StrictlyVC San Francisco: The State of Investment and AI With Elad Gil & Sarah Guo

So elad uh has been is is maybe known Most recently as the world's biggest Solo venture Capitalist but is I'm not that tall but I appreciate It um but has been involved in a variety Of projects and companies over time um And I'm delighted that he's here uh Sarah is um so interesting to me to the Child of entrepreneurs who founded a Communications equipment company Kasa Systems uh she worked for them as a kid Um like the children of entrepreneurs Often do and um then went into Investment Banking onto grey lock where She became one of the first female um General Partners there and then about a Year ago announced that she had raised Her own fund for conviction Capital Which is really great congratulations on Your oneyear anniversary thanks Scot um So you know as a starting point there Aren't I mean this whole sort of idea of Solo GP is still kind of New and fascinating especially at some Of the scale that we're seeing which is Unprecedented so elad I know that you Hate talking about numbers but um can You as a starting point just confirm I Mean there were reports in November that You had closed a$ 1.1 billion um third Fund yes yeah so I think um it's really Interesting because you mentioned solo Capitalism is a new thing or you know

This the there's all these articles About the rise of solo capitalist or Things like that and it's kind of Interesting because if you go back Through the history of venture one of The things that I think is really Striking is Arthur Rock for a long time Back in the 80s and part of the 70s was Investing by himself and then in the 2000s you had a variety of people Nar Ravikant Michael Dearing a variety of Folks um sort of running various funds Or Investments or doing other things on Their own and so I think it's actually a Very old concept and something that People have been doing for a long time And then in parallel there's a number Evental Capital firms where there's Somebody who's hired a small team and That person really is the central Decision maker but it's it's called an Institution and so I think there's Always been kind of this blurry or hazy Line that maybe has more to do with Branding than reality in terms of this Trend um and so in many cases I don't Think a lot of this stuff is actually That new and you've seen people with Small funds with big funds with the Whole range uh throughout I think the Last 30 40 years I think it's it's um Well you're right that it's not new but The scale is is new I mean I don't think We've never seen anything quite like

This before I mean according to reports You've raised about $2 billion do since 2021 I think there's a a few firms that Um are called firms but then you look Into it and it's really like one person With some leverage and in some cases They have um you know very large funds And billions in management and all the Rest of it and I think really it just Comes down to um again sort of how They're positioned or what The Branding Is or other things like that so I think A lot of these Trends are are it's Interesting when I invested in Airbnb um every couple years what they Were called shifted even though they Were doing the same thing they were Sharing economy there were marketplaces There were all sorts of things and I Think that's true of venture capital Like a lot of things have been sort of Repositioned over time in part I think Because people find it more interesting Um but I think the reality is that again There's been these longer term trends That have just been ongoing so um I Honestly think uh what I've been doing Uh has gotten more attention than Warranted like I don't think it's that Interesting but I think it's fine if Other people do so it is it is Interesting well um and Sarah I want to Know a little bit too about how you Operate your firm so I know that you

Have you're not completely alone but Virtually I mean tell me a little bit About how conviction operates and maybe How much I mean I think you're investing Aund million doll fund into AI Companies yeah we uh we have a baby Little1 million fund um uh doing early Stage AI investing and uh we have five People today so so uh it's me two Investors a talent partner somebody Working on uh sort of portfolio Operations and so I'm used to having um A small But Mighty team I think rlock is A relatively small partnership but you Know Venture changes a lot of the funds That we're more focused on early stage Over the last decade became much larger Multi-stage firms and so I think the Landscape of people actually focused on Like early stage Boutique investing is Is much thinner than it used to be and Elad just to be clear you still don't Have I mean you don't have other Investors on your staff uh yeah I don't Have any be uh making Investments I've Always had people who provided me with a Little bit of Leverage like um Shin Kim Was my chief of staff before and uh he Went on to start a company and so I've Hired some people to sort of backfill That role um so I I have people sort of On on the team now but they do a variety Of things some of them are quite Technical so for example um uh my new

