00:10so yeah I'm very happy to be to be here
00:13I'm gonna be talking for about 30
00:14minutes and then open it up to
00:16after that the talk is to half's the
00:19first half is really about what is this
00:21big data thing to educate you guys so
00:23you are aware what big data means and
00:25then the second half is the startup
00:28story of cloud era how cloud that I came
00:29to be and some lessons that I learned as
00:32as I was founding this company so with
00:36that said briefly about myself I got my
00:39my bachelor's and master's from Cary
00:42University in Egypt and then I came here
00:44to Stanford to get my PhD and my goal
00:48was to get my PhD from Stanford and then
00:50go back to Egypt and each I really liked
00:53to teach and my dad is a professor in
00:55can university of economics he also
00:57really likes to teach so since I was
00:59very young he told me you gonna grow up
01:01and you're gonna be a professor like me
01:02and that was my goal to get the PhD and
01:04go back to Egypt but then Stanford
01:06corrupted me Stanford as you know is a
01:10very intrapreneurial school yeah you
01:11have speakers from the industry coming
01:13in to talk all the time you have lots of
01:16classes like this one on
01:18entrepreneurship and and so on so I
01:19became very curious about this
01:21enterpreneurship thing very quickly and
01:23that led to me actually taking a leave
01:26of absence out of my ph.d program in
01:281999 and I started my first company
01:31which was called viva smart and we were
01:34acquired by Yahoo a year later and we
01:36were very lucky to be acquired by Yahoo
01:39just before the guillotine fell down
01:42which is the explosion of the what was
01:45internet bubble that was happening in
01:46the late 90s and early 2000s we had just
01:50as a point of reference we had about 20
01:52other startups doing the same exact
01:54thing like us and they all failed they
01:56all shut down during this bubble
01:58explosion was a very bad exposure
02:00actually so we were kind of very
02:02fortunate unlucky that that happened to
02:03us and then my dad kept nagging me and
02:07saying telling me hey I'm er you have to
02:09finish your PhD and I told him that I'm
02:11richer than you would ever be I was
02:12already sold my company to Yahoo made
02:14some very good money from Yahoo and he
02:16insisted no you have to finish the PhD
02:17so I told Yahoo I want to go back and
02:20finish the PhD and they said okay we'll
02:22let you go back and finish it but you
02:23have to keep working for us it was hard
02:26to find the professor at Stanford
02:27willing to take me as part-time most
02:30to be full-time because they're gonna be
02:31working on many other things beside your
02:33PhD but luckily a professor Mendham
02:35Lowe's Mendel rosenbloom who's as you
02:38know one of the founders of VMware and
02:39one of the best professors and computer
02:42science department that biased obviously
02:44he was my advisor he took me in and I
02:46finished my PhD with him and around
02:49middle of 2008 and then I left Yahoo at
02:53that time and started cloud era which I
02:55will tell you about in the next couple
02:57of slides but before I do that I wanted
03:01to briefly tell you the story of
03:02artificial intelligence and machine
03:03learning which is one of the key things
03:06that we are helping to enable as a
03:07company our product is around and about
03:10that and I'll explain what it is in a
03:12second but these three guys obviously
03:14they're very famous and lots of the
03:16credit for this machine learning and AI
03:18falls back to them and entering 1940s so
03:22that's 65 years ago he came up with the
03:24Turing test which many of you would know
03:26as one of the very first tests to check
03:30whether we have built something as smart
03:32as a human and the idea there was to
03:33have two rooms one room has a computer
03:36running AI the other room has a human
03:38and he cannot see which room has which
03:40and he's talking with them and he cannot
03:42tell and that's kind of when we achieved
03:44intelligence John McCarthy and Marvin
03:48Minsky and they kind of coined the term
03:50artificial intelligence and Joe McCarthy
03:53he came up with the Lisp programming
03:55language one of the very first symbolic
03:57languages for artificial intelligence
03:59and Marvin Minsky Marvin Minsky actually
04:01came up in again again we're talking
04:05about 50 in the 50s so that's 60 years
04:09ago came up with the first neural
04:10networks that actually work and you hear
04:13all this hype about deep learning and
04:14neural networks that was already
04:15demonstrated many many years ago but in
04:19the 80s a the AI winter hit we had this
04:24AI winter were all the funding from the
04:26u.s. from Japan all of the research
04:27stopped in AI it just completely stopped
04:30and it kind of died AI research died and
04:34then if you roll forward to today we
04:38have AI around us coming back everywhere
04:40we have machine learning algorithms
04:41coming around us everywhere we
04:43the assistants like Siri and Google and
04:45Alexa that we can converse with we have
04:47real-time translation and Skype and
04:49Google that can translate better than we
04:52can as humans we have facial detection
04:54in Facebook that again can do facial
04:56detection quicker and faster than we can
04:58and then we have these deep learning new
05:02network algorithms like alphago that can
05:05not only beat the best go players in the
05:07world but come up with new techniques
05:09that the go players are learning from
05:11the machine how to play the game in a
05:13better way so the question that I'm
05:15gonna ask you and I'm hoping a couple of
05:16you here can give some answers and I
05:19have three answers to that question is
05:20why why is machine learning coming back
05:23why is AI becoming more feasible and
05:26doable and happening today so just raise
05:29your hand and say something all the way
05:30in the back yes that's one of the three
05:33reasons is that computational power
05:35needed to do this at scale isn't today
05:37becoming more economical than ever
05:39before back then 30 years ago 30 years
05:41ago only organizations like the army or
05:44space organizations that had budgets to
05:46buy super super computers could actually
05:48do this now anybody can do this what
05:50else that that is the key reason the key
05:56reason is today we are able to get the
05:59data that is needed to make this happen
06:02the third reason is harder to guess but
06:05if anybody wants to venture a guess of
06:07the third reason okay we'll get to it
06:10then we'll get to it so the first reason
06:12is absolutely what you said is today we
06:15are more sophisticated than ever in
06:18terms of our ability to collect data not
06:21just from the machines and and the stuff
06:24happening in the online world but from
06:26the offline world as well because of
06:28sensor networks mobile devices
06:30satellites cameras we are able to
06:32collect data about everything and I'll
06:34give you many a couple of examples later
06:36on from our customers actually of how
06:38they're using this now one of the
06:40examples I really like to use is the ATM
