Billionaire Vinod Khosla: AI Presents Opportunity To Build A Trillion Dollar Company

[Music] thank you but let's start with where AI is leading us so imagine a word 20 25 years from now and let's work backwards not forwards because forwards tends to be incremental I suspect 80 percent of 80 percent of all jobs that are economically valuable and I don't care about AI philosophers but 80 of all jobs that are economically valuable in 80 percent of those functions will be capable of being done by an AI and I say capable because I don't want to get into regulation and all that kind of stuff where things may slow down but that's 64 of all valuable economic jobs globally we'll be capable of being done by an AI that's the opportunity for entrepreneurs in the audience if they want to build a trillion dollar company that's an incredible amount it's orders of magnitude bigger than Google's current market if you look at the world that way so what is the opportunity my answer is almost everything and let's look at some areas in and I talked about this I wrote a Blog on this 10 years ago do we need doctors almost certainly the best doctors will be an AI and I suspect the first AI only doctor not AI doctor but Ai and not AI assistant doctor but the AI only doctor will happen within five years from today that means the FDA will approve an app to be a primary care doctor and we're working on that I'm pretty sure they have the FDA will approve an app to be an AI psychiatrist or an AI oncologist probably a little bit later if somebody is an oncologist in the audience call me because oncology nobody is doing with AI uh so that's medicine but will we have ai musicians we have a company whose goal is to produce a top 10 music hit Untouched by a human being on singing it composing it playing the instruments composing the instruments everything end to end music media art generation game generation and then all the Enterprise functions business analysts that'll be done by an air so no matter where you look you're gonna see that so one of our startups actually uh this morning pitched on that company that you just said you want to know if anyone's doing something on that so I think I can help you out um so vinod I'm sorry that you missed out on investing in open AI uh you know here's the way I should explain this you know why did we invite four or five years ago actually just just to clarify I was joking vinod was the first to invest in open AI so I bet no one has asked him how do you feel about not investing it because he did invest in it so that was being ironic ladies and gentlemen vinod uh well thank you in the in the process is pretty interesting right um let me take you back a long time ago I started sun in 1980 uh sorry 1982. right and the internet was just beginning but it was already growing exponentially and the early parts of an exponential are almost flat and the knee of the curve on the internet growth didn't happen until 1996. in 1982 we made tcpip standard in every machine and people just subjected to that customers did uh I won't spend too much time 1986 when the browser happened and I was the Kleiner when we invested in Netscape and Amazon and then Googled soon after that the internet was happening but the largest Venture return and this is an important story came because we expected the exponential growth of the internet in 1996. when every major Telco said they would never use tspip in the public network and I want to take a dig at experts every experts at the core of the public backbone would be ATM technology for those of you old enough to have heard that term and when we started Juniper which turned out to be a 2500x return so 7 billion in as distributed profit for Kleiner on a three or four million dollar investment it was because we believed in this exponential long way of saying fast forward to five years ago or actually when I first wrote my blog on do we need doctors and I wrote a companion blog in January of 2012 do we need teachers because I figured most teaching would be done by an AI and uh and on the non-profit side our families developing an AI tutor my wife's non-profit is doing that uh it was clear we were in the 1980s era of AI where it was growing exponentially innovating exponentially but not visible by five years ago when we invested in open AI in 2018 or 19 I forget one of those times it was clear in five years we were going to see an exponential capability increase and and that's when we invested but it was very uncertain nobody else believed it but that's the best time for entrepreneurs to jump into an area aim frankly to your investing audience that's the time to jump in it's when your most uncertain when the conventional wisdom isn't everybody believing in it which is when valuations go through the roof it's when you're early in predicting these tracks um so really really important to pick these points here's what I would say I think GPT 4 is a transition point and of course the world's imagination caught fire with chat GPT last September which by the way wasn't even a planned product open AI hadn't planned on releasing that product uh didn't know whether they were going to release that they were going to wait for gpt4 as an alternative think of GPT 5 and gpd6 they will be as big a