing, food delivery, fintech, e-commerce startups copy leaders
Unicorns can’t survive without subsidies
12. 0.00%
2.00%
4.00%
6.00%
8.00%
10.00%
12.00%
2002 2006 2010 2014 2018
Declining VC Investments in
Science-Based Industries
Semiconductors
Communication
Equipment
Medical
Instruments
Where are
fusion, super-
conductors,
nanotechnology
(graphene,
CNTs), bio-
electronics,
quantum
computing?
Money isn’t
issue.
Government
R&D funding
been high for
decades
13. Why So Few Science-Based Technologies?
Change in Division of Labor
1940s – 1960s: AT&T, IBM, Motorola, GE, RCA,
DuPont, Monsanto did basic research
Today: universities train PhDs, write papers, obtain
funding, but little work with companies
Hyper-Specialization at Universities
Exponential growth in journals, papers, and citations
to papers
Growing emphasis on science in engineering research
>144 Nature journals
Today’s top university scientists are drowning in
academic papers, journals, patents, and admin work
15. Selected Publications
What Drives Exponential Improvements? California Management Review 55(3): 134-152, Spring
2013
Rapid Improvements with No Commercial Production: How do the improvements occur? Research
Policy 44(3): 777-788, 2015 (second author is Chris Magee)
Assessing Public Forecasts to Encourage Accountability: The Case of MIT's Technology Review,
PLOS ONE, August 2017.
What Does Innovation Today Tell Us About the US Economy Tomorrow? Above all, that the nation
needs to get a lot better at linking scientific advance to economically and socially valuable
technologies. Issues in Science and Technology December 2017
Technology Change, Economic Feasibility and Creative Destruction: The Case of New Electronic
Products and Services, Industrial and Corporate Change 27(1): Pages 65–82, February 2018
Beyond Patents: Scholars of innovation use patenting as an indicator of both innovativeness and the
value of science. It might be neither, Issues in Science and Technology Summer 2018.
What’s Behind Technological Hype? Start-up losses are mounting, and innovation is slowing. We
need less hype and more level-headed economic analysis, Issues in Science and Technology Fall,
2019.
AI and Economic Productivity: Expect Evolution, Not Revolution. IEEE Spectrum, March 2020
Three Part Series on Startups, Mind Matters, May/June 2020. Where are all the profitable startups?
Why do Today’s Startups Disappoint Investors? Why are there no new Googles and Amazons?
The Increasing Limitations of Academic Experts: Narrower Specializations and Less Practicality
Even as Problems Become More Complex, Working Paper
16. Falling Research Productivity
Drugs
Number of drugs per billion dollars of R&D dropped about 80 times in last
50 years
Number of researchers per commercialized drug rose by almost five times in
last 50 years
Number of researchers required to maintain the same rate of increase
in crop yields rose 6 to 24 times (corn, soybeans, cotton, wheat)
between 1970 and 2010
R&D needed to sustain Moore’s Law has risen in recent decades
Number of drugs per $billion from Nature article by Scanlan et al, 2012
Other data on drugs, and crops and Moore’s Law from Are Ideas Getting Harder to Find
17. Falling Research Productivity - Continued
R&D productivity has fallen across a wide variety of industries
Revenue growth per research dollar has fallen by about 65% over the last
30 years (Anne Marie Knott)
Importance of Nobel Prize winning research in physics has
declined over last century
Few Nobel Prizes have been awarded for research done since 1990 not
only in physics, but also for chemistry and medicine (Atlantic article)
22. Moore’s Law is
slowing and
evidence
of other
technologies
experiencing
rapid rates of
improvement
Is difficult to find
I covered these
issues in my course
at NUS from 2009
to 2016
23.
24. Moore’s Law enabled these product by reducing their costs and
improving their performance
With Moore’s Law slowing, new types of electronic products (VR, AR,
robots, commercial drones, blockchain, AI) will take much longer to
emerge and diffuse
25.
26. Improvements in Other Technologies in Table
No more improvements in cost and performance?
Microprocessors, memory chips, camera chips
Superconductors, DNA sequencers (nothing since 2015)
Improvements but little impact?