06:43machine example when you went to take
06:46money out of an ATM machine 20 years ago
06:49you would put in your ATM card you put
06:52in the pin number and the only
06:53information that was captured was person
06:56he took out Y dollars from account
06:58number Z that was it we call a
07:00structured data today when you go to an
07:03ATM machine to take out money from the
07:04ATM machine there is a camera in the ATM
07:06machine taking a picture of your face
07:08the touchscreen on the ATM machine has
07:11sensors that measure how you move your
07:12hand and how you move your hand is a
07:14biometric signature that can be used to
07:15identify you the cell phone if it has
07:18the app for the bank then we know we
07:21were standing right there in front of
07:22the ATM machine so in addition to this
07:25very small piece of information person X
07:27time T Y dollars account see we have
07:29this mountain of information and that's
07:30what we call big data that allows us to
07:33truly tell who is standing right there
07:36in front of that ATM machine and whether
07:37this is a fraud transaction or an actual
07:39transaction so now today we joke and we
07:41say the only people able to steal money
07:44from your ATM account are Mission
07:48Impossible people because they need to
07:49copy your face and copy your phone and
07:50all of these kind of things though I was
07:52in a trip in Malaysia about two months
07:56ago and I was talking to the Malaysians
07:58about how ATM machines now become much
08:01harder to penetrate because of all of
08:03this big data and AI facial detection
08:05and so on and they said we have a much
08:07bigger problem you're hoping you can
08:09help us with and I told them what's your
08:11problem and they said our thieves they
08:13come in and they come use a bulldozer
08:14and they just take the whole ATM machine
08:16somewhere else and open it up and take
08:18the money from inside it and of course I
08:20told them that's not a big data problem
08:21that's a much harder fault that we can
08:22deal with but the number one reason why
08:25this is happening today is we have the
08:27data today we did not have the data it
08:29was very hard to get the data in the
08:30past today we can get the data much more
08:33efficiently you can see some statistics
08:34here like 90% of the data that we have
08:37in the world today was generated within
08:40the last two years 90% and that's
08:43because within the last two years that's
08:45when our sophistication with sensor
08:47networks with mobile devices just
08:49exploded in terms Batman and Batman
08:51and we're only leveraging a half a
08:53percent of the data effectively today
08:55and the growth trend is really expected
08:57to continue in the next four years going
09:00up from 4.4 to 44 zettabytes
09:03so the number one reason number two it
09:06reason which you guys also identified
09:09the abundance of scalable computation is
09:11today we have the storage the memory
09:14subsystems the CPU and GPU subsystems
09:17the scalable cloud architectures that
09:19allow us and allow any company to do
09:22this at scale and then the number of
09:24three reason is open source allowed us
09:28to democratize the advanced algorithm
09:32algorithms needed to do this at scale so
09:35this started with Google back in 2004
09:39Google published these two papers called
09:42MapReduce and the Google file system
09:44which were the foundation of how to
09:47start to do scalable computation using
09:49commodity hardware and Yahoo to that
09:53credit and to Google's credit as well
09:54but yeah how to years later they took
09:56the concepts from these papers and they
09:58implemented them into open source which
10:01gave us this give us the birth of Hadoop
10:03and this Big Data Platform in 2006 we
10:07only had these two things HDFS and
10:09MapReduce but they were open source they
10:11were being developed in the open source
10:12community with collaboration from many
10:14many companies including ours and then
10:16if you look what happened over the next
10:17ten years is this massive innovation
10:20people building other systems on top of
10:22these systems borrowing ideas from them
10:24leveraging them in new ways and that
10:27created this very very rich rich
10:28ecosystem of software and tools and
10:32programs that allow any company today to
10:35go and build these super advanced
10:37machine learning and AI algorithms so
10:39really the answers the question I post
10:41you at the beginning is this is
10:43happening today because of the
10:45combination of three factors the first
10:47one is the data is available the second
10:49one is computation is available and then
10:51the third one is the algorithms are more
10:53available than ever before so what do we
10:56do at Cloudera cloudera we take all of
10:57this open source goodness that's out
10:59there and we package it in a very easy
11:01to use fast secure platform that our
11:05customers can deploy within their
11:06organizations to solve these problems
11:08using machine learning and AI let's see
11:11what we do at the company you can think
11:12about us as very similar to Red Hat for
11:14Linux so Red Hat they take Linux and
11:17they make it more usable for the
11:18enterprise we take all of this big
11:22open-source projects that you can see on
11:24the far right there and we put them
11:26together into a form factor a platform
11:28that is easier to use by organizations
11:30some of our customers uses on-premise
11:32meaning inside of their own data centers
11:34and some of our customers use in the
11:36clouds on top of Amazon or Google or
11:39otherwise and the key thing about this
11:41platform is the ability to work with any
11:43type of data whether that be structured
11:45data there's an X at time T checkout Y
11:48dollars from account Z or
11:50semi-structured or unstructured data
11:52like images social media etc and at any
11:55volume and then do any type of analysis
11:58one of the most common ways to analyze
12:00data is a language called SQL sequel
12:02sequel is a very powerful language but
12:04there is so many problems that we have
12:06today that cannot be solved by sequel
12:08that need more than sequel and that's
12:10what's special about our platform
12:12compared to standard database systems as
12:14our platform can work with any type of
12:16data structured or unstructured at scale
12:19and then can allow you to tackle that
12:21data extract value from that data using
12:23sequel or going beyond sequel if you
12:25need to go beyond sequel so we have
12:26another number of other frameworks that
12:28cannot analyze data in your ways so now
12:32I want to give you some examples to make
12:33it more concrete what exactly do we do
12:35and the first example is engine from the
12:39insurance industry and before I talk
12:42about that I want to talk about a hack a
12:44trick that we many of us came up in
12:47different industries to deal with the
12:50fact that we could not scale we could
12:51not scale our processing to millions of
12:53users or millions of customers so we
12:55came up with the trick that's called
12:57segmentation so what is segmentation
13:00segmentation says instead of going and
13:02trying to target or model the behavior
