E221: OpenAI $500b tender valuation; Cohere raises at $6b valuation; n8n raises at $2.3b pre-money

0:00Clint Sorenson:Okay, Nick, my man, OpenAI, big tender announcement this past week at a$500 billion valuation, right? So that's up 66 % from the$300 billion valuation that they announced earlier this year. So that's quite an uptick. And this is a tender round. So it looks like$1.5 billion for existing shareholders in this tender. Thrive Capital is going to participate. So that's probably money good. like thrive is in it. Those guys, those guys do big deals. So that's kind of on point. ARR is estimated at 20 billion by December, 2025. So that's a, if they pull through and they deliver on that, that's a 25 X I'll call forward revenue multiple, which isn't crazy.

0:46Clint Sorenson:Right. But what do you think about open AI? It's a big race now. Who's going to hit a trillion first? Is it open AI? Is it SpaceX? Like, what do you think, man? Like, what do you think about this tender offer, the$500 billion valuation? Sure. So I think, yeah, OpenAI could hit that trillion dollar valuation first. It's just, there's a lot more volatility at play. And the fact that this is a tender offer is bringing a different dynamic. So tender offer for anyone that's kind of hopping through this podcast late is really bringing money towards the existing shareholders for an exit or to transfer those shares or sell their shares into a new buyer, which is Thrive in this case, along with a few others.

1:32And, you know, that changes the dynamic a little bit because it's less money going to the company, more to, you know, make everybody happy that's been an employee and put in the blood, sweat and tears into growing a rather dynamic company, to say the least. So it'll be interesting to see if anybody departs after this, You know, they get a bit of money in their back pocket. Maybe they want to go on to something new. I know we're talking about Coher later, you know, something a little bit different than OpenAI. So I actually think that's a super attractive valuation for any existing shareholder to take an exit if they want, or at least pare back and take a bit of money off the table.

2:12And it's also not an unreasonable entry point for someone like a Thrive, you know, that really has a long dated interest in where this company can go in the big entries. So yeah, I think it's a bit of a win-win situation here with that valuation, which is still, like you said, it's remarkably higher than the last round they did. And somewhere in between, really, where that last round was, then the secondary. And then if you put a layer on top of that, you're a little bit closer to that 400, 450. So it's not a big reach. So I'm actually curious if the secondary market will snap up to that in the next two months.

2:52Clint Sorenson:Yeah. I mean, I'm right there with you. I mean, it's fascinating how fast the valuation has grown. I mean, it was like a year and a half ago, two years ago, this thing was at$150 billion. They did that$150 billion valuation round. So, I mean, this is an incredible jump. But the revenue has been ramping, right? So like 25X revenue yield is not crazy. That's exactly how I look at it. So if I'm going to really say - But Nick, hold on. On that point though, they're at$10 billion right now. So not for nothing, they got to go from$10 billion right now to$20 billion by the end of the year. That's no joke.

3:31Clint Sorenson:That's a doubling of revenue in five, six months time. That's quite a feat in and of itself, right? Right. But even if they don't hit that, their revenue multiple from kind of throughout all of 2024 was between around 30 to 50. Right. You know, so just the fact that they're around that 25 mark, it's still pretty attractive given how fast they're growing. Right. And the TAM is just so massive to take over search, to take over, you know, the way people are using these LLMs. Yeah. I think that's a fair play. it's an attractive multiple. And that's the right way to look at a lot of these stocks. Well, I mean, I will tell you that the amount of revenue, especially compared to some of the other large language model providers, I mean, these guys are like gobbling up the whole market.

4:25Clint Sorenson:Like I think Anthropic is probably the next closest. And they're doing really well on the revenue side, right? But XAI is just getting started on that front. Cohere, which we're going to talk about in a minute, is like looking at a couple hundred million dollars of revenue, that's like pales in comparison. So I mean, these guys, if they in fact - Dropping on a couple hundred million, yeah. I know. It's like everything's relative, right? But 10 to 20 billion, I mean, this is going to be pretty impressive to see. I mean, they're talking like 100 billion,$120 billion revenue ARR numbers by the end, by 20, like 2029.

5:04Clint Sorenson:I mean, that's like, that's wild, right? That's a lot of that's like Google level type amounts of revenue, right? Pretty shocking. So we'll see. We'll see. But at any rate, it's going to be fun to mark this one up.

5:21For sure. But yeah, I just recall last night I was using OpenAI just because it was what I had to hand. And I guess I was at it for a little while and I didn't log in and it didn't prompt me with the paywall. It said, if you want to keep going with whatever you're up to right now, you're going to have to pay us some money. Like, oh, it's probably time to close the laptop anyways.