Magnetic storage, Organic transistors
Soon to be slowing?
WiFi, cellular speeds and cost; liquid crystal displays
Batteries? As car batteries catch up with laptop batteries?
Continued improvements in cost and performance?
OLED displays
Silicon, organic, perovskite, quantum dot solar cells
LAN, Internet speeds
27. Mag lev to hyperloop
Micro-finance to fintech
Stem cells to gene editing
Telematics to IoT
Ride sharing to MaaS
Forgotten about solar water heaters, fusion,
cellulosic ethanol, strategic defense initiative
Hype about new technologies:
Proponents Replace Old Ones with New Ones
Even Though Old Ones Provide Lessons
28. Why I am Pessimistic about AI
Growth much slower than forecasts
$15 trillion in economic gains expected by 2030 but only $10
billion in 2018, $15b in 2019, and $23B (est) in 2020
Growth still stuck in news, advertising, and e-commerce
Few startups offer products and services that directly
impact on productivity (IEEE Spectrum)
Solow’s Paradox and small impact of bar codes in retail
(reduced grocery costs by 1.3%)
Little success in driverless vehicles or manufacturing
29. Why I am Pessimistic about AI - continued
Limitations of Big Data revealed in 2016 book, Weapons
of Math Destruction by Cathy O’Neil
Limitations of AI revealed in
AI Delusion by Gary Smith (2018)
Rebooting AI by Gary Marcus (2019)
Computational power used to achieve higher accuracies has been
doubled every 3.4 months
300,000-times increase in capacity after 2012
Head of Facebook AI (Jerome Presenti) says this is
unsustainable. "If you look at top experiments, each year cost is
going up 10-fold. An experiment might be in seven figures, but
it’s not going to go to nine or 10 figures, it’s not possible,
nobody can afford that."
32. Lot of Misleading Hype
Misleading hype in health care: failure of Watson
Misleading hype in energy:
DeepMind did not reduce energy usage at a Google data centers nor for
UK economy; Economist claims “some insiders say such boasts are
overblown,”.
Nest did not reduce energy usage in homes, nor did general subsidies for
smart meters do so
And these propagators of hype are big money losers
DeepMind’s 2018 losses reached $572 million in 2018, up from $154
million in 2016 and $341 million in 2017, on revenues of $124 million.
Nest lost $621 million on revenues of $726 million in 2017.
33. Lots of Misleading Hype - continued
Stanford University’s Artificial Index 2019 Annual Report is filled with hype; no market
data or examples of successful products and services
Presents 300,000 times increase in computational power used in training exercises as
good sign, but industry people say otherwise
Head of Facebook AI says this is unsustainable. "each year the cost is going up 10-fold. Right
now, an experiment might be in seven figures, but it’s not going to go to nine or 10 figures, it’s
not possible, nobody can afford that."
Report fails to address impact of increase in computational capacity on improvements in
accuracy or reductions in time and cost of training exercises, such as in image
recognition.
How much are these trends a result of better machine learning algorithms or more parallel
processing with bigger computers? If it is latter, limits will likely cause a slowdown in image
recognition improvements
34. MIT Technology Review’s Predictions: Many Sound More
Like Scientific Disciplines Than Products and Services
2005
Airborne Networks
Quantum Wires
Silicon Photonics
Metabolomics
Magnetic-
Resonance Force
Microscopy
Universal Memory
Bacterial Factories
Enviromatics
Cell-Phone Viruses
Biomechatronics
2004
Universal
Translation
Synthetic Biology
Nanowires
T-Rays
Distributed
Storage
RNAi Interference
Power Grid Control
Microfluidic
Optical Fibers
Bayesian Machine
Learning*
Personal Genomics
2003
Wireless Sensor
Networks
Injectable Tissue
Engineering
Nano Solar Cells
Mechatronics
Grid computing
Molecular imaging
Nanoprint
lithography
Software
assurance
Glycomics
Quantum
cryptography
2001
Brain-Machine
Interface:
Flexible Transistors
Data Mining
Digital Rights
Management
Biometrics
Natural Language
Processing
Microphotonics
Untangling Code
Robot Design
MicrofluidicsOrange: <$100 Million sales
Blue: too broad and vague to gather data
Green: Over $10 Billion sales; Black: >$100M but <$10B *machine learning also in 2013 predictions
35. Scientific American’s 40 Predictions (2015-2018)
Vague
Next Generation Batteries and Robotics, IoT Goes Nano, Sustainable
Design of Communities, Sense and Avoid, Affordable Catalysts
What is the specific technology?