13:04of all of my 10 million customers I will
13:07instead build a thousand segments
13:08married with children married without
13:10children this age group that age group
13:12this income level that income level
13:14living here living there and then I'm
13:16trying to map fold all of these 10
13:18million users into one of these segments
13:20and then I do my targeting or my fraud
13:23detection or my offers or my advertising
13:25based on these segments and the problem
13:28with this approach is that the good news
13:30about this approach is it worked and it
13:32helped us do this for many many years
13:35the problem with it is at least to false
13:37positives it will end up many many times
13:39making wrong conclusions about specific
13:42people by lumping them in segments that
13:44they yes have the same characteristics
13:46as a segment but they don't behave
13:47anywhere like that segment so for
13:50example in the internet age many 10
13:53years ago 20 years ago before we did
13:54proper segmentation and we use these
13:57techniques of lumping everybody together
13:59whenever I would go to browse the
14:01Internet so I'm a male in the age range
14:03of 40 to 50 years all of the ads I would
14:07see would be golf ads and with all due
14:10respect if you love golf golf is not my
14:11sport I don't watch it I don't play it
14:13in fact I almost hate golf
14:15it's not my thing yet I would get golf
14:17as all the time because I was folded
14:20into that segment on the other hand my
14:23son who is a teenager and loves video
14:26games continuously gets video games ads
14:28when I also love video games I play
14:30video games all the time and I have to
14:31go and ask him what it's ad that you
14:33have seen here or Lily's ad you have
14:34seen there because his age group was
14:37being folded into the right segment he
14:38was just lucky that that was happening
14:39for him at the time so same thing in the
14:43insurance industry in my house I am the
14:49lowest monthly premium for car insurance
14:51I play the lowest insurance per month
14:53because I am a male in the age range
14:564050 my son is the highest insurance
14:59premium in our house
15:00because he is a teenager male which is
15:03much worse than teenage females I guess
15:05in terms of my daughter actually also
15:08teenagers she has a lower premium than
15:09he does and that is again incorrect and
15:14a function of the segmentation because
15:16if you look at my son's driving behavior
15:18the actual behavior he's actually a much
15:20safer driver than me
15:21he's always careful very paranoid never
15:23speeds up stops at the stoplights all
15:25the time and I told you earlier that I
15:28am from Egypt originally I don't know if
15:30you have ever been to Cairo Egypt but in
15:32Cairo Egypt when you're driving in the
15:34street the stoplight is a hint you can
15:37take that hint or you can go it's up to
15:39you is it's not really enforced very
15:41heavily speed limits are also hints so
15:43my driving behavior is not as nice as my
15:46son's driving behavior because of that
15:48I'm still getting the lowest premium in
15:50the house so what many companies are
15:52doing now insurance industry across the
15:54world actually is putting these devices
15:56in the car that measure your actual
15:58driving behavior how you speed up how
16:00you slow down where you stop when you
16:02stop when you take a turn left or turn
16:05right are you using the signals when
16:06you're driving on the highway are you
16:07steady or are you wobbling all these
16:10kinds of parameters and that is what is
16:12used to compute your insurance premium
16:15not your age or your your gender or your
16:19income level or where you live it's now
16:21more a function of actually how you do
16:23and what's the probability of you having
16:24an accident because of how you're
16:26driving your actual behavior so that's a
16:28very very big shift in many industries
16:30not just the insurance industry we call
16:32this the segment of 1 the segment of 1
16:34so moving away from the segment of
16:36everybody to create a segment for every
16:39single customer for every single user
16:40for every single patient every single
16:42citizen so mentioning patients this is
16:45another example of using data to give
16:47voice for premature babies so premature
16:51babies there when they are in neonatal
16:55intensive care their brains have not
16:58really fully formed yet so their neural
17:00network that they have doesn't know when
17:02to cry so sometimes they will cry when
17:05they're happy and sometimes they will
17:08not cry when they're very heavy
17:09distressed which is a big problem for
17:11the for the nurses and the doctors
17:13trying to care for them so what our
17:15customers have done in this case is they
17:17built a predictive model that analyzes
17:20the signals coming out of the body of
17:21the baby how they move the sum of the
17:23brainwaves the pressure the heart rate
17:26the sounds they make and a screen would
17:29show up on the monitor above the crib of
17:32that baby saying what's going on like
17:34the message will show up and say I'm too
17:36distressed from the light in the room
17:38right now is too high please lower the
17:39light a little bit and the nurse would
17:41go and lower the light or there's too
17:42much noise right now in the room or I
17:44need to be fed so a language words are
17:47being now prescribed to them based on
17:50the signals coming out of their body and
17:53then this is a very standard example of
17:55big data being used for doing one of the
17:58most common use cases in big data that's
18:01action anomaly detection or the unknown
18:04unknowns or finding weird stuff
18:07finding weird stuff so MasterCard is one
18:10of our largest customers they have more
18:12than 10 petabytes of data in their
18:14warehouse and one of the most common
18:16attacks that hackers do when they're
18:19trying to steal money from credit cards
18:20is take a very small amount like two
18:23bucks three bucks to stay under the
18:25radar because none of us really call
18:26their company and when they say three
18:27bucks but we if we notice 100 bucks we
18:30will call them and by doing that over
18:32many many credit cards it becomes very
18:34hard for any individual or even any bank
18:36what we would mastercard to catch it but
18:37mastercard themselves because they have
18:39the view of the full data set they are
18:41now able to catch this very very
18:43sophisticated under-the-radar
18:45attacks using what's called anomaly
18:48detection so that's a very quick flavor
18:49of some of the use cases that we have in
18:52our customer base but in summary and
18:56because before I move on to the next
18:57section which is about startups we at
19:01Keller I strongly believe that there is
19:03a data revolution going on right now
19:06and this data revolution is as big if
19:09not bigger than the Industrial
19:10Revolution so in the Industrial
19:12Revolution what we learned as
19:14organizations and countries and
19:16enterprises is how to start leveraging
19:18machines steam engines and electrical
19:21engines to make carpets and make chairs
19:24and make stuff instead of using our
