Deepdive in AIML venture landscape By Ajit Nazre Rahul Garg
1. A Deep Dive in the Venture Landscape of
Artificial Intelligence and Machine Learning
September 2015
Ajit Nazre
Rahul Garg
2. Artificial Intelligence Evolution
Artificial Intelligence is on its 3rd reincarnation since its first birth in 1950 and looks like it has
longevity this time around
1950 -1970s – Start of
AI as a concept but no
real applications
1980 -2000 – AI/ML
mainly used in Military &
Academia
2005 onwards – Large tech companies such as IBM, Microsoft, Google, and Facebook have
invested in AI/ML for commercial applications
1956
John McCarthy
organized a
conference at
Dartmouth &
named the field as
Artificial Intelligence
1950
Alan Turing
published a paper
about the possibility
of machines with
true intelligence
1995
US Department of
Defense uses
predator UAV in
Balkan War
1997
IBM’s Deep Blue
wins chess against
World Champion
Gary Kasparov
2011
Debut of Virtual
personal assistants
like Apple’s SIRI &
Microsoft’s Cortana
2011
IBM Watson
computer defeats
Jeopardy game
show champions
Jan 2014
DeepMind team
uses deep learning
algorithms to create
a program that wins
Atari games
Oct 2013
Vicarious breaks any
‘Captcha’ passing
the Turing test
Jun 2015
Facebook launches
Moments that
detects faces and
shares photos with
friends to whom
they belong
May 2015
Google self driving
cars complete 1M
miles autonomously
Jun 2015
DeepMind teaches
program how to
read
3. Artificial Intelligence / Machine Learning Classification
Artificial
Intelligence
Deduction,
Reasoning,
Problem Solving
Knowledge
Representation
Planning Perception:
Computer Vision
Machine
Learning
Robotics: Motion
and
Manipulation
Natural Language
Processing
Social
Intelligence
Supervised Learning Unsupervised
Learning
Reinforcement
Learning
Decision
Tree
Learning
Association
Rule
Learning
Neural
Networks
Inductive
Logic
Programming
Support
Vector
Machines
Bayesian
Networks
Similarity
and Metric
Learning
Clustering
Deep
Learning
Manifold
Learning
Sparse
Dictionary
Learning
Genetic
Algorithms
4. Multiple Factors contributing to Dramatic Improvement in
Prediction Accuracy
Increasing Prediction
Accuracy of
Algorithms
=
Improved Decision
Making
Increase in
affordable
Compute
Power
Big Data
Faster
networks
Cloud Infra-
structure
Sensor
Networks
Advances in
Neuroscience
“Whenever we can replace human judgment by a formula, we should at least consider it.” -
Daniel Kahneman (2002 Nobel Prize in Economics)
5. Applications of AI/ML in Daily Use Already
Info
Search engines
Sentiment analysis (or
opinion mining)
Information Retrieval
Spam filtering for email
Speech and handwriting
recognition
Spoken language
understanding
Stock analysis
Structural health
monitoring
Syntactic pattern
recognition
Topic spotting: categorize
news articles
Weather prediction
Tools
Face Detection
Finance – Derivatives
Trading
Game playing
Software Testing
Internet fraud detection
Machine translation
Medical diagnosis
Mood analysis
Brain machine interface
in prosthetics
Optical character
recognition
Recommendation
systems
Robot locomotion
Services
Advertising - Targeting
Bioinformatics
Automatic word completion
Chemical Informatics
Classifying DNA Sequences
Computer Vision – Object
Recognition
Customer Segmentation
Detecting Credit Card Fraud
6. R&D Activity in AI / ML as measured by Patents granted
Source: MarketRealist.com
IBM and Microsoft had more patents in AI / ML granted in 2014 than all others
7. M&A Activity in AI / ML is Heating up
Google
Dark Blue Labs, Vision Factory, Jetpac, Quest Visual, Titan
Aerospace, Deepmind Technologies, Boston Dynamics, Bot
& Dolly, Holomni, Redwood Robotics, Industrial perception,
Schaft, Flutter, Wavii, Behavio, DNNresearch, Viewdle,
Pittpatt, Saynow, Phonetic Arts, Metaweb
21
10
6 6
4
0
5
10
15
20
25
Google Facebook Yahoo Apple IBM
Facebook
Pebbles, Surreal Vision, Ascenta, QuickFire,
Wit.ai, Oculus VR, SportStream, Jibbigo,
Face.com, RecRec
Yahoo
Tomfoolery, Cloud Party, Skyphrase, LookFlow,
IQ Engines, Qwiki
Apple
Metaio, Cue, Novauris Technologies, Polar
Rose, Siri, FingerWorks
IBM
AlchemyAPI, Silverpop Systems, Curam
Software, Languagae Analysis Systems
No. of companies acquired in AI/ML
Source: Wikipedia
Google is most acquisitive (21 companies) in the space followed by Facebook (10 companies)
8. AI / ML Venture Funding Analysis: Methodology
• Created a database of over 2800
AI/ML companies from several sources
including Angel list, Crunch base,
Tracxn, Venture scanner, Portfolios of
major VCs, News Reports & Secondary
Research
• Weeded out public, acquired, closed
down and non existent companies
(with no website) arriving at a list of
1781 companies
• Filtered the list further to 312
companies based on funding > $100k
• In addition to the focus areas in AI /
ML areas defined on slide 3, we have
extended our analysis to areas like
augmented reality, image / facial
recognition, and drones that overlap
multiple areas.
