00:00Do we move away from any of the learned performance already, or do you double down on the conviction that
00:06you already have?
00:06Is it still about generative AI and the tooling?
00:09Thanks for having me. Yeah, I mean, we have been investing in AI since the very beginning.
00:17And our focus has really been partnering with founders at the earliest stages,
00:21when founders are really starting out with a brand new idea.
00:24And we've backed hundreds of companies, and we've been fortunate enough to partner with these companies from the very beginning.
00:29Some of which are in an apartment, or our founders that have never even been in business before.
00:34And our focus is on helping them scale up and really find the next round of funding,
00:39and to really partner with these founders all the way through IPO in some cases.
00:43I mean, you've been doing this since 2017, since before ChatGPT, since before AI got cool, dare we say it.
00:49You've backed, what, 500 AI founders already.
00:52Talk us through the wins that you've already seen, because look, the LPs need convincing at this moment.
00:55And where were you able to show your prowess to those to come back and reinvest with you?
01:02Yeah, I mean, you know, I think over the course of the last 10 years that we've been investing in
01:06AI,
01:07we've seen the number of companies that come through our doors really accelerate and increase.
01:11We used to see about 100 companies in AI per year, and we were sort of the oddball venture fund
01:17that was really, really niche in looking at really technical people that were researchers
01:21or people that were not necessarily focused on building big businesses in the eyes of a lot of other venture
01:26capitalists.
01:27But we saw that the ChatGPT moment was going to come.
01:30We saw that the Transformer paper, which came out a month before Gradient launched,
01:35was really going to be the epoch moment for AI.
01:38And, you know, my partner, Zach, and I, we saw that the ability for people to use these models
01:43as sort of an intelligent buddy or an intelligent co-pilot was really going to be the future of the
01:48Internet
01:49and the future of technology.
01:50And we partnered with these founders.
01:52Now we see 2,000 companies a year come through our doors and pitch us with some new AI technology
01:59or AI capability.
02:00And we have to pick about 10 to 15 of those that we decide to partner with.
02:03So the job has gotten more difficult in terms of volume, but it's also gotten a lot more interesting
02:08because of the number of technologies and number of founders that we see and what they are working on.
02:13Everything from wafer scale compute companies to open claw renditions or, you know,
02:18individual models that are for specific tasks.
02:20What's also got more difficult is just the size of the rounds.
02:24Look, we've just had the biggest sort of seed.
02:26What was it, billion that Jan LeCun has just been raising over in Europe?
02:30I mean, coming out with 3 billion in terms of valuation,
02:33you're saying Miramarati raised an extraordinary fund.
02:35People are calling these like mango seeds or avocado seeds, as Ed loves to say.
02:39How can you compete?
02:41So that's a really interesting question is that we actually think that there is a bubble
02:45in certain parts of AI.
02:47There are bubbles.
02:47There is a bubble in foundational model companies, contenders to open AI, contenders to Google,
02:53contenders to XAI and Anthropic.
02:55And those are companies that we don't fund.
02:57What we really want to be focused on are AI infrastructure companies and AI application companies.
03:01And those seed rounds tend to be really well priced and tend to be ones that we can invest
03:06in and lead with an average $3 million check and own 10 to 15 percent of those companies.
03:11We actually don't see the valuation issues in the areas that we want to focus in when
03:16it comes to AI.
03:17When it comes to a lot of these other companies that are raising a billion dollars here or there,
03:21it seems as though those are a lot of venture capitalists or investors that want to chase the
03:27next open AI. But realistically, the moat at the model layer is just capital.
03:33How much money can you raise, you know, because you buy data and compute.
03:36And so where we focus, the valuations are favorable and the round sizes are favorable.
03:40And we believe that that will continue.
03:43What is it like in terms of diversifying your LP base? Because it has a very interesting story
03:48gradient. It came out of Google, but then you brought on external money. Does that change the way
03:52in which you have to invest? Because lots of people have different strategies. The why they're coming
03:57to you with their money.
03:59It was extremely challenging to raise from LPs, especially in this climate. I mean,
04:03we're very fortunate that we were able to get it done in about nine months. But, you know, before for
04:08our prior four funds, Google was the single LP and they've been an incredible partner. I don't think
04:13we could have launched an AI seed fund in 2017 when everyone was looking at crypto without the help of
04:19Google. And so now I think we decided we wanted to diversify our LP base mainly because it gives us
04:26a lot more capabilities. We have partners in the LP base for our companies and we also wanted the
04:31ability to sort of scale up our capital base and have that flexibility. But it was challenging because
04:36a lot of people haven't seen liquidity, especially in the institutional LP network. Fortunately, we found
04:42LPs that believed in our vision and mission and wanted to partner with us over the long term.
04:46And now we have a collection of not just Google, but many other amazing institutions,
04:50including companies like MercadoLibre, which we're very closely partnered with, as well as a number
04:55of other insurance companies, institutions and foundations. And so I think that given the climate,
05:01it is challenging to raise, especially for a seed fund when you have Andreessen and General Catalyst
05:06raising 10, $20 billion funds. But I think we've been able to articulate our strategy in focusing on the
05:13areas where there isn't a bubble, when valuations are favorable and where we can partner with these
05:18founders when it's just an idea and mature and grow with them.
Comments