Chief of staff is sh and Jane who joined Me from ramp where he's an engineering Manager so uh he and another person David song Who Um dropped out of Stanford CS to join memi have joined me And the two the two of them have done Things like um they've built an Embedding playground so they basically Swap in and out any underlying Vector DB And any embedding framework so that we Can play around with different tools so They do a lot of stuff like that which Is really fun because then you can get Really Hands-On with like some of this Really interesting new technology I'm so Sorry I'm actually having trouble Hearing you everybody so loud guys could You please just quiet down a little bit Um I'm glad everybody's having fun but Yeah I feel very left out I'm like Should guys let's get is over here how's It going you know like the Uh okay not this I promise the last Question but this is you and I talked in The spring and this kind of blows my Mind somebody who's been covering the Industry for so long so you you're Managing a substantial amount of money And you can also still you're I think You'd said because I'd asked like how Does it work and you're like well Sometimes I invest my person personal Money and then if it's a bigger deal I Can invest my personal money or I can

Use fund money or put package together Spvs so basically you know like they've Backed up the truck for you your LPS and Then they've also said whatever you want To do is fine as long as we get a piece Of these Deals yeah you know I think Fundamentally what I've learned in life In general so for example when um my First company that I started was Acquired by Twitter uh one of the things That I pre-cleared was the ability to Invest while at Twitter and we had an Agreement that anything that I wanted to Do would go to the um GC for approval Okay and if you didn't approve it in Five days I would just go ahead and do It and it was less about whether Conflicts of interest existed or not but More that they were disclosed that there Were bright lines that there was clear Guidelines and so I think um a lot of What I do now I kind of took from that World of operating in terms of thinking Through how do I make sure that Everybody is clear in terms of what the Intentions are everybody understands how I'll operate and nobody should ever Question it because I'll always always Try to do the right thing but I want to Be very clear up front about what I'm Doing and not doing and vice versa and I Almost feel like anytime you have a Business deal with someone you almost

Want to create an upfront fiduciary Contract what do I owe you and what do You owe me and how are we going to work Together over the long term because These things are hopefully multi-decade Relationships for the set of people you Work with over and over again over time And so I think if you have that Clarity Of how you're going to act and I often Advise this to Founders or Executives I Work with um it makes a huge difference It gets rids of it gets rid of ambiguity It gets rid of uncertainty it gets rid Of the feelings in it and you just point To here's what we discussed and we wrote It down and so everything's fine and so I've taken that approach here as well Yeah there's a lot to be learned from That that's great well both of you I Mean people want to be in the business Of you know Saga and and allad guilt uh For obvious reasons you've got great Portfolio companies historically and Presently you're an investor in mistol Which is an open AI competitor that just Raised something like $415 Million um from entry in and light speed In December at a $2 billion valuation I Did want to ask I mean you you're so You're writing very great Le Texs both Of you lots of firms are coming in Behind you Nea Sequoia and these are Big rounds and These are really big valuations I guess

I just wondered do you think the Valuations in the space are getting Ahead of themselves we've seen this Story play out time and again or there's Something fundamentally different Here so I think you can do two things One is you can look um imp IR L at the Outcomes that are happening like Outcomes like I think about real data That's either you know users or revenue Or an Enterprise Value in the public Markets right um and uh in the interim Of having like public AI native Companies we have open AI with you know Publicly disclosed Revenue numbers um North of 1.5 billion now and uh and a Lot of companies that have real Traction In AI based on users and revenue I Forget what is public about Harvey but It's a company that um a lot and I are In together and they had like a really Good first year selling like legal Co-pilot and so um I'd say like yes There are definitely going to be Companies that are very overpriced I'd Also say like I am biased in this point Of view but um I'm a large investor in My own fund right like I actually need The companies to work over time and like I think there is real customer value I Did a reference with um Uh a end user of one of my portfolio Companies the other day and he was Telling me that he went to a trade show