Old
Fuel Cells, additive manufacturing, distributed manufacturing,
catalysts for vehicles
How are these technologies new?
Not a Technology
AI Ecosystem, Sustainable Design of Communities, Sense and Avoid
Drones
Similar or Recycled Ideas
Dimensional Materials (nanotech?), AI and Deep Learning (5 Times),
Many genetic technologies (7 Times), Quantum Computers (2 Times)
36. 2015 2016
Fuel-cell vehicles OLD
Next-generation robotics VAGUE
Recyclable thermoset plastics
Precise genetic-engineering techniques
Additive manufacturing OLD
Emergent artificial intelligence VAGUE
Distributed manufacturing OLD
“Sense and avoid” drones VAGUE
Neuromorphic technologies
Digital genome
Autonomous Vehicles
The Internet of Things Goes Nano VAGUE
Next-Generation Batteries VAGUE
Open AI Ecosystem TECHNOLOGY?
Optogenetics for Therapeutic
Neuroscience
Organs-on-Chips
Perovskite Solar Cells
Systems Metabolic Engineering
Blockchain
Dimensional Materials NANOTECH
RECYLCED
Scientific American’s PredictionsSimilar
similar
similar
37. 2017 2018
Blood Tests for Scalpel-Free Biopsies
Draw Drinking Water from Dry Air
Deep-Learning Networks
Artificial Leaf Turns Carbon Dioxide Into
Liquid Fuel
Human Cell Atlas
Precision Farming Increases Crop Yields
Affordable Catalysts for Vehicles VAGUE
Genomic Vaccines
Sustainable Design of Communities VAGUE
Quantum Computing
Augmented Reality
Advanced Diagnostics for Personalized
Medicine
AI for Molecular Design
AI That Can Argue and Instruct
Implantable Drug-Making Cells
Lab-Grown Meat
Electroceuticals
Gene Drive
Plasmonic Materials
Algorithms for Quantum Computers
Scientific American’s Predictions
Similar
More Genetic
Engineering
Similar
38. Number of
PhDs
% with
PhD
% with PhD
or MS
% with PhD,
MS, or MD
% of Total
PhDs
Biotech 791 35% 41% 53% 32%
Education & Research (mostly biotech) 346 33% 40% 47% 14%
Medical Instruments 159 13% 24% 32% 6.4%
Sub-total, life science sector 1296 28% 36% 46% 52%
General Instruments 104 24% 38% 40% 4.2%
Semiconductors 158 18% 41% 41% 6.4%
Electronic Equipment 79 15% 31% 31% 3.5%
Communications Equip 86 11% 32% 33% 3.2%
Sub-total, electronics Sector 427 16% 36% 37% 17%
Computer Programming 51 8.9% 22% 22% 2.1%
Computers 50 8.4% 29% 20% 2.0%
Computer Systems 34 7.8% 20% 21% 1.4%
Software 136 6.3% 20% 20% 5.5%
Telephone & Telegraph 27 5.2% 15% 15% 1.1%
Sub-total, Internet Infrastructure 298 8% 22% 22% 12%
Computer Services 29 5.1% 16% 19% 1.2%
Information Retrieval 22 4.7% 4.7% 13% 0.9%
Retail & Wholesale Trade 16 4.5% 12% 12% 0.6%
Finance, Broadcasting, Transport, Securities, Insurance,
Real Estate.
8 2.6% 11% 12% 0.3%
Business and Other Services 26 4.0% 12% 12% 1.0%
Advertising, Employment, Leasing 7 2.9% 9.4% 9.4% 0.3%
Sub-Total, Internet Content, Services, and Commerce 108 4.2% 13% 14% 4.3%
Number and Percentage of Advanced Degrees by Industry and Sector