this morning legendary venture capitalist vin owed khosla in a rare interview. he sits down with me to talk about what he sees in the year ahead, and big data company ayasdi, finding answers where others didn't each know to request the questions. asking the questions this week, our reporters far rad man jaw of the "new york times" and laura sidel of npr. this week on "press: here." >

. >> good morning, everyone, i'm scott mcgrew. the point being even elite pilots have levels of elite. the same is true in the world of venture capital on sand hill road. there are lots of venture capital firms and like fighter pilots they are a confident bunch but there are some that are more elite than others. at the very top you will find just a few. kleiner perkins and jason who are wits, draper fisher and khosla ventures run by vinod khosla. joined this morning by far rad manju and laura sidel of npr. >> you know a lot about a lot of things and you had a background in sun micro estimates, but as

you invest are there things that you don't know but you suspect would be good investments and how do you learn about the thing that you know nothing about? >> do you know what's misunderstood in my view about venture capital is it isn't about investing at all, it's about backing big ideas to go pursue new things and most of the returns in venture capital come from new ideas that you don't imagine before. >> how do you understand the company with the new idea that has nothing to do with your background, it's in biotech or something in which you are not familiar? >> i think you have to have a growth kind of mindset where you don't make a lot of assumptions about what you know. what i know is how little i know and i start with that assumption and i imagine the possible. i let great entrepreneurs take their ideas and try and make something happen. it's surprising how many of the

large ideas didn't start out the way they -- they initially thought things would turn out. >> so i was wondering what are your thoughts, though, right now, the markets are going down, the nasdaq is way down, all these tech companies, some people have thought we are in a bubble. do you think this is a bubble bursting around here? is it going to affect the investor climate here? >> so two things are important. first, in the venture capital business the real venture capital business, what happens in china, what happens with oil prices has nothing to do with whether an investment is a good idea or not. in fact, it's puzzling to me why stock prices should go down when oil prices should go down, they should go up, the global economy will do better. and china is such a small part of the u.s. company. the u.s. companies should do just fine, but too much of wall

street is about following the herd and believing each other and reading press articles and sort of panicking. >> so you are not panicking right now? >> i'm not panicking. i don't pay attention to the stock market. i barely ever look at what the stock market is doing. what's important is when a new idea in venture capital comes your way, whether it's a good idea important will not be here for five years. >> one of the five years that i think is true in the vc world is some of the best investments have been made in down periods. are you in some way -- if we're heading for a down period are you in some way looking forward to that? >> you just want a down period to happen, don't you? >> buying is on the cheap now, right? >> there's buying and selling but in our business you can't do buy and sell in the usual way because our timelines are so long. five, seven years in an investment.

but it is true when everybody is negative the best ideas don't get funded and that creates the best opportunities. >> now, the converse to that, vinod, is that in the greatest of times all kinds of ridiculous ideas get funded. have we seen that? >> we see both. when it comes to money there's only two emotions i think investors go through, one is fear, the other is greed. people bounce between these two walls and the key is to stay in the middle. do we get over optimism and things funded that shouldn't be funded? all the time. it's people's mindset. >> your fund has funded some things, door dash i'm thinking specifically, wonderful service, love it to death. it seems to me those kinds of servicing, delivering your dry cleaning, are the ones that are most vulnerable because if the kid who is working over at twitter loses his job he is not going to get a door dash which

means door dash, et cetera, et cetera, et cetera. why those sorts of companies? >> well, every company is different and at each stage it becomes a different kind of investment, but if you're fundamentally adding value for the long-term, what door dash is or isn't isn't going to be clear in the next year or two, it will be who dominates that space in three or four years. i only ask you to look back at uber and say -- >> this is true. >> -- specialized limo service, who would take that and how could it ever be worth more than all of the taxi businesses in the country combined. >> you said you don't look at the stock market but there has been this problem in the tech industry in the last year and a half or so where it's been difficult for these unicorns, these private companies to go public and that's ultimately how you guys get paid. does that worry you? >> it doesn't because those unicorn valuations are