2800
AI & ML companies from
sources: Angel List, Crunch
base, Tracxn, Venture
scanner, VC websites,
News articles & secondary
research
1781
Active companies after
weeding out public,
acquired, closed down,
non existent companies
312
Companies that
have external
funding of >$100k
9. Active AI / ML Ventures by Country in 2015
More than half of the AI/ML companies are based in the US
929
127 108
77
52 44 34 32 29 19
330
0
100
200
300
400
500
600
700
800
900
1000
USA UK India Canada Germany France Israel Russia China Netherlands Others
n= 1781
10. AI / ML Ventures based on Funding and Location
SF Bay Area has the highest density of funded startups followed by New York & London
Out of funded companies , nearly 40% are in the SF Bay Area &
70% in North America
Nearly 18% of companies are funded
68
49
26
12 12
8
0
10
20
30
40
50
60
70
80
Silicon
Valley
San
Francisco
NewYork London Seattle Berlin
312
1469
Funded
Companies
n=312
11. AI / ML Startups are still in their Infancy
62% of the startups still pre series A
193
72
26
10
5 1 5
0
50
100
150
200
250
Seed/Angel Series A Series B Series C Series D Series E Other
n=312
12. Venture Funding and Most Active VCs in AI / ML
Khosla Ventures:
Blue River, Scaled Inference, Atomwise, Lumiata,
Kaggle, Idibon, Metamind, Pymetrics, Ayasdi, Catalia
Health, Theatro, Thync
Google Ventures:
Expect Labs, Wonder Workshop, Clarifai, Orbital
Insights, Airware, Agent, AltspaceVR, Bento, Building
Robotics, Framed, Skycatch
Intel Capital:
API.ai, Cloudmade, Emotient, Expect Labs, Eyesmart,
PrecisionHawk, Hooklogic, Prelert, Reflektion, Total
Immersion, Fortscale,Whoknows, Incoming Media,
Occipital
Two Sigma Ventures:
Anki, 3D Robotics, Rethink Robotics, Jibo, Kasisto,
Dextro, Socure, Floored, Canary, Zymergen, Indico
Data,
RRE Ventures:
DigitalGenius, Yhat, Clearpath Robotics, Jibo, Palantir,
Giphy, Viglink, Airware0
50
100
150
200
250
300
350
2010 2011 2012 2013 2014
14
12 12 12
8
0
2
4
6
8
10
12
14
16
Intel Capital Khosla
Ventures
Two Sigma
Ventures
Google
Ventures
RRE
Ventures
Source: Company Websites
Source: CB Insights
No. of Investments in AI/ML companies
Venture Investments in $M in AI/ML companies
300% increase in venture funding from ‘13 to ‘14. Intel Capital, Google, Two sigma & Khosla
ventures most active in investing in AI / ML
13. Funded Ventures in various fields of AI / ML
Machine learning, Robotics, AR and Image recognition are most actively funded areas
n=31292
58
36 34 33
26
11 10 8
4
0
10
20
30
40
50
60
70
80
90
100
ML
(excluding
Deep
learning and
predictive
analytics)
Robotics &
Drones
AI
(excluding
NLP, ML,
Robotics,
Computer
Vision)
Augmented
Reality
Image &
Face
Recognition
NLP Predictive
Analytics
(clustering,
SVM,
Bayesian
networks)
Computer
Vision
Deep
Learning &
Neural
Networks
Others
14. Major Fields for AI / ML Venture Investment by Geography
North America is leading the way but Europe has significant presence in AI & AR areas
n=312
68
44
22
20 20 19
11
6 6
2
17
10 9
14
6
13
0
2 2 2
7
4
2 2 2
0
2
0
0
10
20
30
40
50
60
70
80
ML
(excluding
Deep
learning
and
predictive
analytics)
Robotics &
Drones
AI
(excluding
NLP, ML,
Robotics,
Computer
Vision)
Augmented
Reality
Image &
Face
Recognition
NLP Predictive
Analytics
(clustering,
SVM,
Bayesian
networks)
Computer
Vision
Deep
Learning &
Neural
Networks
Others
North America
Europe
Asia/Other
15. Areas of Application of Venture funded companies in AI/ML
72% of the funded startups targeting enterprise applications
n=312
(consumer apps, consumer electronics,
educational applications for consumers
& toys)
224
88
0
50
100
150
200
250
Enterprise Consumer
16. AI / ML Ventures with Applications across Industries
Impact across a wide range of industries
50
38
30
23 23 23
16
14 14
9 8 7 6 6 5 5
35
0
10
20
30
40
50
60
ConsumerApps
IndustrialAutomation
AI/MLasaService
Healthcare
Media
Marketing
Advertising
Security
FinancialServices
Ecommerce
HR
Agriculture
Education
Enterpriseproductivity
ITServices
CustomerService
Others
n=312
17. Aerospace
Cumbersome & manual process
of maintenance and repair using
field manuals that have to be
frequently updated
Field service agents can call or chat with
support from his/her augmented reality view