To look for a solution that did what This company offers because he thought It was so important to his business and I'm like that you know he doesn't care What investors think or what the hype Cycle is he just sees an opportunity um And I I I think the capability sets Quite powerful um can I also ask Sarah So you're investing $100 million Everything counts how are you investing That money I mean how many how many Portol companies do you aim to have at The end of this fund yeah I I feel very Lucky in that um I guess this goes to Like less planned investment strategy When I left Greylock I said I don't know Much more than I want to do early stage Investing I like being very Hands-On With companies I like working with them As they scale as well but I like being There early and then I know how uh I Know how portfolio math Works in terms Of like more ownership means better Returns um and so I I was very lucky to Have a set of LPS that were very Long-term oriented and didn't want much More information than that and so if we Just go back to lad's framework of like What is the Um uh I guess like Financial agreement Mine with my LPS was like I think I'll Have a relatively concentrated portfolio We're going to be doing early stage Investments and I promised to make you

Money so I didn't commit to a certain Construction um and I think we're Figuring it out as we Go elad if you don't mind my asking Because you do have a lot of money to Deploy I'm guessing because you won't Confirm it um How um what are the like biggest checks That you're writing what AI company has Raised the most money from You uh it depends on how you define AI I Mean maybe andreil Andre yeah the Defense tech company I mean I've been Investing with them and working with Them since their seed round so I met Them when it was just the four Founders And then um I participated in almost Every round since then I led their Series D um which I think was a $450 Million round not for me but overall is Around um but there's a variety of Companies that I've invested with over Time where I just invested early and Then kept going stripe would be another Example of that um some SAR I invested In multiple rounds that's now a public Company um and a number of them are uh Sort of Prior generation of Aid driven Right samsara is an AI company coupled With Hardware Andre is AI company Coupled to Hardware so uh those I think Those may be two examples great um you Know I also wanted to just talk to you About some of your portfolio companies

Because I'm still still trying to figure Out how different these companies are Than predecessors one of yours is Perplexity which has raised more money From Nea um so this is a company that's Trying to take on search uh essentially It's taking on you know Bing and which I Guess is not that hard to do maybe um And Google but what gives you the faith That this company can uh you know this a Sustainable Enterprise yeah I think there's a couple Things so um I invested in perplexity I Actually LED their preed round when it Was just Arvin the founder and so um I Heard that he was very smart he reached Out to me and I pinged uh he worked with Peter Abel at Berkeley and I think Peter And Peter was like this is one of the Smartest people who's worked with me and So we started meeting and brainstorming Ideas and originally he was going to go Into a very different direction in terms Of being a NextGen um sort of data Co-pilot and as he was iterating on the Product more and more people kind of Wanted to use it for search and he just Followed that pull of what people wanted And he ended up building this really Interesting res search and search and Answer tool and you know fundamentally There's a variety of ways to think about It um you know it's it's I think one of Them is basically some proportion of

Share of search and a tiny fraction of Search is still a massive market so it Could just be Tam Centric um I think It's Pierre lond I don't know if you if You ever yeah Pierre is um I think he's 80 years old now or something he's still Working and you know amazing um person Who operated for decades and then joined Seoa and he was a partner at SEO And one of the things he always used to Say was that um and I'm sure he still Says again he's still working um I don't Want to make it sound like pumis when it Doesn't um is that he'd much rather be In a giant Tam going after a small piece Of it than in a a small Tam with a big Piece of it because at least in a giant Tam there's always moved a room to move Around there's room to expand you always Have kind of Headway and in a tiny Tam Or total addressable Market you get Constrained by the market market and Then you get stuck and I think Perplexity is a great example where even A small slice of this massive Market Could be quite interesting and I think The you know Arvin is one of the most Talented people I've worked with in Quite some time and in general I should Say um in the AI world one of the things That's been amazing over the last two Three years that I've been doing Generative Ai and I've been doing AI Investing for maybe 15 years now is um

