OpenAI: A $300 Billion Non-Profit?

[00:00:00] Exciting week here at Newcomer. We just announced our Cerebral Valley London AI Summit. We are going across the pond. How are you feeling about that? Daunting? No, I think we're ready. We've got the venue. We've got a great speaker lineup. Very excited to have Dara Khazr-Shahi, the CEO of Uber. Oh my gosh, wait, you're the Uber guy. What's it like booking him into your own event?

[00:00:21] It's flashbacks. I covered the ins and outs of Uber, including the CEO search when it was, what, it was going to be Jeffrey Immold, it was going to be Meg Whitman. And then out of nowhere, it was sort of the dark horse, Dara Khazr-Shahi. And I've met with him a number of times. When Uber went public, I was there on the floor of the New York Stock Exchange. When Newcomer launched, Dara was kind enough to give me an interview.

[00:00:46] They've been pretty savvy, honestly, about independent media. I'm sure you saw that they did one of the inauguration parties with Barry Weiss at the Free Press. So they're definitely up on this sort of indie media. I've known them for a long time.

[00:01:00] It helps that they have a partnership and investment with Wave, the great self-driving car company of Europe. Alex Kendall is going to speak with Dara. And Uber, of course, needs to show how it's relevant in a sort of self-driving future with Tesla beating in strum very loudly and Waymo partnering with Uber, but showing, hey, we can do our own thing. So yeah, I'm definitely excited about that one.

[00:01:24] Yeah, that'll be a great hit. Also, I mean, I'm super pumped about all the, you know, like spicy, hot, early stage startups that everyone's talking about going to be on the lineup. We got the CEO of Granola coming, the hottest AI note taker.

[00:01:38] I know I've started using it though. I'm like not good at being chill about it. You know, I feel like part of the sales pitch with Granola is that you just like let it run and don't bother everybody. But I'm like, oh, I need to tell everybody I'm using that, which sort of complicates everything. And obviously we have Dylan Field at Figma, which we're equally excited about. Yeah. June 25th, CerebralValleySummit.com. Check it out. It's, you know, we want founders to apply. Basically, if you're a real insider, you should join us and apply or shoot us an email.

[00:02:08] This episode is presented by Brex, the financial stack founders can bank on. Brex knows cash is king for startups. So they build a banking experience that takes every dollar further. Unlike traditional business banking solutions, Brex has no minimums and no transaction fees, and you can get 20x the standard FDIC protection.

[00:02:28] Even better, you can earn industry-leading yield from your first dollar while still being able to access your funds anytime. Get the business account trusted by one in three U.S. startups at brex.com forward slash banking dash solutions. That's brex.com forward slash banking dash solutions.

[00:02:51] Jumping into our main story of the week, spoiler alert, it's not tariffs. We're probably the only podcast not talking about the tariffs right now. We cover private companies, not public stocks that are tanking across the country. I know, if anything, that's one of the great privileges of being a private company. You're like, we raised the money before, we think on longer time horizons, we don't have to sort of have the ebb and flow. Obviously, private markets will not be insulated, given that this is likely to be an ongoing problem.

[00:03:20] Of course. But for now, that's over in public markets and over in private markets. They're having an awesome week. OpenAI raised the largest venture-backed round of all time at $40 billion. Wow. We've grown so numb to it. It's always like, wait, this is the largest? You know, we did make a big deal of whiz when that was like, what, like the biggest M&A? And now we have the largest round?

[00:03:44] For all the downturn naysayers right now, this quarter has been ridiculous. Just to catch everyone up on the bullet points of the round, I know it's been out in the ether for a while, but OpenAI finally announced it this week. Here, kind of the rundown of the main stats you should know. It was a $40 billion total round, but it's kind of trunched, so it's a little funky, but it raises the company's valuation to $300 billion, which is wild for a private tech company. That's like one of the most valuable companies ever.

[00:04:11] We have SoftBank leading initially with a $7.5 billion investment and then $2.5 billion from an investor syndicate that includes Microsoft, KOTU, Altimeter, Thrive, people who've backed them before. And then the round is trunched, so the remaining $30 billion of capital, which is again divided between SoftBank and those investors, will come at the end of 2025 as long as OpenAI converts to a for-profit company. So it is dependent on that transition happening before the end of the year.

[00:04:39] Yeah, that's kind of staked funding there, but it's still a huge raise. And just for some reported stats behind how OpenAI has been performing to justify this crazy valuation, they have projected $3.4 billion in revenue, ARR around $4.4 billion. So the implied valuation multiple is around 68X, which is high, but not the craziest we've ever seen.

[00:05:05] They just have to do this one little challenging thing of figure out how to go from being a non-profit to a for-profit company. They don't figure that out. They don't get all the money. Hard to see how investors return all this money if they're investing in a non-profit. So there's a big flip. The big question on everyone's minds, of course, is OpenAI worth this? Is this just in la-la land right now?

[00:05:32] Or is it actually the most valuable private company of all time? Eric, you made the bear case a few months ago at the $157 billion valuation. Now we've basically doubled that. I know. I've been searching through the piece because I was like, I definitely caveated it. And I said, there's inevitably going to be more momentum investing in OpenAI to come. OpenAI is probably going to have to raise money again next year, even with its $4 billion revolving credit line.

[00:05:58] OpenAI could very well raise an upround based on the same vibes it raised money on this time, especially if investors continue to prioritize revenue growth over cash flows. So I'm prepared for people to mock this story in a year or less when OpenAI raises an upround. I should have foreseen SoftBank specifically. I should have been, yeah, this is October 24. I should have said, and SoftBank hasn't even sufficiently marked it up. And obviously, until they top ticket, the bears are not safe.

[00:06:27] I'm still holding on to bear case at $157 billion. I mean, obviously, it's easier to transact private market stocks. So there's an argument that I was just wrong because I should have invested, pulled a Thrive, invested $157 billion. I don't know that they're selling, but you could exit the position potentially in the next round. And so there are certainly people who are riding a momentum stock and there's plenty of money to be made there.