artificial anyway, right? so, you know, when public market stocks go up and down people don't worry. somehow they seem to attach more importance to a private company's stock going up or down. i don't pay attention to it. things change. the environment changes, but the fundamental value you're building shouldn't change and what door dash becomes or uber becomes in four or five years will depend on what they do. if uber replaces all public transportation, for example, because uber pools with driverless cars is cheaper than sam trans, your local bus service. >> and more efficient. >> and more efficient and it's door to door you're going to create value. >> one sector you're interested in, clean energy, and that's a sector where i do wonder about the price of oil having an impact on that because it seems

like if the price of oil goes down people get less interested in clean energy. >> no, it's absolutely true that depending upon the price of competitive technologies, in this case oil or gas, alternatives get more or less interest. but at this stage if you look at the clean energy sector, things like lighting are doing very well, that's clean energy. in certain parts of the world solar still doing great. it really depends. if you have a much better power generation technology or a refrigeration technology it will do fine. we've done a lot in clean energy in agriculture, that's very, very interesting. so it's a very broad area, not a narrow voi. >> how depend are your investments or interests in government subsidies in clean neshlg? would you increase subsidies if

you could, decrease them? how dependent are they? you know, my car and my solar are partially paid for by subsidy and i love it. it's great. but they are not going to last forever. >> so we never i'm vest on the basis of subsidies. you know, if they are there you take advantage of them, but you can't bet the long-term that subsidies will increase the value of an area. if something scales to be large it can't have subsidies, it's just a rule of budgets. >> but isn't there come sense -- i mean, if oil is cheap people particularly in the consumer market it wouldn't seem as important to them to put in the solar panel or do all those things to bring down their bills that are helpful to getting -- >> i think it's fair to assume that today and five years ago you could have assumed all price will go up and oil price will go down and that over any ten-year cycle both will happen.

you can't be sensitive to that so you pick areas. some are more dependent on price of oil, others aren't. but frankly in our business there's so many new areas that are changing. like even data science, for example, effects energy. it's a different way to last in that area building iet men zags. >> i need to interrupt you as you introduce the concept of data sensors, there's a big data company i need to ask you about. i need to do that after i pay a bill. we will be back with "press: here" in just a minute.

we're talking with vin nod khosla. there's one thing i want to get to but some of my viewers know you as the guy with the beach. may i ask you about how your beach is? you have a great coastal property and you are in a battle with the state of california. before we get to the thing i really want to ask you about which is ajasdi, what about your beach? it's not a topic i want to talk about, very clearly it's a dispute around property rights and most of the press hasn't gotten the basic facts right. >> it dates back to the treaty of hidalgo. >> no, it doesn't. it dates back to the coastal act in california. >> there's a press guy getting it wrong. have you taken it as soon as it needs to go? will you take the case as far as it needs to go? >> i won't comment on what i

will or won't do? >> will you comment on ajasdi? >> it is this big data company and we can talk to the ceo in a minute, but the way it works is complicated. this goes back to my very first question of when they came to you or did you come to them? did you understand immediately what it was they were doing? >> i -- as soon as i met ajasdi i was very excited about the company. what has happened is the proliferation of data is so large that the way it's used is changing very, very rapidly. any change creates great opportunities, but what's even more exciting is there's so much more value to be created out of the same data that we already have that we can process differently. >> no question. big data companies are trying to do that, but did you -- top logical analysis, did you understand what that was when you met with them?

>> i think i did. >> okay. >> i have to find out if they think you did. >> here is the fundamental thing, when data increases orders are issued and it is increasing. the idea of an oracle database becomes irrelevant. you can't do things the same way. it's still valuable, oracle still valuable. what you do new and what's possible changes because there's so much more data. i will explain it a simple way. almost all the time when people used to look at data they'd say if i do this will sales increase? that's a very linear way of looking at data. new tools like ayasdi opened up the world to questions you didn't know to ask. is a different word or phrase going to be better at advertising on tuesday morning