How good people are again like how good The founders are there's been a huge Step up in terms of drive hours worked Intelligence Ambition um uh p uh genuine organic Salesmanship right and so I think the Level of Founders and that's you know um A variety of people that um I've backed Over the last two three years I've been Very impressed by them and Arvin is one Of those people who keeps stepping up Incredibly smart uh very aggressive Cares a lot about what he's building Cares a lot about his users and I think When you see things like that those Usually uh lead to very good companies And good outcomes and I'm sorry I don't Know this is he a a Serial entrepreneur No this was his first startup he did his PhD at Berkeley and then he went to open Aai okay and then he started this out of Open AI got it got it yeah you mentioned Peter ail um and I wanted to talk to you Sarah about mistol so it's going against Going up against some stiff competition Um it publicly released its Tech so that People can develop their own chat Bots Open AI others said this is not you know Necessarily responsible just wondering What you think about that like this is Not responsible yeah not responsible Yeah Um Ian the threat of disinformation Etc The threat of dis the threat of abusive

Use exactly yeah um okay so I I would Start with um I think all technology That is like General platform technology Can be used by actors good and bad right And the question is just like where Those safeguards sit and who gets to Determine um what is good and bad and Who gets to use it Um because they have to sit somewhere uh I I Think traditionally like we don't figure Out how to restrict the internet we Figure out how to restrict applications On the internet um and what that allows For is like a million flowers to bloom All the applications that we want to use You know people can put them online in a Decentralized way and then we can decide Which ones when they grow and are Impactful uh are you know uh responsible To society in some way and should be Constricted or face obligations uh I I Think mostly people feel software should Generally operate that way um I I think The MRA point of view is they care a Great deal about safety but there is a Balance between safety and Innovation And it's not clear uh that you know a Very small number of labs with a very Small number of decision makers should Make the decisions about safety um if You draw that parallel to the internet Do you worry that they're French and Therefore May impose French values on

The rest of the world you know 4-Hour Work Week wine for Lunch like that's the thing I'm worried About with mrr actually is the potential For that to happen and so I wonder if we Should kind of shut down alarmist when I When I go to bed Sak blue you know it's Like what it's just so I don't know That's my concern okay I will say so Talking about being a alarmist I mean What is I do Wonder as investors just Generally speaking when you are um Funding these companies yes a lot of People will say it's the end of the World you know where is a taking us You'll say that's not you know don't Worry about it but I mean if things are Used For um purposes they weren't intended For like how do you protect yourselves If you're involved with these companies And there you know it's really Interesting because if you look at the History of the 20th Century um if you so basically if you Look at AI safety there's really three Areas uh one area is like Digital Risk Misinformation bias things like that Um a second area is physical risk can an AI make a virus can it do bad things to You can it create a bomb whatever derail A train and then the third is Existential risk like are we worried That there will be some species

Competition over time right and people Tend to mix those three risks and some People will shout about regulation Because they're worried about the Existential stuff but they realize that The misinformation and everything else Is stuff that we've dealt with in other Contexts throughout the history of uh Humanity honestly if you go back to the 20th century and you go through each one Of those risks r s um misinformation we Had the Spanish American war because of Yellow Journalism uh physical risk we had two World wars we had Mass famine Existential risk we had genocides and Attempted genocides repeatedly humans Did that without AI people did that to Each other and so I think we need to Remember that the thing that is the Biggest threat to us in the short run is Other people and we see that Increasingly in the global theaters in Terms of everything that we're seeing in Current events right now and I think People kind of forgot that and um so I Think it's really important to kind of Reground and say okay what are we Actually worried about why do we already Have tools and legal Frameworks for it And in the context of misinformation and Other things we do um and then how do we Think about these threats evolving over Time relative to the Arc of progress The

Arc of technology and also most Importantly The Arc of humanity because At least right now I'm dramatically more Worried about other people than I am About AI although in the long run I Think AI could be an Issue um you know I'm sorry I'm still Having a little bit of trouble hearing But I heard most that um I wanted to ask About foundational models I think when We talked you'd said you kind of had I Don't know if you'd said you avoided Them or you hadn't really funded them But I wondered if that was by choice or Circumstance I'm guessing that you could Have if you yeah I've made some um Foundation model Investments over time So mrr despite the French thing okay um Is one um there's Other it was a big hurdle right I had a Yeah I was up late at night you know I Was kind of questioning what if the Values spread what if it's French Canadian value vales is that good or bad And you know yeah exactly they have Poutine so you're like that's a good Thing but it's also Canada so you know you kind of go back And forth on these things and it's a lot Of late nights a lot of drinking you Know I don't want to talk about it Actually I think um it's TMI right now But uh but I think if you look at I Think one one of the things that's very