[00:06:50] But if we're talking on a sort of four-year time horizon or sort of long-term settled out public market price, yeah, yeah, I still think I'm in the bear case. I mean, if you review what I said, here are the bullets. Money losing machine. It's a company that loses a lot of money. Tenuous and expensive intelligence pole position. Man, we've talked about that a lot. We had the DeepSeek moment this year. Anthropa keeps getting better. There's serious competition. Google has one of the top models at the moment.

[00:07:18] It is a tenuous and expensive intelligence pole position. Talent exodus. Altman's limitations and OpenAI's weak alliances. I mean, I do think the Microsoft relationship continues to be a little muddled. I mean, Sam is depending more and more on Brad, the chief operating officer, and Sarah Fryer. It'll be interesting to see. I don't know as much about them, so I'm excited to see. I mean, I know Sarah Fryer. I don't know Brad. Brad Lightcap.

[00:07:43] He was running the investment fund out of OpenAI that was backing some early stage AI companies. I talked to him briefly about that one time. He seems legit in that fund. He's the chief operating officer, clearly an important guy. On top of all that, they are a non-profit company being sued by Elon Musk that has not made the transition to being a for-profit. And it's an insane risk factor. And also, Elon is now one of the most politically powerful people in the world.

[00:08:11] Obviously, Altman's working his own relationship with the Trump people. But I'm going to come in and make the bull case for this round. Yes, the price is crazy. I won't deny you there. But they have really, really upped their sales game. I mean, just the demand. There's already, I think they saw reports that people have already generated 700 million images just since they launched, like within the last few days, the image generator tool. Right? People are clearly using it. Madeline, usage means they spend more money.

[00:08:42] But yes, it's good to have customers want something you want to do. But, you know, you give something if it costs more to deliver it. Obviously, they will have to find some way to financialize and make some money off of all of their demand. But in the classic Silicon Valley form, you get the users first and then figure out how to make money later. Right? Like, that's how this always goes. And I think they've captured the market, especially in consumer. Like, there is no chatbot that comes close to them.

[00:09:11] Enterprise is a little bit rougher. Like, obviously, there's like, you know, more competition there. Anthropic is really popular, has a really popular API. It's a little bit more varied. DeepSeek, obviously, we've talked to death. But I do think in the consumer market, people love ChatGPT. That is the killer AI app. There really has not been anything that's come close to it. And so I think they have kind of like lightning in a bottle with their product. And they're like the main productizers. So I could see this being worth it at some point. Yeah.

[00:09:41] I mean, I agree with everything you're saying. You know, I've personally probably spent $300 on OpenAI, you know, because I paid $200 for their ultra premium version and then subscribe to the plus version. So they're clearly making money. I mean, they deliver value. And I do think, you know, there's no question. It's not like sort of Uber. You know, they never really proved out like Poole, for example.

[00:10:04] Feels like for sure they're going to be able to figure out a way to make Studio Ghibli photos in a profitable way. To me, it's more of the is that commoditized? Do sort of specialized apps win? And the final creative bear case that I will make, if OpenAI does too well, like say we are in AGI and it's still this sort of nonprofit, like are they really going to be able to like harvest everything?

[00:10:32] Or is it going to be like you were set up to be for the benefit of the world. Now you have the world controlling technology. Are they really going to be allowed to be the $2 trillion company that runs the only sort of non-human competitive intelligence? That being said, OpenAI, I think, is really going to work to make this for-profit conversion work. They'll lose half of this funding round if they don't, right? So they are working around the clock to make this. They want it. Definitely. There's no question about that.

[00:11:00] From the biggest funding round of the year to the juiciest espionage spy drama of the year. So I had a fun story this week. I'm going to give you the brief facts so that you remember where this all started with a Rippling lawsuit, right? So it's like Rippling, the company run by Parker Conrad, the sexiest, you know, payroll and company information, you know, like internal corporate management company you could have. It's so funny that this all like came down over just like specifically payroll. Well, that's what's funny about it.

[00:11:28] It's a very boring sector. You know, you got Rippling and Deal competing with each other. Deal is the company backed by Andrews and Horowitz. It's the firm that screwed over Parker Conrad and his old company, Zenefits. So you have all sorts of competing alliances. And Deal and Rippling, they hate each other clearly. So Rippling then drops this lawsuit and says, Deal has been spying on us. We had this employee. It was working for us and feeding information to Deal, which was like a pretty crazy explosive allegation.

[00:11:57] And then even crazier is basically Rippling says, we know this for sure because we set up a trap and we suggested the idea. We sent a letter to three Deal executives and like I think they're outside counsel and mentioned a particular term that we hadn't sort of mentioned anywhere else, a fake Slack channel.

[00:12:17] And then that mole, the spy, searched frantically for it in our Slack channels and obviously was like totally made up, but he wouldn't have been able to hear about it if not for sending this to the executives. And then we went to confront the guy. He's in the bathroom and they're like, hey, we've got like a court order and like a court representative here that says you need to hand over your cell phone. And the guy's basically like, I'm, you know, there was some interplay of like, I can't find it.

[00:12:47] But eventually it's like, I'm willing to take the risk and sort of fled the scene. It is wild that like, this is maybe the first case of like a Slack orchestrated sting operation I've ever heard of. Like, oh my God, the honeypot trap, the Slack channel, the sexiest spy thriller breaking down on Slack. And so, you know, this was already like a great story. We'd gotten far more than we ever deserve in the boring world of, you know, writing about sort of payroll startups.

[00:13:13] And then this week, this is all playing out, by the way, in Ireland. Like this employee is, I think, in Ireland. So the guy's name is Keith O'Brien. He is the alleged spy mole, sort of whatever we're calling him. And according to a sworn affidavit, he confessed like in court, basically. He was like, yes, this is all true. And not just is this true, but I was coordinating with the CEO of deal.