Today Future
Companies making it happen: Atheer Labs, Augmate, Infinity AR, Total Immersion,
Vuzix
Pic Courtesy: Atheer Labs
18. Agriculture
Farming decisions based on tradition and
intuition
Machine learning algorithms using sensor data
and aerial imaging help farmers make
intelligent data based decisions increasing
yield
Today Future
Companies making it happen: Blue River, Farmlogs, Greensight, Mavrx, Pulsepod,
Terravion
Pic Courtesy: Farmlogs
19. Automotive
Cars driven by humans, prone to errors
(nearly 1.3M people die in road accidents
every year)
Driverless cars – leading to comfortable
experience & less human fatalities
Today Future
Companies making it happen: BMW, Daimler Benz, Google
Pic Courtesy: Google
20. Background Checks
Checking of background information,
documents and identity manually taking 1-2
weeks
Automated background checking process
through APIs, real time ID authentication using
image recognition. Nearly 60% reduction in
costs
Today Future
Companies making it happen: Onfido
Pic Courtesy: Onfido
21. Communications
People can talk to one another only in a
common language that they speak
People will be able to talk to one another
without knowing the other person’s language
– Nothing lost in translation
Today Future
Companies making it happen: Lexifone
Pic Courtesy: Clker.com
Hi, How are
you
I’m fine thank
you
Hi, How are
you
मैं ठीक हूं
धन्यवाद
22. Customer Care
Time consuming process of
Authentication
Biometric Voice recognition based instant
authentication
Today Future
Companies making it happen: Agnitio, Nuance, Verbio, Voicebase, Voiceitt, VoiceVault
Sir, Please help me with
the authentication & tell
me your DOB, Address,
Mobile No.
My Password is XXXXX
23. Customer Care
Cumbersome & time wasting
Voice Menu
Virtual assistant that can converse like humans
& assist without Menus
Today Future
Companies making it happen: DigitalGenius, Expect Labs, Nuance
Press 1 for Credit
Cards, Press 2 for
Debit Cards, Press 3
for Loans, Press 4 for
Netbanking……
24. Healthcare
Doctor reads static patient charts
before or after visiting a patient
Doctor can read all relevant reports
dynamically using gestures during her patient
visit
Today Future
Companies making it happen: Atheer Labs, Augmedix, Infinity AR
25. Media
Watching ads on TV is avoided as they are
not actionable and mostly irrelevant
Automatic content recognition will allow
advertisers/ programmers to inject
interactivity & context relevant content
Today Future
Companies making it happen: Cognitive Networks, Datascription, Dextro, Enswers, Magic
Pony Technology, Persado, Videntifier Technologies
Wonder if
that Jersey
came in my
size
Just ordered
the Jersey
directly
Pic Courtesy: Forbes
26. Navigation
Driver needs to divert his gaze or
has to give rigid voice instructions
On windshield navigation with heads up
display that recognizes the surrounding and
gives instructions
Today Future
Companies making it happen: Nuviz, Wayray
Pic Courtesy: Wayray
27. Office Productivity
Mundane tasks in office leading to decreased
productivity
Automatic meeting schedulers, note takers,
speech to text transcription & virtual personal
assistants will improve productivity
Today Future
Companies making it happen: Assistant.to, Gridspace, Idibon, Robin Labs, Thoughtly,
X.ai, Zahdoo
Pic Courtesy: Typepad & Coolcontourproducts.com
28. Oil and Gas
Crack Detection: It takes human 5
hours per inspection with 92%
accuracy (Avg cost ~ $7.5)
20 seconds with 94% accuracy through image
recognition (Average Cost ~ $2)
Today Future
Companies making it happen: Tractable
Pic Courtesy: Wikipedia
29. Recruiting
Impossible to manage candidates with
required behavioral skills
Neuroscience based games will match
candidates with required skills
Today Future
Companies making it happen: Connectifier, Pymetrics, Talentoday
Pic Courtesy: Pymetrics
30. Wealth Management
Stock trading & investing happens mainly on
personal skills and divine luck
Sentiment analysis, crowd sourced research &
algorithms will make investing more
transparent and better informed
Today Future
Companies making it happen: Betterment, Estimize, Narrative Science, Personal Capital,
Sigfig, Wealthfront