19:27hands right and that allowed us to
19:31significantly improve the speed the
19:33volume the throughput of manufacturing
19:35across the world and countries that were
19:37able to leverage that and of that became
19:40the leaders of the world and the same
19:42thing is happening now with a data
19:44revolution which is there is a lot of
19:46sophistication being built out there
19:48right now about how can we take data and
19:50automate decision-making using data
19:54versus having humans have to be in the
19:56loop every time before decision can be
19:57made and these decisions can be business
20:00decisions at a very high level but they
20:01call Sobe autonomous cars how we drive
20:03at the car if you look at how we drive a
20:04car that's really just about making a
20:06decision what should I do exactly at
20:07this right at this time and that will
20:09transform the world in very significant
20:11ways so that's kind of the motion the
20:13market the movement that Caldera
20:15trying to enable and now I'll move on
20:18from that to entrepreneurship I just
20:22wanna check how we're doing timewise so
20:23we have until okay good so I'll keep my
20:30eye on the clock so the club dress story
20:32so far we have four founders in the
20:34company I'm one of them so I came from
20:35Yahoo Yahoo acquired my first company as
20:38I told you earlier Mike Olson comes from
20:40Oracle and both me and Mike are still
20:42very heavily involved with caldera after
20:43the day and that was eight years ago and
20:46we sought the company there's two other
20:47co-founders who are less involved today
20:49Jeff hammer Becker from Facebook and
20:51Krystal Basilio from Google we raised
20:56our first round of funding round a five
20:58million dollars by VC called accel
21:00accel partners this famous from being
21:03the first VC behind Facebook is one of
21:05their largest investments and we raised
21:08that money in October 2008 that's when
21:10the financial markets were falling
21:11everywhere left and right because of the
21:13mortgage crisis it was technically so we
21:16were lucky at the beginning of our
21:17journey we were very lucky that we
21:19raised the money and we were still able
21:21to raise that money and find an investor
21:23that's willing to buy our pitch we only
21:25had a PowerPoint presentation and the
21:27four founders at the time telling them
21:29that we think there is a massive data
21:31revolution going to happen and is going
21:33to be called big data there was no need
21:34the name big didn't even exist back then
21:36when we said this so we were fortunate
21:39from that perspective but we started to
21:41deliver we started to get liver results
21:43start to have customers another VC came
21:46and said we want to fund you a grail of
21:48partners so we took another six million
21:49a few months later actually it's not bad
21:51it's like six seven eight months later
21:53and then another one came later because
21:57we kept doing well and the story just
21:58kept going we today a cloud era raised
22:01in total more than 1 billion dollars in
22:04capital in the company that's not our
22:05valuation that's how much money we
22:06actually took as a company and we're
22:09using that money to expand globally so
22:11we proved that the business model works
22:13in the US we proved how this can solve
22:15problems for financial companies for
22:17health companies for many texturing
22:19companies for governments and now we're
22:21scaling that across the world and when
22:22you're scaling you need money to scale
22:24actually especially in an enterprise
22:25software so we now have more than 1500
22:29thirty countries across the world now
22:32when you raise this much money and you
22:34make so many rounds you're gonna hit
22:36with what's called dilution and that's
22:37what I want to talk about in this slide
22:39so first show of hands in the room how
22:41many of you know what dilution means the
22:45word dilution means if you raise your
22:47hand it's okay if you don't know I just
22:49want to get a rough estimate so about
22:50maybe thirty percent of the room raise
22:53so dilution is the ownership you present
22:56ownership of a company so you can see
22:58this stage 1 and stage 2 as an example
23:00here when you start a company at the
23:02beginning ground 0 just like there's
23:04nothing in the company other than you
23:05and your co-founder you can you make up
23:09this random thing called shares shares
23:11is a random concept it's a myth that we
23:13made up that lawyers are made up and
23:14business and financial people made up to
23:17signify your ownership in a company
23:21right so we divide it's like you think
23:23about the company is a real thing the
23:24company is not of your thing the company
23:25itself is a story and I'm it as well
23:27but think about the company as a real
23:29thing like a pie and then you're gonna
23:31slice this pie into pieces and each one
23:33is going to own a piece of the pie and
23:35that piece is called the share and your
23:36present ownership in this company is how
23:39many pieces of pie you hold how many
23:41shares you hold so you start the company
23:43say at the beginning you're gonna make
23:4410 million shares so you're going to cut
23:46the company into 10 million pieces
23:47you're going to give each one of the
23:49founders or both of the founders
23:50together 8 million pieces 4 million
23:52pieces each one of them and then you're
23:54going to set aside two million pieces
23:56for new people they're gonna join your
23:57company new employees they're gonna join
23:59you so they get some ownership in the
24:01company as well and at this time when
24:03you're starting a ground zero every
24:05single piece in the in the pie is really
24:07worth nothing there's nothing yet the
24:09company is still starting up we didn't
24:11make anything so the actual value is
24:14it's very nominal so the way to think
24:17about this if you're looking at this
24:18slide is your angle your angle of
24:20ownership you present ownership in the
24:22pie is the number of shares that you
24:24hold say in this case 4 million shares
24:26divided by the total number of shares
24:28that exist right now but that number
24:30will change in the future so 4 million
24:32divided by 10 million is 40% that's how
24:35much you own that will be the angle the
24:37value how much money you own which is
24:40different than the number
24:41yours is going to be a function of the
24:43share price how much value we have per
24:45share which is the radius right so the
24:47more the radius the more value you have
24:48when you're beginning the company at the
24:50beginning you have nothing the radius is
24:52zero so you own zero even if you have a
24:54million shares ten million shares ten
24:56million shares doesn't matter the value
24:57is zero right now you go and start to
25:00raise funding so say Stage two here you
25:02went to raise money and you're talking
25:04to investors and then Wester said we
25:07will give you five million dollars but
25:10we want 10 million shares in the company
25:12right so you have to print 10 million
25:15more shares this fake thing that the
25:17ownership in the company the company now