[00:13:42] Like literally we were like going back and forth. And also he was only getting paid $5,000 a month for this. So it's like $60,000 a year to like do what seems clearly like criminal, you know, just for like some small amount of money. The CEO of deal, Alex Bozes, likened what Keith was doing to mean like James Bond. And again, like if you're James Bond, you need more than $5,000 a month. Anything else stand out to you?

[00:14:12] I mean, I feel like, first of all, just glossing over that in the chat with Boisius, there's also his father, Philippe, deal's CFO. And this mysterious figure named The Watchman, not explained what's going on there. This is just like deep throat, but like boring. I don't know. Do you watch Billions? You ever watch Billions? I never watched Billions. There's a character in Billions who's like this sort of like hatchet man, basically. Is he The Watchman?

[00:14:40] Yeah, it feels like that sort of role. You have to imagine all these like, by the way, all these executives like watching like the villains on TV being like, oh, I guess that's how business is done. We need our like corporate spy, you know. Are we the baddies? Or more like that's the model. For sure. And I'm just still not over the fact that deals lawyer allegedly told O'Brien to destroy evidence where he smashed his phone with an axe and flushed down the toilet. Allegedly.

[00:15:07] Whenever you make allegations like that against a lawyer, you have to shout like allegedly a billion times. But yeah, it is a pretty strong allegation to say their lawyer was in any way involved because that's that's not good. You know, absolutely not. And of course, you know, allegedly, O'Brien has claimed in his confession that, you know, he was coached by deals legal team to shift the narrative here. And they were they were like suggesting he like potentially like flee to the Middle East. I'm sincerely excited to see deals response.

[00:15:38] I know deals been quiet this whole time. This is a wild case of accusations that they have not yet commented on. Most of this, again, of course, is coming from this confessional and the rippling lawsuit. Right. So what are they going to say about this? But also, how do you come back from this? Like even if like this is some degree of it's proven untrue, like where did you land in the court of public opinion after all of this as a company? I can't claim to totally understand Keith O'Brien and all this.

[00:16:03] I'm sure there will be reporting at some point being like, how does this guy totally flip? Fascinating. Also this week, Politico reported that Elon is exiting the Trump White House in the coming weeks. Oh, my gosh. The number two guy moving on out. Define coming weeks. Nobody else is saying this. Is the coming weeks sooner than late May or early June when Elon's special advisor status is supposed to end and he sort of signaled that maybe I'll wrap it up by then.

[00:16:33] The administration is not flat out denying this story, but they've reported back to that, you know, signifier initially that was like, this is always going to end. So it's not like a full denial, but that's kind of the thing. So maybe it's just like they're saying he's not extending upon this. But I guess the main thing was there was like ruffles around, you know, how well Elon and Trump are really getting along, which was sort of the intrigue they're going for. I also wonder if tariffs like are good for Elon because it's like this this one isn't his problem.

[00:17:01] And like he was generating a ton of negative press with like Doge and just like, you know, they're destroying random things. You could also make the case that he like lost that Wisconsin race by getting right. Oh, my God. Exactly. Right. Yeah, exactly. Not only did he lose it, like fine, they were going to lose it. It's classic politics that you don't say I'm staking my whole reputation and the whole world hangs on this race. If you can't probably pull it out, you know what I mean? It's just like why?

[00:17:28] Why raise the stakes of some random Wisconsin Supreme Court case? Like I understand it has implications for like members of Congress and like it's important. But why put your whole reputation behind it if you know anything about politics? It was a little bizarre. For sure. So anyway, that's not looking so hot for him. And they're saying he could, you know, stay on as some kind of advisor in some capacity. Still, it's not like he's going completely away and leaving D.C. I can't imagine that's going to happen unless he has like a major Trump ego clash.

[00:17:58] But that hasn't happened yet. I'm not sure that that's going to happen, given that they've done so well together this far. Well being a very relative term, given everything we just discussed. But in terms of the rumors about, you know, the big personality billionaires eventually just going to flame out like that has not borne out yet. So I wanted to bring up what Elon's doing here is what he did at Twitter, right? People talk about he's like running the Twitter playbook in government.

[00:18:25] And so the Twitter playbook, the similarities are you come in, you rip everything out and some things break and then you put those back in. But if nothing really breaks, you're like, oh, that wasn't that important. We just saved a whole lot of money. We got rid of it. It's at least a way to cut costs. It doesn't necessarily drive revenue. And I don't think the federal government is going to have a hype-y foundation model company that we can merge with to save everything.

[00:18:53] But putting even that aside, if we're just running the Twitter cost-cutting playbook, a key part of it is that you sort of like get the feedback from the world and you feel like, oh, we caught Social Security. You did that. People get really upset, you know. And I feel like government and society is slower and bigger than Twitter, right? It's not like looking at code did the app break or like waiting until Disney like calls you shouting about something.

[00:19:19] And so like I don't see how Elon can expect to run even in the most charitable sense this Twitter playbook and exit by late May or early June because there's no way he's going to have had the feedback to know like, oh, we should put back the pieces. And he hasn't cut the budget. You know, as the other – besides, he doesn't know if he should put anything back and he hasn't even achieved the savings he's supposed to. We should remember the Doge, you know, savings website continually being debunked.

[00:19:48] So the savings aren't even what they're saying they're doing. Are you making the point, Eric, here that government and businesses are run differently and you don't actually run the government like a business? They're different. Right. The metaphor isn't very good, perhaps. Yes. It does feel like that whole argument is basically like if USAID was so important, Elon and Trump personally would hear people psychologically like, you know, making – would be so upset.

[00:20:15] It's like, oh, I'm such an elite disconnected from people who are hurt by my decisions that I don't feel it and therefore that is proof that it has no impact is like exactly the problem. At the end of the day, we are going to be able to judge their success because they're not very transparent about all the details on just like, did the deficit go up or did it go down? And I think it's clear that the deficit is going to go up. Yeah. If you're combining this with tax cuts, yeah.