jump from GPT 4 as GPT 4 was from GPT 3 and GPT 3 was from gpd2 so this is just the beginning of exponential capability and for the entrepreneurs in the audience I'd say skate where the puck is going not where it is almost ignore gpt4 say it's a great prototyping tool where do I want to be two or three years when GPT 6 is around so you sort of touched on this but how can Founders and yell if you're you're a founder [Music] how can Founders build something defensible in AI I feel like you've given a little bit of advice but anything more before I go to the big picture so this is the hard part if I'll give you the following example I looked at the recent YC batch when they when open AI released one feature which is plugins about half the batch got obsolete literally they were doing something minor with chippity or gpt4 and almost half the bags plans were obsolete as soon as plugins were released and you could connect in interesting ways with other systems so it is hard for another no most of the YC what's rappers around gpd4 which is also not defensible so I'm talking about what's not defensible yeah you're not answering the question I asked I I am getting to the questions I'm first saying they don't okay then I'll do the tell you that they'll give you five stars in this question until I hear hear the yeah yeah the the by the way I don't need five stars from you I don't need validation for you so important as I said he's irreverent I it does an important message for Founders don't listen to what others are saying the most important thing the founder does is decide Whose advice to take on what topic way too many people will give you advice and are qualified to give you advice so as that was in the side back to what you do do first you assemble great teams teams that can stay ahead of the Curve um uh and I don't think uh Patrick will mind me saying that Patrick Collison and I had lunch and we both decided after an hour and a half we were confused about how things will evolve now think about it he's played with every model himself he's confused I'm confused what can you do to solve that the big big deal is assemble a great team that will follow the development and get ahead of them and know how to leverage that so many startups I say are taking a software engineer who just happened to be relabeled ml engineer that's not good enough to build a resilient long-lasting company um the second thing I would say is look hard at what everything that others can do quite easily and why aren't they doing it what happens when they see you being successful and follow you that will happen and so that's another don't but that says look at the areas where you will build defensibility there is industry specific knowledge I don't believe industrial specific data will matter as much but industry specific knowledge will matter so take a larger vertical paint a five-year vision and say gpd4 can help me do this and I'll build the other pieces when gpd5 comes it'll be a lot more powerful uh we'll do these other things so build a path over the next five years with a great team that can anticipate the development of features um that's to the people building applications on gpt4 but I suspect there's some people in the audience who says this is not the end-all and large language models is not the only way to do AI so we are actively investing in alternatives approaches to large language models bigger and bigger models um those of you who are in the audience who understand what I'm saying um hold on raise your hand if you understand what he's saying all right A lot of people put their hands up raise your hand if you are willing to admit you don't understand what you're saying okay the majority did understand just so you know smart group yeah all right so I have another question but keep keep going so there are approaches Beyond large language models that should be investigated in a way and whole startups could be started uh what will look like open AI five years from now as a technique or an approach and I can tell you there's some really interesting work at MIT I've seen over time um that could be an alternative to large language models or add a tip to that model so there are ways to build differentiation both in the application domain in the fundamental model domain and in assembling the right teams to go after uh these products most startups either don't have weight AI co-founder or they don't have a great product sense co-founder and you can't hire a product manager in the space you have to have a Founder with founder Authority who can file who can Zig and Zack to the right product formulation and hasn't the moral authority of a founder so assembling your team is key most of all you need the right in advice um and I'm well known for saying most investors add little value and many add negative value if you Google that my name pops up why because they take a very and and this is a message to investors take a longer view than short term because people take a very short term where can you generate Revenue fundamentally the big problem with YC startups I saw in this batches they're all focused on one metric what is their MRI or ARR and what is the growth rate of that not for their defensibility will be wrong approach in this market because