25:19we're gonna have 20 million shares
25:20instead of 10 million print 10 million
25:22more for us and we're going to give you
25:245 million dollars for them which means
25:25that the value per share is 50 cents
25:28which means the total value of the
25:30company is what in this case 10 million
25:35so the value of the company is 10
25:36million because they gave you 5 million
25:38for half the company so the other half
25:40reasons to say that it's also worth half
25:43a million so the total company is worth
25:4410 million but now what happened is this
25:47dilution effect I was telling you about
25:48because now what happened is the
25:49founders together instead of owning 80
25:52percent which was 8 million shares out
25:54of 10 million they now own only 40
25:57percent which is 4 million 8 million
26:00shares still but the 8 million shares
26:02now are out of 20 million because we
26:03printed more shares and that's what
26:05that's what happens every time cloud
26:06that I went to is more money we printed
26:09more shares we printed more shares in
26:11the company so my ownership in cloud era
26:13I started with an R shape this big and I
26:15kept shrinking shrinking shrinking
26:16shrinking shrinking and it's a much
26:18smaller ownership right now and some
26:20entrepreneurs they freaked out about
26:21that and say oh I'm losing all of my
26:23value I'm losing no money you have to
26:25keep in mind that you're always
26:27comparing the percentage that you own as
26:30a function of the area of the slice
26:32you're getting yes at the beginning you
26:34had a very big percentage but the radius
26:36was zero it was worth noting now I have
26:40a smaller percentage I'm not gonna tell
26:42you exactly how much I own but let's say
26:43own 5% of the company so 50 percent but
26:46the radius is much bigger now the share
26:48price we have is up from a few pennies
26:51to now dollars tends to went to 10 20 30
26:55and the area of that slice is
26:57proportional to the radius squared every
27:00time you increase your radius by a
27:02little bit your value goes up squared
27:04compared to that the angle that you have
27:06so it's a very long way of saying don't
27:09obsess about dilution when you're
27:10building a very big company your goal
27:12should always be about value creation
27:14how can I increase the value of this
27:16company and increase the per share value
27:18that's what you should focus on 100%
27:21dilution is important you don't wanna go
27:24and just give shares to anybody of
27:25course you have to control it but but
27:26don't be shy about doing it because
27:28that's how you end up building a very
27:30big company at least that was my story
27:33and it's holding out well it's playing
27:35out well for us so now I have some
27:39almost done this is my last slide though
27:41this slide takes some time because it's
27:43a it has lots of stuff on it it's some
27:45of my advice that I'd like to give
27:46lessons learned as I was building out
27:48this company that I'd like to share with
27:50with other potential entrepreneurs and
27:53the first lesson is whatever you do make
27:58sure that the idea the thing the mission
28:01that your company is after is something
28:03that you feel very passionately about if
28:07you extremely passionately about and the
28:08reason why is the analogy and announce a
28:13funny analogy but the energy it's like
28:14you're getting married you're getting
28:16you're gonna be married to this thing
28:17you're gonna be working with this thing
28:18day in and day out actually more than
28:20more time probably then you put in an
28:22actual relationship and you're gonna
28:24have ups and downs you're gonna have
28:25many fights within the company gonna
28:27have many problems you can have a
28:28customer that's gonna cancel on you on
28:30the last minutes you have an investor
28:31that's gonna cancel in you on the last
28:33minute there's going to be so many much
28:34emotion and so much strife and the thing
28:40that's going to keep you going over and
28:42over and over again is the passion and
28:43the belief that you have and what you're
28:45trying to do and it's the same thing
28:47that I think keeps a relationship a
28:49marriage going as well is the passion
28:52and the love that you have for the other
28:53person despite all the bad things that
28:54are happening so some companies go
28:57through what's called the pivot a pivot
28:59is when you change what you were doing
29:00so you're doing this thing it didn't
29:02work out so now we're going to do this
29:03other thing if that happens and this
29:05other thing is not something that you
29:06feel passionate about you don't have
29:09art in then you should leave that
29:10company you're not gonna survive it as a
29:12founder you're not going to survive it
29:14as an employee you might but as a
29:15founder very hard because we carry a
29:17much bigger load than an employee so
29:20that's my number one advice by far and
29:21in fact I do some angel investments
29:23right now like I don't do a lot of my
29:24investments but I do maybe a couple of
29:26year and one of the key things I look in
29:28the eyes of the intrapreneur I'm talking
29:30with is are they do I see that the
29:33passion jumping out of their eyes I do I
29:35see this absolute love of the thing
29:38they're trying to do and if I don't see
29:39that they're not getting my money and I
29:41think all the VC is actually they invest
29:43in that way as well so it's very
29:44important that you have that and be
29:46genuine about it not don't take it you
29:47can't fake it it's very hard to fake
29:49passion number two and sometimes and
29:52sometimes people find this controversial
29:55you need to know how to hire great
29:57people but more importantly you need to
29:59know how to fire them very quickly as
30:01when you're starting at the beginning
30:03and you only have four people ten people
30:0720 people up to maybe 100 even more
30:10closer and I was 1500 so were past that
30:11point but in this very early days of a
30:13company of a startup you're competing
30:15with massive other companies like we
30:18were competing with IBM IBM has like an
30:20army of people that can totally
30:21obliterate us in any account we try and
30:25and and attack and the only way you're
30:28gonna win despite all these odds and
30:31despite these bigger competitors is by
30:33having amazing amazing people in that
30:35founding team that sounding team needs
30:37to be flawless in other words it's like
30:40a sporting team playing football or
30:43playing baseball or whatever everybody
30:45needs to be playing and everybody needs
30:46to be the best at the thing they're
30:48doing and when you're a very young
30:50startup as you will see in my next
30:52bullet point you always have the sense
30:54of urgency where you need to be moving
30:56very quickly as well and and advancing
30:58very quickly as well so you don't have
31:00time to train people like some people
31:02are good and I know if I were spend more
31:04time with them I can change them I can
31:06make them better but you don't have that
31:07time you don't have that luxury of the
31:08time when you're a small start-up so the
31:11right thing to do is to let them go