[00:20:41] And if you're combining it with cutting revenue, you know, like IRS agents and like other ways to make things in and if you're sort of hurting the economy by putting a bunch of people out of jobs. Like it's not – tax cuts are a part of it but it's also like the things that they're doing with Doge are going to hurt our ability to collect revenue, have people employed, have a strong economy that produces taxes.

[00:21:07] I want Elon to stick around because I think he should finish the job, like prove that he can do it. Not just say, oh, I tried. I hit the deadline. I got to get out of here. You know, it's like do the thing you said you were going to do if it's so great, you know. I mean I'm still in the like run the country your way. I mean we're getting it right now with tariffs. Notably a lot of the VCs that are, you know, also involved in these efforts are not part of this exiting plan.

[00:21:30] So I can assume, right, like Shriram and some other people may stick around, the A16Z people doesn't seem like they're leaving anytime soon. Scott Cooper, you know, who is a top guy to Andrewson Horowitz, I still think he hasn't even been confirmed yet. And I like him. We can bring this episode full circle honestly. So Parker Conrad, the CEO of Rippling, the one going after Deal, was also in my piece after the election saying like, this seems like a disaster but like a godspeed. I wish you the best for the country.

[00:21:59] But Parker to me was like, you know, they're probably going to get their faces ripped off like everybody does with Trump. And he specifically made the point, everybody thinks Trump is going to do their crazy thing but then he does, you know, some other crazy thing instead. So anyway, Parker tweeted today, I hope we don't forget Mark Andreessen's grand proclamation about the economic coiled spring that would be unleashed with the Trump presidency. It's like, yeah, exactly. Where is it? That's our episode. Stick around.

[00:22:26] I have a conversation with the CEO of Clay, the top company on our mid-stage Enterprise 30 list with Wing and a fun conversation digging into his go-to-market AI company. Give it a listen. I'm here with Kareem Amin, the CEO of Clay, number one mid-stage company on the Enterprise 30 list that we did with Wing.

[00:22:51] And so this is our second video after talking to Paul Klein, the browser-based CEO, digging into companies that, you know, are now thrust on my radar. Where I'm like, oh, I want to know everything about what you guys are doing. Kareem and I were on a panel the other day at NASDAQ celebrating the Wing 30 list and happy to have you on the podcast. Hey, Kareem. Yeah, thanks for having me, Eric. Great. Well, yeah.

[00:23:18] So I think, you know, we sort of want to get to like, how did you get to number one on the mid-stage list and then where you're going? You know, the best place to start is how was Clay founded? And I think, you know, you guys have been at this a while. Yeah. Tell me about the early days. Yeah. So we started over maybe just seven years ago. The motivating idea was how do we give the power of programming to an order of magnitude more people?

[00:23:45] So we were very much in this continuation of the idea of how do you make programming accessible? And the reason we wanted to do that is that we knew that if you could essentially have that power, you would be able to level up kind of like the work, the type of work that you're doing and just have access to capturing a lot of more of that value. And we were anticipating that there will be more and more kind of like AI set of features that would automate work.

[00:24:14] Obviously, we were not kind of like seeing that LLMs would be the way to do that. Was Clay Filmed? 2017. Yeah. And who is the first customer type? Great question. I think when you look back at it, yeah, you do need to have a clear customer and you need to have all of like who's your TAM and all of that stuff. But I don't think that most people, I think that's just one way of like articulating your idea.

[00:24:41] Like we started off from what is it that we want to have in the world? And then we thought, okay, well, if we want to give the power of programming to more people, do we want to give it to engineers and make them a lot faster? Or do we want to give it to non-engineers? And we started kind of exploring to see like what is possible. At that time, there was already things called programming by example. So in Excel, I haven't talked about this actually before, but you could go in and let's say that you had like a phone number and some of it was missing the brackets around the first three digits.

[00:25:10] You could add it and you show kind of Excel like, okay, you add it once and it can add it to all of them. But sometimes it would add a double bracket. So you go delete that one. And now it learns, okay, if there's a double bracket, don't do that. And so we started to say, okay, there's going to be more and more ways to program by example, to program by showing the machine what you want. And we were kind of exploring like what is it, what's possible today? We eventually decided that actually it's much, much more powerful to give it to non-engineers.

[00:25:38] And we started thinking about like, what are the metaphors that people understand and decided, okay, the spreadsheet's the world's most popular programming environment. And Google made it collaborative. So many people could work on it at the same time. But the next step was to make it connected to all of the internet's data. And so that's kind of like how we came to Clay. Because once we did that, we were like, well, who is going to use this? And there could be financial analysts or there could be people building lists.

[00:26:05] When we looked at people building lists, it was recruiters, salespeople, marketers, customer success people. And so we decided to focus it on go-to-market teams. So I think at the end of the day, you need to focus it on who you're selling it to, to get the product going, to build the business. But it's ultimately in service of this bigger goal. And so just the core customer action so that people just understand, like, are we going to use the product the way that most people use it? First off is what?

[00:26:32] The core customer action is you want to grow your business, right? And to grow your business, you need to build a list of potential new customers or existing customers that you want to reach out to. And you need to figure out what properties predict that someone's going to be a good customer. So Clay allows you to aggregate all the data sources on the internet to build the list of potential customers that you want to message. Using those data points, you can figure out what you're going to say to them.

[00:27:01] So let's say you find all the people on the internet who don't have SOC 2 compliance but are SaaS companies that have been built. And you sell a SOC 2 compliance product. And that's a perfect moment for you to reach out to them. So was that possible pre-LLM? Or how much of the product really existed before sort of the model explosion? Yeah, it was totally possible pre-LLMs. There were all these different data providers that scraped data from the web and packaged it.

[00:27:26] What was difficult, so that's what we did first, was say, okay, like why use one data provider when you can use all of them? And it depends on what you're looking for, which data you need, and makes it easier for you to work with that data to build the list. What LLMs made it possible to do was to then also get unique data properties that no data provider kind of had. So let's say you want to say, I want to find all companies that have remote offices in South America, right?