versus thursday afternoon? it's not a question you would normally think about and we humans are limited in the questions we can think to ask of data. ayasdi is awesome because it can figure out what the right questions are. that has been an area about expertise and judgment in the past and this is a whole new feed radically different. >> is it the kind of thing where you can -- i mean, there are a bunch of these kinds of data analysis companies. is it the kind of thing where you know when you see what these guys are doing versus what other guys are doing that this one has a special thing? there are a bunch of search engines and then google came along and it was obviously that google is better. is it obvious here or is it hard to tell? >> early on it is never obvious, but what ayasdi was doing is so dramatically different. we've looked at hundreds of data

companies. every other company was trying to solve a different problem. google is very good at driverless cars, that's a machine learning problem. you can do specific problems. the idea that you don't know what the right questions are has frankly not been addressed by any company. every company tries to take one mathematical or al gore rhythmic approach, not to get too technical. what ayasdi does is tries all the possible approaches. >> i want to interrupt and let laura have one more question. we have only a minute left. >> the idea that it's visualized, that you can see the data and it appears like in a 3-d way how or not is that or how different is that? is that what everybody is doing now, they are trying to show us, visualize data for us? >> no, it's much more than visualization. there's plenty of visualization companies in their surface and those companies and traditional

data companies are about looking for your keys under the lamp post no matter where you lost them in the night. >> right. >> that is what they do. you say this area i can ask a question, i can think about it. what ayasdi does is say i don't need to know what you think the question is. we will help you figure out what the right question is. >> vinod khosla thank you for being with me. i need the rest of the time to talk to gurjit singh. and we will talk to ayasdi --

welcome back to "press: here." if you are just joining us we were talking to venture capitalist have a node could say las about ayasdi. he has joined the u.s. government's department of defense in funding the big data company. big data is a big buzzword. one of our challenges is describing how ayasdi does big data differently through topological analysis. the shape of data can tell you something about the data. stay with us, it's easier than you think. this is a graph of my salary over time starting with my first job out of college. the shape shows -- the shape of the data shows i'm generally on the right path. here is a graph showing eruptions of old faithful, shape tells us there are two a day, one is much bigger than the other. that's easy stuff one year

regression. let's step up to what ayasdi done. this is a graph of data of knee replacements performed by a president who. i cannot interpret this data at all but the mathematicians and computers at ayasdi can and it tells them all kinds of things. gurjeet singh is the ceo and one of the founders at ayasdi. i was readings your dissertation i got pretty much past the title page. how would i do explaining what the hell it is you do. >> it's very simple. we learn to see the world in shape. imagine fonts, we can express the letter a in so many different forms. topology is a branch of mathematics that was invented to deal with this notion of shape. what we did at stanford and at ayasdi thereafter was essentially take this old area of mathematics principles and apply it to data.

>> go on. i'm still unclear as to what that means. so does that mean -- is it like geometry, so, you know, we know what the shape of a triangle is and it has a mathematical formula, is that what you mean? >> that is exactly what i mean. in fact. so think about we saw a shape on the screen which was your salary, doing brilliantly of course. >> thank you very much. >> but also imagine that you look at your customers, right, and oftentimes as a large enterprise you would look at your enterprise ant you will want to segment them. segmentation, it's actually a shape problem, you are trying to decompose your data into multiple distinct pieces that don't talk to each other roughly. similarly imagine that you are an economist and you are looking at the growth rate of the gdp versus the fed interest rate. it's a cyclical phenomena, when the interest rates are low the economy goes. >> your math and your computers can handle all kinds of

questions. >> we were talking knee replacements, but i have a graph of violence in iraq and that is a graph somehow of the violence in iraq and your data scientists can learn things that the soldiest did not know they needed to ask. that's phenomenal. >> and what would that be? you know, this was something that we were just talking about, which is the idea that you would help people ask the question. >> right. questions. >> how? >> imagine when you search something on the web, you go to foogel, you have an idea, you turn it into pretty much an english sentence, hit go and see a list of results and then you select one or whatever. now, when you're dealing with data i imagine in knee surgery you have as a hospital system millions of patients and for every patient you have tens of thousands of interactions. formulating your question into an english sentence is no longer possible. right? so these shapes in the data they