Underd discussed is the breadth of Things on the foundation model layer Right so there's the big llms there's Image gen model there's video there's Audio on the diffusion model side as Well as some Transformer stuff there's All the various biology models there's Chemistry models now Material Sciences Robotics and so there's so much to do With the foundation model layer in so Many different ways um and some of these Are going to forward integrate into Applications right that's open Ai and Chat GPT that's a variety of Things um but some of them will also Vertically integrated in biology and Healthcare and they may vertically Integrate in robotics and then some of Them will just be Standalone models and So so I think the future world is this Really interesting mish mash and to me The biggest wild card right now that I Don't understand is um what happens with Meta inv investment in llama and there's A few scenarios that could happen there And really the way to think about llms Is what's at the frontier and what's Behind it the frontier most advanced Model right now is Gp4 and this year there's going to be at Least four companies that are likely to Launch a gp4 level model and I think Open a and I don't have inside knowledge About this but I'm assuming open AI will

Hit like bt5 or 4.5 or they should be a Step ahead because they've been so far Ahead against everybody but the question Is 5 years from now does llama catch up And do we have an open- Source model That is as good as any closed- Source Model and if so that creates somewhat of A flip in the ecosystem because the Value will go in part it'll still stay With the foundation models but it'll Also migrate to the cloud providers the Hyperscalers who are running these Models will absorb a broader share of Dollars and that's a really big shift if It happens and so I think there's a lot Of very big unanswered questions that Are both really interesting Intellectually but also make certain Investments difficult to understand or Contemplate and sometimes uh as a Founder or as an investor you have to Take the leap and sometimes you say well The world isn't certain enough then I'm Also okay waiting and so I think that's One of the big open questions about Foundation models that's really Interesting I mean I keep thinking There's going to be oligopoly but I Maybe the players are different than I'm Guessing they're going to be I I guess Also you know so you're you're we're all Guessing but like in 10 years what it Still look like um it does today where We have some Power Players and then

There the hierarchy kind of looks the Same or do you think we could live in a World where there are just the three Power players and that's it I mean how How do you yeah yeah yeah Sarah what do You think I don't know anybody that Works in AI today that is comfortable Making a 10-year prediction I think most People are pretty uncomfortable making a Two-year prediction actually just in Terms of pace of change in the industry Um and so like as somebody who uh you Know was a seed in series a investor in Mraw like I'm going to honestly say um I Uh I am shocked at how effective models At the like small or 8 by SB size are Right um and if if you just take this Framework of like what's at the frontier What's behind the frontier and behind The frontier you're going to have very Capable things that are open source um Uh you know I I think that the question You asked is is really complicated Because um you know there's a version of This that's an AGI future we like I Don't know that there's three players Maybe there's one right um I think There's a version of it where uh the Availability of llama 2 and mraw have Enabled a whole generation of Application companies um to explore like New use cases that are much more Sophisticated and it is very very hard For a single Foundation model player to

Go do the product and distribution and Post trainining and data collection to Get into all of the crannies of the the Universe that use technology or could Use more Ai and so I think there's a Huge I think there's a trillion dollars Of value to be created and captured in All of these different applications and I think it's very unlikely that most of It goes to a few players I think the Foundation model layer is likely to be Very important no matter what um uh I Think the open source movement has Changed my view of how much of an Oligopoly it's going to Stay You Agree I think um at the frontier it's Most likely an oligopoly and then behind That it's Fragmented right guys thank you so much For joining me so fun and I'm thank you For not making fun of you for Mispronouncing Mraw a thanks for having Us really a treat thank you so Much