[00:27:53] And maybe that information you could kind of extract from existing data providers, but it would take a lot of work. You would have to say, okay, do they have an office in South America and an office outside? And where is the headquarters? And that means that they have a remote office in South America. So it extracted a lot more value out of these data sets that you'd gotten together because there's logic and sort of pulling out pieces. Yeah. And you could do more real-time things, right? So you could say, check if this company has the logos of some of our current existing customers, right?

[00:28:23] So you can do things that maybe you'd have humans SDRs normally do. Do customers pay to use all the different data sources or you just license it yourself and it's abstracted away from them? Or how much interaction do they have with? Yeah, we have an interesting model here. So basically you can use most of the data sources using Clay credits, right? Because we have licenses with them. If you already have a contract or agreement with a data provider, you could just use your API key in Clay and it's free to do so.

[00:28:50] So our idea is very much a collaboration with data providers. We focus on a segment of the market that they don't normally have the resources to sell to. And some of the data providers are smaller and they have really high quality data, but they just don't have the full UI or go-to-market team to sell it. And the other thing is once a customer grows big enough, we prefer that they actually have a connection to the data provider and they have a direct kind of like contract and API key with them. Our belief is that, yeah, we don't sell. We don't necessarily make the money off of selling the data.

[00:29:20] It's how you make use of the data to create a campaign that we think is like where we provide value. Yeah. And then on sort of the content generation side and sort of the outreach scheduling, how much are humans in the loop? Or like how much is it sort of still a human firing off emails versus setting all these triggers? Or how do you think about that? The way we think about it is humans are there to come up with the ideas, like which triggers, when, how often, to whom. We just facilitate doing all of that work.

[00:29:49] So the way to think about Clay is like if Figma is for designers, Clay is for go-to-market teams. You're still involved, but it's a creative tool that allows you to express your idea quickly. We don't necessarily believe that there is one solution that fits all or that there's like one magic button that you should turn on and let it go. You have to constantly be iterating on who are you? What are the guesses that you think makes a good customer and what predicts that? What should you be saying to them? Your products are changing.

[00:30:17] Their products and their needs are changing. So you need to stay up to date. And you also need to stand out from the crowd. Does this only work if it's like, oh, I can sort of A, B test my customers and there's a huge universe? Like if I have a business where I'm sort of like whale hunting is AI is useful, where it makes sense to focus a lot of energy on in particular. Right, because it's not just for outbound, right? Clay is letting you build, let's say, a deep research dossier about this customer, right?

[00:30:43] You might want to figure out, you know, who are all the, how many people have they hired in a particular department? What is their organizational structure? What have they been publishing lately? You can look at their financial statements of their public. So what are their actions online? What are they doing on social? You can like do social listening and you don't have to send them a message. You could send all of that information to your own team. The other thing is that, you know, I think of outbound as similar to advertising. So Coca-Cola still advertises.

[00:31:12] You need to have it on your mind. So just you need multiple touch points for someone to know who you are and why you matter. So an even example, I think in Clay isn't just to be used to find new customers. It's also to expand your existing customers. So imagine that you're Figma, you release this new like product dev mode, and maybe you reach out to all of your customers who have a lot of designers, but, you know, they, none of the engineers have onboarded and you figure out who their head of engineering is.

[00:31:40] And you message them and say, hey, your designers are building a lot in Figma. Should your developers be using it too? Do you send the actual email through Clay? And so we basically connect to a bunch of third parties to send the email, but we have an interface that allows you to draft it and figure out what you're doing. And the reason we, we stay kind of separated from the channels is that people do all kinds of things with Clay. So there are companies that send physical mail with Clay that send gifts, personalized gifts that have created personalized websites.

[00:32:08] So you can, people have created personalized ads. So we're kind of agnostic about where the end channel is. You've obviously been intentional about the product besides sort of coming up with a great feature set. Like what's it, what explains your like explosive growth at the moment or what, what would you point to that sort of like really clicking for Clay and why you're having this moment where investors are so excited about what you're doing?

[00:32:33] I think investors are excited because they see kind of the potential that is made possible through Clay. So we, you know, if you're, every company needs to grow, right? And there are only three ways to grow. You either find new customers, you convert existing ones, or you expand existing ones. Clay helps with all three, right?

[00:32:53] But we found a particular kind of gap in the market, which is that most schools in go-to-market want to sell you a solution that helps you automatically grow your company. It's sold to reps or people who are on the front line and lets them kind of pull a list quickly and send some messages. Whereas what Clay does is it assumes that there is a new type of role, rev ops, growth marketing. We call them go-to-market engineers.

[00:33:22] And we think that AI has actually enabled this new role that's going to be really well-paid and really expansive. That is a more technical person that can take on that job and can 10X the whole team. And that, and you don't need to A-B test things, right? It doesn't have to be that scientific. It's mostly like you have an idea, we can turn it into reality quickly. And so we're a very, I think we're the first creative tool in the go-to-market space that actually allows you to speed up any idea to reality.

[00:33:51] And you can apply that for sales or for marketing or customer success. So the TAM is huge and people are making a lot of money using Clay and also getting promoted using Clay. So that's why customers like it. If you get this tool, you're going to, yeah, get a promotion. On the flip side, is it hard to get a company to embrace a new tool when you say, oh, you need really to rethink what employees you have if you have Clay?

[00:34:19] Or like, isn't that sort of, normally it's like, we want to meet you where you are so that you're excited. We meet you where you are, right? Like the role already exists. It's called RevOps. I think it's just evolving. Also, advanced AEs or SDRs can pick it up. So it's basically, you start off where you are and then you can essentially parlay it into this job once you prove that it's working.

[00:34:43] So we tend to kind of meet people where they are and then they quickly find that they're getting results and then it transforms into, and don't forget that this is a particular moment where people are willing to experiment because they're like, what is our go-to-market strategy with AI? And the answer is you should be using Clay. What can you say about number of customers, size of the business, employees, whatever snapshot you can give? Right.