31:13right when you discover that one of the
31:14employees and this very early team even
31:16one of the co-founders is not really
31:19playing at the same level that everybody
31:20else is playing it's not putting the
31:22not generating the same impact it's not
31:24about time it's about impact or not
31:27fitting from a cultural perspective
31:28culture is very important as well then
31:31it's time to make a change and it's very
31:33hard to make a change but it has to be
31:35done and that's what I think that's what
31:37makes Silicon Valley's so special when
31:39it comes to this compared to Europe or
31:41many other Asia and many other countries
31:44are trying to replicate the same thing
31:45that we have here is our willingness to
31:47do it but our willings also do it in a
31:49very humane way in a very in a very
31:50decent way like we take good care of
31:52these people when we're letting them go
31:54we have them find a new job we give them
31:57some shares in the company still we give
31:59them a big severance package etc etc but
32:01it's important it's part of what you do
32:03later on when you grow as a company and
32:05you have say I would say more than 400
32:09people then now the impact of any single
32:11person divided by 400 is not read that
32:13as big so now you can afford to spend
32:15more time with them training them and
32:16coaching them and getting them into a
32:19good State so now today at Cloudera
32:20when we are 1,500 we don't fire people
32:22like that we take our time we do try and
32:25make them better but when we when we
32:27were very young when you only 10 people
32:28or 20 people and one is not doing well
32:31the impact is going to be felt by
32:33everybody the whole company will slow
32:34down and that's where you need to be
32:35quicker about making Corrections this
32:41lesson I'm sure you guys hear this over
32:42and over again it's one of the core
32:43ethos of a Silicon Valley and of startup
32:46culture which is don't be afraid to make
32:49mistakes right so we strongly encourage
32:53a cloud era since day one all of our
32:56employees to try to do things in new
32:58ways and not be afraid to screw things
33:00up just part of our culture and part of
33:03almost every startup culture here in the
33:04valley the key point is if we see the
33:07same employee failing in the same exact
33:08way over and over and over again that's
33:10bad but let's code failure there's a big
33:13difference between failure between
33:14failing right so failing one time and
33:17correcting and learning excellence we
33:19love that continuous failure very bad
33:21right and that's that it's very
33:23important to know the difference between
33:24these two concepts sense of urgency even
33:29today a cloud there are 1,500 people we
33:31thirty countries worldwide we're doing
33:33very well revenues wise but we know that
33:36many competitors coming up behind us and
33:38even when we were just five people that
33:41was true even now 1500 people is true
33:43and even if you look at big companies
33:44like Amazon or Google it's still true
33:46for them you always have to have the
33:48sense of urgency right like the thing
33:50that keeps you up at night is what's
33:52gonna happen next from my competition
33:54that might make them
33:55leapfrog me and how can I stay ahead of
33:57them that should never go away she'll
33:59always have that sense of urgency listen
34:03to your customers so I in the first five
34:07years of color I ran engineering as well
34:09now I focused mainly as the CTO so more
34:12on the technical aspects
34:13I don't really run engineering but when
34:15I was running engineering I would always
34:17tell our engineers you have to listen to
34:20the customers and build backwards from
34:22what the customer is asking for and they
34:24would counter me back and say are you
34:26telling us we cannot be innovative
34:27you're telling us we cannot innovate we
34:29always have to do what the customer is
34:30asking that means we are not innovating
34:32and and that's a fallacy that's a very
34:34big mistake innovation is not about
34:37innovating a new problem right if you
34:41innovate a new problem that nobody has
34:43nobody is going to buy the solution you
34:44have for that problem
34:45innovation is about innovating a new
34:48solution to an existing problem
34:49and the hard part that we have as
34:51product designers and as engineers and
34:54frankly the pitfall that we many of us
34:56fall on when we go for a PhD and become
34:58very heavy on the research side is we
35:00lost this connection between did we
35:02innovate the problem or is this problem
35:04a real problem that people have and
35:06before you start any significant effort
35:08you want to make sure that the problem
35:09is a real problem not one that you
35:11invented and one of the famous quotes
35:14that I like from Henry Ford Henry Ford
35:16he didn't really invent the car he
35:18invent the pipeline that allows us to
35:19make assembly line allows us to make
35:21cars efficiently but he's frequently
35:22credited with inventing the cars and the
35:25quote he said when I asked people what
35:27they wanted they said we want faster
35:30horses faster horses and this shows you
35:33your job your job as an engineer as a
35:35product designer as a company creator
35:38and entrepreneur is not asking people
35:40what they want right you always want to
35:42go back and ask them what's your problem
35:43what's the problem that you have and
35:45they will tell you the problem I have is
35:47I want to get from point A to point B
35:49and now you know okay now I'm gonna step
35:52out from that think completely out of
35:53the box all these perceptions about
35:55horses and how we used to do things in
35:57the past I'm going to forget them out
35:58and I'm gonna come up with a new
35:59innovative solution to solve that
36:01problem and that's how you innovate so
36:03never lose track of your customer and
36:05which problem you try to do and never
36:06invent the problem the failure that I
36:08see many entrepreneurs get into is
36:10sometimes they get so passionate which I
36:12love but then they get so passionate
36:14about a fake problem but only they have
36:15but only a couple of their friends have
36:17right and and that's obviously a very
36:20big mistake to fall into have faith in
36:24success so by this I'm saying positive
36:27energy is very important I truly believe
36:30that positive energy whatever that is
36:33helps you increase the odds of success
36:35for yourself and for your company you
36:38guys must have heard the stat from other
36:40people that talk to you only one out of
36:44every 10 companies funded round a funded
36:47not angel funded out of a Angel Fund is
36:49even lower only one out of every ten
36:51will do well two out of every ten will
36:55do okay and the remaining seven will be
36:58either loss for the investors and which
37:00means nothing for the employees once
37:01you're talking loss for investors you
37:02mean nothing for employees and founders
37:04or they will just shut down so in other
37:07words the chances of success is really
37:0810 percent for funded company that get
37:11around a of investment and if you look