most simple things will be obsoleted um so big picture I hear a lot of people saying that mobile and the internet is pretty big but this AI moment is bigger can you imagine something in the next hundred years it could be bigger than this AI moment well I never say never but currently I can't think of it if most human jobs can mostly be replaced um then we'll be in the era of great abundance and GDP growth and productivity growth and being able to be more creative with a creative AI music art storytelling all done by us all the way to oncology and medicine and so in web Robotics and AI yeah robotics is another area we haven't talked about yeah huge opportunity in robotics obviously a lot of opportunity in all the enterprise software stuff so almost everything you can imagine for those interested in in looking at the enterprise software I would urge you to watch Satya nadella's uh talk on the future of work that he did recently I found it incredibly impressive I literally called him and told him how impressive it was um it imagines a different way to work so let me build on that so people won't um will likely won't have to work in uh 40 to 30 years as they're working as we imagine it now but until then AI could create significant wealth inequalities as it disrupts Industries like do you think aai is helping you financially so yeah so you're so you're doing okay but what do you think about this in an inequality for some that may not cash in on this moment so seven years ago when I refer people to this um I wrote a Blog in I think it was Forbes or Fortune um it was in Forbes that said air would cause this was 70 years ago great abundance great GDP growth rate productivity growth all the metrics economists use so great time and increasing uh income inequality that's going to be a major problem and the solutions to that will be social Bill Gates has talked about our robot tax maybe there's an AI tax I don't know the solution but Society will have to address income disparity which will be increasing not decreasing and it worries me probably the single biggest thing about AI that worries me and our inability to realize that we are an era of great abundance and we have the resources to support the neediest and in fact I think Universal big big basic income which I talked about seven years ago in this blog towards the end of the blog is something that will be a possibility because per capita income if I'm right by 2050 or so will be so high and by 2100 just through the roof um so that we will have the ability to support people uh whether we do it or not is a political question yeah so you're pretty busy in 2016. you said a lot um hey blind spots we have a lot of MIT people here who's with MIT yay so don't say something so he can't see everyone so everyone say if you're with MIT go yeah yeah okay so what do smart MIT folks and and the public not understand about this AI moment and and chat GPT like is there things that that the misconceptions like say something profound and provocative well first I would say since I have written a lot uh I would uh so those of you interested in large renovation in medicine uh I wrote a 100 page document about 88 years ago called 20 doctor included and I'd recommend people read the transformation in medicine I invasion uh eight years ago that I said would take 25 years and now I think will take less than 15 years from that time um so what does that mean I think people are expecting a way what's going to hit us is a tsunami all right well I did do you want to finish that thought I have my last two questions um a tsunami is a very different phenomenon than a wave have we had a tsunami in our in your lifetime In Our Lifetime not in business you know the largest transition we saw was in agricultural employment which went from 50 of us employment in the year 1900 to a few percent by the 1970s but that took a long time so we had multiple generations to adjust and relocate people and people to move to the cities I think this would be so much faster we will have a hard time adjusting and that'll be a big challenge for society there's of course Ai and defense and warfare and cyber security I worry about that a lot I've already a little less about AI safety and sentient not because they are unimportant they're very very important and should be the primary area of government funding for AI research it should be the Principal area of research but my my bigger track I worry about is a powerful AI in in our adversaries and like Russia and China uh going after us both for hot power which is Warfare and cyber warfare but also soft power like economic influence over the rest of the world I worry about that a lot all right so vinod I got two final questions and I'm going to say in both at the same time not over on top of each other I'll say one and then the next one all right you talked a lot about the last seven years ago what you said what do you think the next five years in terms of tech progress is going to look like this program is called imagination in action like there are a lot of people are imagining but not making action happen talk about what you think the next five years look like and what action people should be doing that's one question and then my last question which I know is very