[00:35:05] So the last public numbers that we gave out were in the Forbes article when we announced our raise at $1.25 billion and we had 5,000 customers and over 30 million ARR. And we've grown substantially since then. The jobs question. I mean, do you have an intuition on whether AI is going to reduce the number of jobs or from your sector?

[00:35:30] Like how, I mean, if you sort of, you know, 10Xing or more the capacity of an employee, does that sort of imply fewer employees? Or how do you think about that? You know, my current sense is very much embodied in like things that are like previous technologies that we've encountered. So, you know, when you build more roads, you just get more cars. I think that our system is set up to, regardless of the efficiency gains, to do more. And so let's say we get more efficiency gains. Well, we're just going to want to try to grow faster.

[00:35:59] We're not going to say, okay, great. Now we can grow as fast as we did with fewer people. I think some people will do that. And hence, you have a lot of people coming out saying, oh, we're going to build like, you know, a 10-person company with like 10 million per employee. And maybe you're satisfied with that. But I think there's going to always be competition at the top for markets where winner takes all, where you will try to do more. So I don't think that you will have fewer people.

[00:36:25] I think you will have more productive people in certain elements of the business. And you will find new areas where you want, where things weren't doable before, weren't worthwhile. The other thing is, I think a lot of the current predictions are predicated on LLMs becoming one to two orders of magnitude better. And I think that that is different, right? If it becomes two orders of magnitude better, I do think many of the jobs will become less needed. But I'm not totally sure.

[00:36:54] I don't have an intuition on where we are towards that. Do you spend a lot of time as the CEO of a very forward-thinking AI application company gaming out like AGI strategy? Or so much of the business is like new model comes out, figure out how to make use of it. And we react when they come. Or like how much is there sort of like war gaming going on inside Clay about what the capacity is going to be?

[00:37:16] I think, you know, what I've told the team in general is 95% of your time should be executing, 10% meta-reflecting on anything. Otherwise, you're wasting your time. And so we're focused on delivering with what we have now. There's already a lot. And 10% of the time we think about the future and what are the possible kind of options. But regardless of what the options are, we will be in a better place the more successful we are right now. Right. Yeah. To make a decision.

[00:37:45] How do you think about, you know, I, so I use Superhuman a lot. Yeah. Different products interested in using Clay. Every sales product or, you know, could be a reporter tool. And obviously, you know, we have customers too now. But, you know, Superhuman or email providers, it can feel a little bit like you're fueling a war on both sides. Where it's like you're making it easier to email people, but you're also making it easier to like then filter out sort of the emails. How do you think about it with your product?

[00:38:14] I mean, you're sort of, you're more focused on arming sort of one side of the equation. But do you think, oh, there's just going to be another AI company on the other end that's making it, filtering out all this like emails. That's what I'm going to receive. Yeah. But I think we're, so yeah, I mean, I love Superhuman and Rahul. I think they've done a great job. I think that, you know, have you ever heard of the idea of the Red Queen? I think there's a. I have. It's been a while. I feel like it came up in the Uber context. Right. I think, I think the author was Matt Ridley.

[00:38:42] The idea is that, you know, the Red Queen keeps running and stays in the same place. Right. So, and I think it was about viruses and like human genomes and like you're constantly changing and meeting it. So there's always this cat and mouse game. I think it just depends on how you approach it. Right. So we are not thinking about how do we give people the power to send more messages. Quite the opposite. We think that you should be doing deeper research ahead of time, send fewer messages, do it over time.

[00:39:12] Right. You don't want to blast your whole TAM or all of your customers. You don't even know if it's going to work. So you should pick smaller segments that are deeply researched, spend the time deeply researching them and then send a more worthwhile message. We just help you do that at a scale where it becomes feasible. And we give you the tools to do that. If you, that's why we don't own kind of like the messaging channel, because we don't know what you're going to send. And we're not responsible for that.

[00:39:38] If your product sucks, your message won't work regardless of who you target and what you say. So that's also information. Right. And that information is helpful. I think it doesn't make sense to claim that we will help you grow regardless, because what you're selling might not be worth growing. If there are good customers out there that fit, we'll help you find them, but we can't promise. Right. Yeah, we'll help you figure out how to go about doing that.

[00:40:03] And so I think we're much more aligned, actually, with customers and people sending messages that are worthwhile. And we try to guide people to use us in this way, because by the way, if you don't use us correctly, you're the one who's going to get banned, not us. Get banned just by like email providers, if you're sending too many messages. By email providers. Yeah, exactly. Or like by, you know, ad messages or you're the one who's not going to get the results. Right. So, yeah, I think that's a key element of it.

[00:40:30] Can you walk me through just like one of the most interesting customer case studies? Yeah, for sure. Sure. There's a couple, I won't name the names right now, but some of the case studies are on our website. But I want to tell you a couple of them that we don't have on the website that I think are great. One company actually sent us a message where they started looking at our status page and seeing whenever our product was down or had issues.

[00:40:55] And then they were a DevOps tool and they would see what we said was the issue. And they'd be like, well, if you were using us, you wouldn't be down. And here's why. And it was very effective. I read it. And so I thought that was a pretty creative way. There are no data sets that look at status pages, but they used our AI agent to monitor us. There's another company that is looking at... To double click on that, like they built a tool or like how...

[00:41:23] No, so they used Clay as AI agent to look at our own status page and monitor it to see whenever we said that we were down, looked at the reason that we were down, matched it to one of the products that they could sell to us and then sent us a message. So you have a tool that sort of triggers when websites are down in real time or... No, that's the beautiful thing. We have an AI agent that you can program to do whatever you want, including look at our status page. Interesting. Right.

[00:41:50] So what's great is that we don't need to have a tool for every possible use case. Actually, customers discover new use cases. But are they then... They're hosting a tool they built on Clay or that becomes a tool that all Clay customers can use? No, no, no, no. It's just they told the AI agent to go look at our status page and then message us when there's an issue. And tie it to their products. We can then take that use case and make it more templated for everybody else. Yeah.