37:14at company cloud or a cloud there are
37:16now is and what's called a unicorn
37:17category meaning we cost 1 billion
37:19dollars in valuation the chances of
37:21becoming that is 1 in 1,000 so it's even
37:23much lower over every 1,000 companies
37:25founded that have the potential to be a
37:27unicorn only one will be so in other
37:32words luck is very important luck is
37:35very important because the other guys in
37:38this other 999 companies they're also
37:40all of them working their ass off
37:42they're also all of them extremely smart
37:45just like you are and what will make you
37:49of course you'll execution and how good
37:51you are but your luck will play a very
37:52big part and the positive energy that
37:54you have inside of your company so with
37:56that actually also say if you are not a
37:58religious person you have to wish for
38:00luck all the time and luckily for us our
38:03if you are a religious person I usually
38:05say ask your mom to pray for you because
38:07you actually are gonna absolutely need
38:08it to tell things in your favor as you
38:10build out your company so with that said
38:12these are some books that I highly
38:14recommend you guys read they are very
38:18good when it comes to spotting trends
38:19and sponsoring what could be a a new
38:22major opportunity but in general very
38:25good advice for how to start a company
38:27how to build a product that people
38:28really care about they're gonna pay
38:29money for and just how to manage people
38:31and how to be good in life yourself like
38:34the seven habits obviously it's a very
38:35famous book about that and with that I
38:38can open it up for questions how much
38:40time do you have for questions we have
38:4210 minutes for questions and we're not
38:45gonna leave here for 10 minutes until
38:47you ask questions first question is
38:51there any specific industries there it's
38:56a very good question so I'll repeat the
38:57question do you think there is any
38:59industries that are underutilizing
39:01big data so right now we actually are
39:06seeing all industries getting very
39:08heavily involved in this space including
39:10some ones that we did not expect at
39:11Cloudera so for example we have
39:14customers today in the farming industry
39:17agriculture and farming I never thought
39:18agriculture and farming will be using
39:20big data but what happens is they have
39:22these sensors in the field that measure
39:23sunlight conditions irrigation levels
39:25and the rate of growth of the crops that
39:27cut the color changes and they you do a
39:29lot of a be testing experiments to
39:31maximize the yield out of their farms so
39:33I wouldn't say there is any single
39:35industry that's really not leveraging it
39:38but some industries are better than
39:40other in the speed at which they're
39:41leveraging it so absolutely the the
39:43faster-moving industry is now tend to be
39:45the technology industry the financial
39:49industry banks are very usually very
39:51early adopters telecommunication so cell
39:54phone companies and set for
39:55manufacturers and governments to our to
39:57our surprise governments are adopting
39:58this technology very quickly both for
40:00cybersecurity and terrorism but also for
40:03tax collection tax collection tends to
40:05be one of the things government's really
40:06care about and this technology is very
40:08good at catching tax evasion companies
40:11trying to hide stuff to smuggle money
40:12somewhere so lots of governments are
40:17all the way in the back I see you
40:18channeling Jeff besos quite a bit in
40:20your you know customer focus and all and
40:21I'm curious when you're a very small
40:23company with let's say almost no
40:25customers how do you how do you try and
40:27adopt this focus you know what what
40:29problems do people have if you don't
40:30really have people to focus on so the
40:34question is how do you have this
40:38customer focus when you're still a very
40:40young company that has no customers to
40:42focus on in the first place and that is
40:45the hardest thing I mean that the
40:46hardest stage of any company is that the
40:49first stage of discovering what the heck
40:51we're going to build and is this thing
40:53we're building solving a real customer
40:54problem that's the first stage the
40:56second stage where you prove that you go
40:58actually Unseld it and make money doing
40:59that and then the third stage which is
41:01the stage we are in right now is cloud
41:02era is the scaling stage how do I take
41:04that and scale so that first stage it's
41:08trial and error I mean that that first
41:10stage is you is usually where this 9 out
41:12of 10 companies will fail is that fair
41:14stage because first hopefully you have
41:17some good intuition of you yourself
41:19experiencing this problem so in our case
41:21we were lucky that I experienced the
41:24problems we were solving at Yahoo my
41:26co-founders experienced them at Facebook
41:28on Google but even before we had the
41:31conviction even before we started cloud
41:32era we still went out and we reached out
41:34to people in some big banks and some big
41:37telecommunication companies and we will
41:39go sit down with them and we will not
41:41tell them the solution we were thinking
41:43of at all we would tell them what are
41:46some of the problems you're seeing right
41:47now and just listen and and do pattern
41:50matching meaning are we say are we
41:53hitting the same problems caused
41:54instantly consistently over and over
41:56again and that that was our approach
41:58that was the approach that we took to
42:00try and validate that the initial hunch
42:03that we have is this problem that we
42:05know is real because we experienced it
42:06it's not just a problem for us and our
42:08respective companies but it's problem is
42:10bigger than us and yeah it's behooves
42:12you to always try and do that research
42:14before you kick off your company to make
42:16sure that in at the metal level you are
42:17solving the right problem to begin with
42:20and then as you start to actually add
42:22customers and have customers using you
42:23then it becomes easier they're on to
42:25always have these customer panels you do
42:27once a month or once every three
42:30listen to them and see what their needs
42:31are other questions I will ask myself a
42:36Peter question until you come up with
42:38another question which is a question I
42:41frequently get from students like
42:43yourselves should we finish our degree
42:46or drop out and start a company assuming
42:50some of you might be interested in the
42:51answer so if you look at myself I
42:54actually dropped out with my PhD to
42:56start my first company and then went
42:57back and finish it finished finished it
42:59PhD is different so PhD is a is a very
43:02long investment and my advice today
43:04actually to people is do not pursue a
43:06PhD unless you absolutely want the PhD
43:09for the pages sake in terms of your
43:12career assuming you don't want to be
43:15professor if you wanna be professor
43:16obviously PC is the only ways you have
43:17to go get a PhD but if you want a career
43:19in the industry especially in computer
43:21science and computer engineering you
