- 2 days ago
Category
🗞
NewsTranscript
00:00An early Anthropik investor who now runs AMP, an AI infrastructure company building compute
00:04capacity for Anthropik and other AI startups. Andrea, we have a lot to talk about today,
00:09but let's just start with Anthropik. I was listening to this podcast where you talked
00:13about how when you were one of the founding investors in Anthropik over at Andreessen,
00:17you helped basically get them in front of 22 VCs and 21 of them said no. What has it been
00:22like for you to watch this company, which at first couldn't get anyone on board to them,
00:28to have everyone literally lining up the door for what might be a record-breaking IPO?
00:34Well, first off, thanks for having me back. It's always great to be here. So yeah, to correct
00:39the record, I invested in Anthropik before I joined Andreessen Horowitz.
00:42Right. Yes, you were an angel investor. Thank you, Ange.
00:45Yes. And then the only reason I bring that up is because at the time, unlike most institutional
00:53investors, I was an angel investor. And so I had the privilege to partner with the founders when we
01:01were all operators and fellow founders. I was running a company at the time. I had just sold
01:07it to Discord. And I would say it was frustrating because those of us in the AI community, my previous
01:13AI business was called Ubiquity 6, and we all used to hang out in the machine learning community in San
01:17Francisco. And we were all just very technical. And so for those of us who understood what was
01:21happening with the science and who read the scaling laws paper that, you know, Tom and Dario and team put
01:26out in 2020, we just said, hey, this is very simple. It's a very simple recipe. The more compute and
01:33the more
01:33data we combine in the right combination, the more predictably we can keep increasing the intelligence
01:38of the model. It really was not that hard to understand. And yet I found that all those 22 investors,
01:44those VCs, those institutional firms that LPs who are pension funds and sovereign funds who trust,
01:52they trust their retirement money with these VCs, they were just being let down. I mean, I think that
01:57was the most disappointing part was these folks who are supposed to be stewards of capital just were
02:02not technical enough to understand that this technology was going to change the world. And so
02:07I did round, I invested my life savings at the time and rounded up some friends because I felt like
02:12as a steward of capital on behalf of the like, of the LP community, it was important to get
02:18somebody technical who understood what was going to happen to have, give these pension funds and
02:22retirement funds and so on, exposure to this extraordinary wealth creation opportunity,
02:27which by the way, is not anthropic specific. It's a whole ecosystem. You know, since then,
02:30I've made multiple investments in the Frontier ecosystem, including Open Router, which we just
02:34sold to Stripe, you know, this week. And there's extraordinary wealth creation. But I'm just,
02:39I was really frustrated by 21 of those 22 institutional firms who are supposed to be
02:44stewards of capital just failed. They're just a miserable failure, right, of judgment, of
02:50professional upgrading. They weren't upgrading their skills fast enough to understand this
02:53technology. So that the punchline is, we need a reset in the venture capital community. For whatever
02:59reason, the existing set of leaders are just not staying fresh enough. They're not updating their
03:04skills. And they've let down a whole generation of LPs as a result. I mean, if you look at
03:07Anthropics cap table, and if you look at a number of other Frontier labs as well, the institutional
03:12venture capital representation in these cap tables is a joke.
03:16That's really interesting. I mean, yeah, I want to ask more about it. But first, I just have to ask,
03:20because I think everyone in our audience will want to know this. You're still invested in
03:25Anthropic, right? You held on to your stake. You didn't sell?
03:29I did not sell. In fact, I think I've tripled down, quadrupled down since then. It is our single
03:34largest portfolio position at AMP. AMP has a venture capital business called AMP Foundry.
03:40It's our largest position. I can't share publicly how much we've put in. But let's just say we're
03:45meaningful investor. I mean, this is public. If you look up Anthropics, last two rounds, the CSG and
03:50Series H were mentioned there. And yeah, if anybody's buying, sorry, selling, I'm still buying.
03:57Wow, their IPO is going to be a good day for you. And can you talk about what this does
04:01to the ecosystem, though, to have you were talking about the wealth creation,
04:05what it does to the VC ecosystem to have these mega IPOs come and to be sitting on the on
04:11these
04:11huge paper gains? Well, I think the paper gains are turning into liquidity pretty fast, right?
04:17That's what we're I'm pretty confident that I'm actually really excited for the public markets
04:24to get some access to these equities because they are much more rational than venture capitalists.
04:30Traditional venture capitalists think that they're good judges of technology and fundamentals,
04:35whereas what I find is retail investors, public market investors understand math much better.
04:41I mean, the anthropic business is very simple. I love simplicity. We take a dollar of compute.
04:46We turn it into ten dollars of intelligence and that's it. We just rinse and repeat. It's just
04:50scaling laws. And I think the public markets are going to reward this beautifully. And I think
04:54the liquidity will then hope, you know, traditionally flows back into the ecosystem now to the next
05:01generation of firms, right? As startups, the cycle of life of startups is the leaders go public,
05:07the liquidity flows back to venture firms. Those venture firms then invest in the next generation.