dear to your heart uh the environment where do you see opportunities for AI and sustainability what bets are you you looking to make that both can do good and just turn to sustainability if we don't do it this decade you know we're going to be in real trouble but also where are some of the business opportunities because if we get those out there there could be you know a synergistic thing going on so five years in environment well the one thing I would say is the rate of development is hard to predict so I would only say anybody who says what will be possible in five years is wrong and smart people won't predict the where what will be developed or not but what I will say is mostly what people imagine when they say AI won't be able to do X they're more likely to be wrong than right um and especially within the set of things we as human beings can think of developing because we think more linearly than the exponentials we are in the middle of the tsunami or the typhoon this isn't a wind it's a typhoon we're in the middle of uh it'll be more disruptive and by the way disruptive means not fun to be disrupted very hard for those who are disrupted but huge opportunities font knows and also for you the investors in your audience you know I actually just um thought of saying so Mark Twain and and Shakespeare get quoted a lot I don't know if you get put in the same line as that but you get quoted a lot is there anything you've been quoted on that you think people got wrong oh hard for me to thank let me answer your climate question maybe if I have time I'll yeah think about where I've been quoted wrong yeah or they don't necessarily understand the intent of the of the point that they just run with it for sure like a shark talk so climate Solutions will take hard science it's a different problem um AI can play a role there but AI will not be transformative in climate unfortunately that I've seen I've looked at the nips papers on climate uh Ai and climate they're playing on the edges I didn't find a single compelling paper on ai's role in climate so far so if any of your audience has seen something uh and I did my survey probably 18 months ago please send it to me my email is ovk which is office of BK at Coastal adventures.com uh but climate is about hard science I'm very excited about Conwell Fusion Bob momgaard was a senior fellow at the MIT plasma Fusion lab when I first met him and we helped uh sort of get that thing going and I'm very excited it can be transformative in fact Fusion in air the two things that can be most transformative of the planet um there is only a dozen areas in climate that matter and a dozen people could change the climate picture dramatically over the next 15-20 years I won't go get into the details of that but I wrote a Blog about it in July of 21. uh so you can find it someplace um but ai's role will be there like better Material Science discovering a new material AI could do a good job I recently asked what AI can do to develop a room temperature superconductor which is not transformative itself but could be hugely helpful um catalysts room temperature superconductors but by and large the mechanical world the bits world is driven by bits phenomena could you design a better bug to mine uh take lithium out of brines probably I hope somebody is working on it in BIO and lithium uh bio and Mining would be really interesting biomining is a huge area of interest for me personally because of how bad the current extractive Industries are extracting lithium out of seawater plenty of it but just very uneconomic to do today could you improve that 104 yeah you know let me let me interrupt for a second I'm going to retract that last question because I decided it's not so good so our last speaker had dinner with kaifu uh the other day China versus uh um Us in terms of AI I'm curious your your take on that what we need to know I also want you know you have a lot of friends here Kim Pelosi's in the room and she's waving uh Ramesh rasker I know you've done a lot with him over the years Taylor who I introduced you to he's in here he's very happy that you guys are working with him so you got a lot of friends in addition to a lot of people who've never met you so thank you for being you I and and I'm glad you brought this up it's a good note to end up I believe we are in a massive techno-economic war with China this is not going to be pretty it will be good for the planet either party wins it's good for the planet but I want Western values to win over the Chinese political system and that will depend on who has the best technology in a couple of areas AI being dominant because of its Global impact the second is Fusion I think our energy is the second area those two areas will determine who has economic influence and hence political influence globally and I want Western values to win so I consider us at war with China on the front that's a very different kind of War than the one we worry about in Taiwan or Ukraine ladies and gentlemen the node [Music]

Channel: Forbes

Views: 33,770

Keywords: Forbes, Forbes Media, Forbes Magazine, Forbes Digital, Business, Finance, Entrepreneurship, Technology, Investing, Personal Finance

Id: RnjD-nhutK0

Length: 28min 26sec (1706 seconds)

Published: Fri Jun 30 2023