[00:42:19] And, you know, that's something that we can choose to do. But we don't have a marketplace where you can then sell that strategy. Oh, I see. Yeah, yeah, yeah. Got it. That's a super interesting one. Yeah. There's another one that I think is pretty cool where a company has a way of kind of doing brand normalization. So it's just like, is your brand the same in every channel that you're using it?

[00:42:45] So they use Clay to look at like a company's brand identity. Then they look at all the social channels that they send messages through. And whenever they see a discrepancy, they message the company and say, hey, you just tweeted out something that doesn't match your brand identity. You could have been using our tool to have brand consistency. So these are very unique go-to-market strategies.

[00:43:08] And what's amazing about this is it's very custom to your company, but the tools that you need to do that are not custom. They're the, you know, Clay primitives. And that's obviously why we're called Clay, because you can build different things with it. Right. The same way that every website, you know, could look completely different, but you can still build it in Figma. Do you allow customers, like how much, this is going to reveal how non-technical I am, but like direct like API access.

[00:43:36] Or like, you know, I feel like one trend in the AI world is like, we're only an API business and you can sort of flow through us without really interacting with our interface. It sounds like what you're describing, though, is the opposite, where it's like we have a great interface. You'd want to go through Clay. Or how do you think about the rise of, you know, these sort of API first businesses? Right now, we don't have a public API for you to do that.

[00:44:02] Sometimes we work with companies to enable that, but I think it depends who the customer is, right? Is the customer an engineer or is the customer a non-engineer? And so we believe that the more features we make in our UI, the faster you can build some of these things without needing an engineer. And then there's obviously also room for engineers to use some of the capabilities that Clay has via API to build things themselves. That's something that we'll be looking at in the future.

[00:44:30] I want to ask you a couple questions about sort of AI applications broadly and where things are headed. Any, I guess, last words on Clay or like, you know, where you'd like to be in sort of two years or any other reflections that I haven't covered on your business before I ask you to play pundit for a second? I think that the thing that's so exciting about Clay is that we have this really large community. So there's over 20,000 people in our Slack channel.

[00:44:59] There's clubs in over 40 cities. We just opened out a funding round, a community funding round that was basically allowed our own community to invest because we're getting ideas from the community so that they should share our ability to grow.

[00:45:17] But I think we've created this new go-to-market engineer role that is a really like powerful persona where you're able to kind of control all of like the third party data that you can find on the internet about a company and then your own first party data.

[00:45:34] So like, let's say someone's been using your product a lot, but stopped using it recently, but would use it if you released a new AI feature, you should message them or has been complaining in gong calls about your support and now you've improved your support. So this go-to-market engineer can turn all of these ideas for growth into reality. And I think those people are getting promoted really quickly, starting agencies as well to help other companies making a lot of money. And so in two years, we want this to be a role that's in every company.

[00:46:04] Each one of these companies is using clay to grow. The idea of like a service provider who's using your tool to then sell it to other companies. I mean, it's funny because you're also seeing AI businesses, you know, in the legal space, there's sort of an ongoing debate of whether you should be the service business that uses your secret AI tools or sell the AI tool.

[00:46:26] Yeah. I don't know. Do you worry at all like, oh man, if you can build a company to sort of consult just using clay, we should be sort of selling those consulting services. Or how do you think about that ecosystem? No, we're very much wanting to partner with that ecosystem. And we have a very large number of people making a lot of money building on top of clay. It's a small percentage of our revenue. Most of our companies use it directly.

[00:46:50] But if you look at HubSpot, for example, about half of their ARR comes from basically revenue that a consultant has touched. And so you can build a very big business doing that. Right. Yeah. We almost signed up for HubSpot and it's like, oh, we'll give you this person to architect it all. It's like, oh my God, I'm going to be spending this money forever. I'm never going to get out of it. That's exactly what they want. Right. I walked away at the last second. But I know I said I was going to move on from clay, but it's super interesting. I mean, how much are you learning from Figma?

[00:47:19] I mean, I hear like community centric. Are you allowing employees to start using clay before the company starts paying for it? Or what is sort of the adoption sort of journey for most companies? No, we have a reverse trial. So you get all the features for two weeks and then you have to pay. Right. Yeah. And so we have a free tier, but we are really selling to people who have a product that they want to grow. So our profile is a little bit different than like... It's not the rogue employee.

[00:47:47] It's sort of like, oh no, this company is sort of deciding to consider using it. Yes. Yeah. Yeah. Yeah. And that's a better fit for us. Yeah. Because it's... How come? Just more intense or it's like... Well, it's just more that you need to have something to grow before you can grow it. Right. Whereas some other products, like if you're in Figma, you might draw something. It doesn't really matter if the thing exists or not. And you can still be playing around there. So it makes a bit more sense. And we're closer... We're more closely tied to revenue.

[00:48:16] Once you start making revenue, we can help you grow that. Right. Yeah. Which is why investors are excited. It's like, we're close to the money. You can see that you're making more of it. You can see the effect. Yeah. Yes. For sure. And don't forget that the biggest SaaS company in the world, Salesforce, is basically a go-to-market kind of like toolkit. It's ostensibly a CRM, but really it's everything that you need to have the information so that you can grow your company. Do you have the CRM piece of it too? No.

[00:48:44] I don't think that that's kind of where we want to be. We are a system of action, not a system of record. I like that. Yeah. Yeah. What lessons do you think other startup founders can learn from Clay in terms of building applications, whether it's the relationship with model companies that you've taken or anything else that you think is more broadly applicable that you can give away in other categories?

[00:49:08] I think that the key thing, without knowing a specific situation, the core ideas, I think, are you have to have alignment between the various parts of your business. So like, what is your product, right? And how does that align with your go-to-market motion? How do they feed into one another, right? So where do you start? And then how do you take it to the next step? Those things need to make sense, right? So if you are, I'll give you an example, right?