05:11With Anthropic, that's not going to happen, right? Because remember, the institutional VCs failed
05:16at their jobs. And so instead, that that responsibility is going to fall on the individuals
05:21at Anthropic, the co-founders, the early employees. And my hope and I'm actually I announced this
05:28yesterday on Twitter where I'm running an ecosystem program to train angel investors, the younger operators
05:33to give back and use their family office capital to invest in the next generation of VCs as of companies
05:39as angels. Because I think the institutional VCs are it's it's time to.
05:43And what's going on with that? I don't want to go on too much of a tangent. So I want
05:47to talk about
05:47OpenRider too. But this is really a fascinating point you bring up. Are they have they have they
05:53had their come to Jesus moment considering what's happened with Anthropic or is something still broken
05:58with the institutional VC investors?
06:01Completely broken. You know, I think the egos of I hate to say this, but the boomers are just so
06:08hard to get
06:09I mean, they just can't get over themselves. Right. They've had they had successes in the 2003 era.
06:15They made their money in the Internet boom and they're just sitting on these. It's like tenure faculty
06:20positions in academia where where some percentage of them like it as an example, Mark and Ben are amazing
06:24at Andreessen Horowitz. They constantly update themselves and innovate. Unfortunately, the next layer across
06:30the industry, people who are not as innovative as them are are just it's copium. We like to call it
06:36the younger folks in the industry. We like to call it copium, which is the cope where they keep
06:40thinking that the next generation of investors like us keep getting lucky. Anthropics seed round
06:46open routers seed round 11 labs a seed round. I mean, these are all seed rounds I did. Did I
06:52get lucky
06:52three times? Yes. Does luck favor the prepared mind? Yes. Maybe we we know a thing or two of the
06:58next
06:58generation. And and the boomer generation needs to sit back a little bit and realize the right thing to do,
07:04the respectful thing to do is is help the next generation actually scale their impact instead
07:10of sitting back and critiquing all the next generation. So I would look, this might be rumors.
07:14I don't know. But I was told two days ago after we announced the Stripe open router acquisition
07:19that a partnership on Sandal Road, I won't name names, was analyzing my results. And some of the older
07:26guys said, oh, Anj got lucky three times. When I hear stuff and I found out about this because the
07:33younger associates at the firm called me and said, hey, they felt the duty to report it to me because
07:38you know, we're friends. And they said, we're a bit embarrassed by how the the older guys are
07:44carrying themselves. It's just not a good look. And I said, no problem. You can come over to AMP
07:48anytime. We're we're starting a new generation or a new generation partnership. And we will welcome
07:53anybody who does the hard work of reading the papers, understand the technology and investing in
08:00it. It's a very simple recipe. It's not hard. You just got to do the work. You can't shortcut the
08:03work.
08:04Anj, I'm worried you're about to start like a generational warfare right now with these comments.
08:09Point taken, though. I would say it started for it started five years ago, right with the
08:14anthropic seed round. Each one of those 22 introductions I made for Dario and Tom
08:18was a learning lesson for us that the reason we needed a reset in the generational investor base
08:26is because the older generation has stopped doing the work. And I don't know why it is.
08:30Maybe they've made too much money from their crypto earnings or the internet earnings or the Zerper.
08:35I'm not sure. But the rest of us can't wait around. And and I think, luckily, there is a set
08:41of
08:42investors in the previous generation like Marc Andreessen and Ben Horowitz who do want to give back.
08:46You know, they've supported AMP in a big way. And so I think it's an it's a it's a it's
08:50an IQ test for
08:51the older generation. Do they want to be part of the future or do they want to sit back and
08:56try to
08:56hold on, you know, and and not be part of an extraordinary new ecosystem at the frontier?
09:02Because I can tell we all talk the next generation. We're on group chats all the time talking to each
09:06other about what's going on and which of the older generation is supportive or not.
09:10And and I think the judgment day is coming pretty soon.
09:15Anj, I feel horrible because I wanted to ask you about open router. And now we've gone on on a
09:21tangent. Well, we should talk about open router. OK, literally 30 seconds. Can you just tell me
09:24quickly about the justification of buying open router for Stripe? Yes. Yes. Actually, we wrote a
09:30fairly technical paper on our website. People can go to amp public dot com slash research
09:35slash open router and they can read our paper. We published a white paper
09:39based on a hundred million token data set analysis that my colleague Malika and I did.
09:44It's a research paper that argues Stripe is acquiring open router for security reasons.
09:52Open router has one of the most extraordinary data sets on behavioral of transactions on AI.
09:59And if you need to secure the Internet of the agentic era, you want a cross behavioral cross model data
10:06set to figure out when these agents are going rogue. That's going to happen. That is happening at scale.
10:11And so Stripe was very strategic. A lot of people think Stripe is a billing company. It's a routing
10:16company. No, Stripe is a payments fraud company. They defend the Internet. And open router is one of
10:21the best security data sets out there.
Comments