allow in a very simplistic way for somebody to circle a region in a slap and say i see this as being a distinct part of the shape, why is that? and the software will tell you these are all the knee surgeries in which the outcomes are great because the drug that was prescribed was a neuro toxin as an example. >> it sounds like a general purpose thing. who are you selling to? who is your sort of customer base at this point? >> so commercially we sell into financial services and healthcare and we also do some business with the government. >> can you talk about the business you do with the government? >> no, i cannot. >> well, getting back to, say, that graph, you know, what would somebody -- what question would somebody ask looking at a data graph like that, you know, what would pop out? there's all these red dots over here, what do they mean? >> america. football players,?k+w soccer players. >> you look at the shape of all the soccer players and try to find a shape correctly it would

look like the letter y you say what distinguishes the arms from the y to the base of the y and why is that the case. >> do coaches do that or doctors do that or does ayasdi do that and get back to the bank or soldier. >> >> our customers do that. in many cases an end user might not interact with the shape directly. for example, as a hospital system before a knee surgery is going to be performed a physician would look at the emr data, the patient data that's already in the emr -- >> the stuff that doctors can understand. >> yes. and the view that they would see would already be a very simplistic version of the shape actually, that it's der syed rizwan farook from the shape. >> did you have to develop special products for the hospital version of ayasdi and soccer player version? >> that's correct. that's what we do as a company. we have a general purpose platform and as a company we are building specific products for specific -- >> i know this is going to sound like a dumb -- maybe it's a dumb question, but i keep wondering, okay, but so what if i can see

that shape is different. can you give me a practical thing that is learned from, you know, saying, oh, what's that data that then was applied and -- >> absolutely. let me give you one of my favorite examples. so when we started ayasdi before we know that investors -- it was very early on, just three guys and one of the first data sets that one of our collaborators looked at was this old breast cancer study. if you imagine this data, you have a few thousand patients, i think it was a few hundred actually and for every patient you measured 20,000 genetic markers. right? it was a 15-year-old data set, very old data set and the result of this analysis was a shape that looked like the letter y. now, when you colored this network, you color these networks by various outcomes, by the survival rate of the patients, you saw that clearly in one part of the y patients did very well. their survivals were very high and everybody was fine. on the other part things were

pretty grim overall, but it turned out that within that smaller population that the population that did not do well there was a smaller population whose prognosis was very bad but still ended up surviving with breast cancer. if you think what's really profound, it's a decade and a half old data set, it's opinion studied to death, pretty much the first data set about genetics of breast cancer and this researcher within minutes was able to discover a segment of patients who were supposed to not do very well but ended up surviving. >> then you can ask why. >> that's correct. i can't describe it because it's very complicated. >> you can't describe it. >> i can't describe it. i can tell the researchers -- >> you have a ph.d. from stanford -- >> in mathematics. >> okay. fair enough. now, i've only got a minute left. there are lots of big data skps. how do you sell yourself? if i'm a company that doesn't understand big data to begin with and that's why i hired you and far had has a company and you have a company how do i know yours is better if i the customer don't understand what's going on behind the scenes? >> i lot of the big data

companies today are doing bigger versions of the things that we used to do with small data. so we have had business intelligence tools for a while, now there are companies that do business intelligence on big data. we have had quarry optimizers for a time, companies that do quarry optimizers for bigger data and these are all great. like vin nod said, oracle is great, we need those things but the big complexity that nobody is trying to solve today is the complexity of asking the right things. right. we no longer have the ability to ask the right questions. >> i like the way that vinod put it and that was that if you're looking only under the lamp post for your keys you're asking the wrong question. >> that's right. >> and that data can't give you the answers. all right. gurjeet, thank you for being with us this morning.

that's our show for this week. my thanks to my guest vinod could say la and gurjeet signing of ayasdi.

those interviews and all our past interviews are available at "press: here" tv.com. i'm scott mcgrew. thank you for making us part of your sunday morning.

the blizzard of 2016 has thankfully come to a close, but it certainly has left a mark. record snowfalls on the east coast, up and down the i-95 corridor. more than two feet of snow hitting our nation's capital. as a result, the nhl showdown we had hoped to bring you, the pittsburgh penguins and washington capitals, will have to wait for another