[00:49:35] We knew that we wanted to have a PLG driven business. So like people can come and sign up for the product. They don't need to have a demo. So we built the product so that you can use it without needing a demo. And then we figured out who has the most acute problem that is similar to the one we want to solve. And it turns out that there's all these B2B agencies that are helping people grow their companies, but they are at the forefront actually, because they're helping so many different companies. So they have the most complicated set of problems.

[00:50:05] And if we solve theirs, they would go and talk to other people about us. And so we knew that we could help them. And then we can also do the same thing and go directly to companies. The use case is not different. The functionality isn't different. The feature set isn't different. If you're going to help someone and then jump to another set of customers and the functionality is different, that's too far of a leap. It needs to be kind of concentric circles that are overlapping.

[00:50:31] You want your successes to sell your product to somebody else without having to build something totally new. Exactly. And then you need to have the sales motion to be aligned. You can't start PLG and then suddenly be like, well, everything needs to be sales-based in the future because you built a different culture. And you also have to have a unique insight, usually by turning something on its head. So you could build the simplest tool in the world, or you could build a more complicated tool. You need to figure out where are you on that spectrum.

[00:50:59] And you could build a tool for everyone to use, right? Or one person to use. And those assumptions change what you build. What do you make of this idea that software companies are screwed? Or it's like, if I'm going to build... I mean, you talk about Salesforce. That was sort of the situation with Klarna. It's like, I'm going to build my own Salesforce, so I won't have to pay them. Is it just you have to sort of outrun the capacity of somebody to build it internally?

[00:51:28] Or you think that's sort of overhyped? Or what do you think about this idea that every enterprise just builds any application that becomes core to what it's doing? Yeah, I mean, that was already happening, right? And just depending on how much... How willing you are to spend money on this and how willing you are to kind of set up a team to do this. So a lot of people had clay-like things internally, the same way that people had built CRMs internally before Salesforce. So there's always a build versus buy conversation in software.

[00:51:55] Whether it skews more to build now because it's cheaper. I think the question is, it's not about building most of the time. It's about maintaining. Everybody knows this, right? That's the total cost of ownership, right? Who is going to maintain this? Who understands how this thing works? Who already is knowledgeable? Does this look like something else they've seen in the future? Are they starting from scratch? And I personally like the idea of personalized software, right?

[00:52:22] That you can build something quickly, some internal tool that you need. I do think that for tasks that you're regularly doing and that look similar across every business, that you will still have software off the shelf that is just built to be very configurable very quickly. So that has tons of integrations that is easy to manipulate or mold right into whatever you want. So I think that software that is very easy to customize will still be in high demand. Yeah.

[00:52:52] And I think that you're not going to build from scratch. Even if you can build it, the requirements gathering, knowing what to build is still a lot of the work. It's most of the work, by the way. Right. Yeah. Yeah. And so you're going to have to figure that out. Your data relationships give you sort of a pretty big moat if a customer is using a lot of them and wants this sort of sense that I can pull from a lot of different sets. Even if they're not using a lot of them.

[00:53:21] What happens if I'm mostly just using like one of them all the time? Like, do you worry about that? That's not possible. Right. Yeah. I'm not super worried about that because everybody has tried to build the one data set that rules them all, but you're working with living entities, people and companies, and there's so many different ways to view that. So there was so many different properties and you can't have them all. And it's also very different, right? Like you might have data about companies in North America versus companies in Europe. So that hasn't turned out to be the case.

[00:53:47] There's definitely a moat in all of the data provider relationships and data cleaning and all that. But I think a lot of the moat is in the workflows, right? It enabling you to build any kind of workflow quickly, no matter what it is. And then also seeing what are the most interesting workflows and enabling that for everybody else. Right. How much do you adjudicate which data source is more accurate or score it? Or you're like, oh, if you're depending on phone numbers, like we think this is the best

[00:54:14] source versus leaving it up to the customer to sort of pick between multiple lists. We do a lot of due diligence before kind of working with a data provider. And we have a team that looks into that and compares kind of like, you know, whatever the data is in a particular, you know, set of functionality like technographics or demographics. But we leave the choice to the customers.

[00:54:38] In the end, we'll be doing more and more of that as we figure out ways for us to be really accurate so that we can actually give really high quality information here. Right. You know, I love the idea of like truth in AI. And obviously with sort of existing companies, there are actual facts. And so you can sort of decide over time, like what's real and what's translating into the world as it exists. Do you use coding apps internally?

[00:55:06] Or I'm curious what you guys are obviously super sophisticated. There are lots of them running at each other right now, like Codium, Cursor, Lovable, you know, there's Replit. There's obviously Claude Code itself. It goes straight to the source. Like what are you guys using and what's your read on how much that's helping Clay build quickly? A mixture of all of the above. I mean, I think people internally use Cursor. Some people use Windsurf. We don't have a unified kind of ideology. Go forth. Use what you want.

[00:55:36] I think it's too early to kind of decide. And I think actually letting engineers decide what is best for them. You know, if you're using Bolt or Replit to like whip out, product managers could use this to show like a quick prototype versus if you're using Cursor as your IDE, that's a very different thing. Right. Or some people use GitHub Copilot. We're still finding that a lot of like AI code requires, you know, heavy human review. And you also need to know what you're doing so that you understand what you actually implemented.

[00:56:05] Otherwise, we'll have there are like very subtle bugs that you can introduce. You know, my friend Meryl runs a company, Graphite.dev, that does AI agents like code review. Pretty cool. Yeah. So there'll be AI agents doing the review and then you'll review it. It is moving, helping us to move faster for sure, especially if you're writing a lot of boilerplate.

[00:56:27] I do think that the barrier is still the ideas, what you're, how they all work together, coordinating all of these things and then being able to bring them to market through a unique kind of go to market channel. So we use clay internally as well to figure out like what are the tactics we're using to sell clay. And I think that you don't want to just build a bunch of low quality things. Congrats, you know, a mid-stage winner and we'll be following everything you're doing.

[00:56:56] And thanks for coming on the podcast. Appreciate it, Eric. Thank you. Cool.