43:23will do way better in the markets if you
43:25work right after the bachelors or
43:27masters masters also does help but PhD
43:29actually if you look at students that
43:30start at the same time as you and you
43:33you stayed for the PCE they start in the
43:35market their career overall longer term
43:36but the way back would fare will fare
43:38way better than then you would so that's
43:41when it comes to PhD programs and we
43:43have been discussing earlier with Roger
43:45yes he's seeing that demographic shift
43:47happened here at Stanford when it comes
43:50to bachelor's and master's I say
43:52absolutely try to finish at least the
43:53bachelor's degree and the reason why I
43:55say that is what I said earlier is the
43:58chances of success of a startup is one
44:01out of ten in nine cases out of ten your
44:04startup will fail let's lower odds than
44:07many of the games you can play in Las
44:09Vegas and not only that you will only
44:11find out four years or five years of
44:13your life after you invested whether it
44:15works out or not so it's good to have a
44:17bachelor's degree that you can fall on
44:19as a back-up plan to go and work
44:20somewhere well if if the startup did not
44:23work so my advice always is yes please
44:25do go finish your degree first to have
44:28that backup plan in your pockets
44:30other questions take a new question here
44:34and then we'll come back to you Oh
44:37computer science degrees what do you
44:39think some of the majors are they're the
44:40most valuable for careers at the moment
44:42I did a science so data science meaning
44:45statistics and operations research and
44:48that kind of trajectory it's I mean it's
44:51one of the most I would say even more
44:54variable than science right now like
44:56that the demand for these kind of people
44:57that can program so they have some basic
45:00programming principles under their belt
45:02but they really understand math and and
45:04and predictions and and statistics and
45:09advanced machine learning kind of topics
45:11way more needed as a scale right now so
45:14I would if I can force my kids to go
45:17towards some career it would be that I
45:19know it's gonna be an amazing career in
45:21the next 10 10 years yeah very good
45:23question all the way in the back
45:27thinking about like with PhDs and
45:30research and stuff is it common that you
45:32see people taking their HD research and
45:34trying to spin off startup coming from
45:37there because I know especially the tech
45:39industry is much easier to do that and
45:41under in other majors or other
45:43industries and I was wondering on your
45:47yes absolutely so I mean many many of
45:49the great companies here in Silicon
45:50Valley started because of that right
45:52because of the PhD research triggering a
45:55idea for a solution that would not have
45:59been otherwise possible VMware is a very
46:01good example of that Google itself
46:03actually is a good example of that so so
46:05I guess my answer the question is yes
46:14you talk about how your day-to-day job
46:16has changed from like each stage and
46:20like talking about we're day-to-day so
46:24the question is can you tell us about
46:25your day-to-day job and how changed from
46:28the first early days at Calvera to how
46:29it is today so I'll start today and then
46:31I'll go backwards today I'm the chief
46:33technology officer for the company and
46:35for an enterprise software company like
46:38louder the chief technology officer 60%
46:42of their time is about evangelism
46:44so it's traveling across the world
46:46talking with customers and at public
46:48events about how this technology can
46:51change their industry in very
46:53significant ways so that they become
46:54very attracted to the technology so I
46:56sometimes joke and say CTO stands for
46:59chief talking officer or also chief
47:02travel officer I travel a lot I travel
47:04I'm like last year I traveled like five
47:08hundred thousand miles last year alone I
47:11made the three loops around the earth
47:13last year in the opposite direction so I
47:15lost three days of my life
47:16because of that you don't get them when
47:17you cross the deadline this way you
47:18don't get the day back so that's today
47:21the remaining 30% of my time is about
47:24the strategy of the company overall and
47:28the culture of the company so as a
47:30founder with my founder hat I still play
47:31a very big role in setting the culture
47:32of the company the type of people we
47:34want to have working for us and then
47:35from a strategy role are we building the
47:38right things not for the next six months
47:40but for the next five years that will
47:42keep us alive so I think Gregg
47:47Papadopoulos who was the I think he was
47:49the CEO of some CTO of Sun was CEO of
47:51Sun has this very famous quote he said
47:54to give you the nagi of what a CTO does
47:57versus the VP of engineering VP of
47:59engineering doesn't implementation CTO
48:00does that kind of division it's very
48:03similar also the VP of Sales and the CFO
48:06so if a company misses their forecasts
48:10you don't blame the VP of Sales as much
48:13as you blame that CFO that did not
48:16project the math and the science that
48:17they're going to miss their forecasts so
48:19in the same sense if a company misses
48:22the deadline for delivering a product
48:23this quarter or next quarter yes you
48:25blame the VP a vision engineer
48:27but if the company builds the wrong
48:29product for the next three years you
48:31fire your CTO like the CTO is the one
48:33that kind of supposed to own that and
48:35make sure you're building the right
48:37thing so that's kind of what I do today
48:39and in fact I have a very long article
48:40on this called what does a CEO do so if
48:42you want to learn what the CTO does from
48:43my perspective you can search for that
48:45online and you will find it if I roll
48:48back to four years ago I was mainly
48:50playing the role of the VP of
48:52engineering which is taking the
48:54requirements from our product teams from
48:55our customers and making sure we deliver
48:56that on time and hiring all the smart
48:59people that we have in the company
49:00and if you rolled back eight years ago
49:01when we were starting the company we're
49:03doing everything right like like we I
49:05was assembling the desks for people when
49:07they come in I'll go buy from Ikea desk
49:09and make the desk for them where I was
49:11setting up their laptops with the
49:12software they're gonna need to be able
49:13to be productive in the company from
49:16accounting and having QuickBooks to
49:17manage the the salaries of people and
49:19what we gonna do so it was really at the
49:21beginning you're doing everything which
49:22was fun but only for a short time and
49:24now it's fun but in a different way so
49:26that's kind of the trajectory of how how
49:28it went and with that I would like to
49:30thank you very much for for being here
49:31today and hope this was useful and
49:32informational for you thank you
![Welp, Here's what the Fed *Just* Said [Fed FOMC Minutes]](/_next/image?url=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fl8u23oq1E9E%2Fhqdefault.jpg&w=3840&q=75)