5:44Clint Sorenson:Oh yeah, sure. That's funny. That's where people are going to find so much value. When you hit that wall, you realize the amount of value it brings you because whatever I was up to, I knew it would have taken me a day of my time anyways. So everybody should be willing to pay that money. And if I'm going to pay a subscription or be bludgeoned to death by watching a thousand ads, I'd much prefer to pay a subscription. So if they can rival revenue of a Google at some point, I'm actually pretty bullish about that. Yeah. I like, I mean, listen, I prefer the subscription-based model versus, you know, where you are the product, right?

6:28Clint Sorenson:And your data is the product. I think that's, I'm witty on that. Okay. Let's talk about Cohere. So Cohere is a large language model business, AI large language model business as well. It competes with OpenAI and the rest. It's smaller Canadian firm. And they have a different approach. I mean, they came out of the gates. They solely focused on B2B channel, right? Which naturally, I mean, in my opinion, businesses haven't even really fully adopted AI in a proper way. But we'll get to that in a second. So they did$500 million at a$6.8 billion valuation. What's interesting about this, Nick, and I think you're uniquely positioned to discuss this, being the CEO of PM Insights, is I had the secondary market valuation at like$4.4 billion.

7:12Clint Sorenson:So it's trading at a discount, right, to the last round, which was at a$5.5 billion valuation. And this new round is coming in at 6.8. So if you were taking a position like currently at the current second-ever market price, you're looking at like a really attractive return here if, in fact, the market starts to trade up. But like, what are your thoughts? Like, is it, I was taking this as, as it's trading at a discount because it's just, it's like not really keeping pace with the open AIs. There's not a lot of buzz around these guys like there is with XAI. You know, you don't really hear their name kind of mentioned very much, but they're kind of digging in and like really working with companies and helping them implement AI.

7:55Clint Sorenson:It's a very different model, right. Then I think the other, the other firms are looking at. So like, what are your thoughts? What are your thoughts on that? Is Can you, in fact, dig in with a company and help them implement AI, but also be a closed model provider like Cohear? Or is that just you don't need your own models. You can just use open source models and do that same kind of service. What do you think about that? Or maybe most importantly, what do you just think about the secondary market valuation and having a traded discount? Then we're kind of coming in at a premium to the last round.

8:28Yeah, I think that's a good place to start. We'll start with the valuation and maybe some of this unpacks why they haven't grown, you know, at the same rate as some of the rest. So very quickly, and I'll pop this into the comments section once we post this. But Cohere for the last one year, last 12 months, is down 5%, whilst OpenAI, Perplexity, and Anthropic are up 170, 344, and 253, respectively. So, I mean, it's just really getting left in the dust in terms of returns. And that from a revenue perspective, they're all kind of singing a somewhat similar song with the exception of perplexity. So between Anthropic, OpenAI, and Cohere, they're between about 25 and 45.

9:17Okay. However, with Cohere being purely B2B, I think we need to focus and see if they're going to be successful with the long game. because it will take them a whole lot longer to ramp up and have those businesses fully ingrained onto their platform, which will be very different than someone that does B2C and B2B together, which will be the other three that I mentioned. If they have success here, you're operating in a bit more of a closed system. You're not letting, if you're really clever about your security with any of them, You shouldn't be having your data going out in the world to train. But with Cohere, I think you have a tighter ecosystem to manipulate, which could be to your benefit for businesses working with that and maybe with some of the others.

10:10And if they are operating in some regards differently, because I know Evan and I both kind of jive on this. We're not just going to use one LLM. We're going to use a couple and see how we arrive and make sure that, you know, the end result gets us to a better place. So I think Cohere is probably just playing a long game. And maybe they're kicking themselves a little bit for leaving money on the table for the short term. Right. But if this works out, it is a pretty attractive one because the valuation is just dwarfed by the other three.

10:40Clint Sorenson:Right. Yeah. Yeah, I mean, it kind of begs the question, like in the long run, if this is a$100 billion, $150 billion company, like this entry point's incredible, especially if they can deliver those kinds of returns in like the next five to seven years, that's like really, it would be really attractive. There's just, I don't know, I'm like a little concerned, like, you know, you get like some of the other large language model providers out there where they're just gobbling up market share. Now, to your point, that may not translate, the direct to consumer market share may not translate to b2b market share but there's this component of brand right and like risk around the you know provider that you're gonna you're gonna partner with too so like i don't know i just cohere you know cohere i have under research coverage i just um the ceo is like a very smart guy i've heard him talk a couple different times really smart guy as opposed to the ceos of the other ai companies right no i mean they're all like you know what i mean but like he's he's a thoughtful guy and like even though like he but he's taking you're right he's taking a very different approach right so like i i just wonder if that is in fact the right approach um uh or if it's going to be like too little too late because the other people have like run away with it you're just too far behind you know but the growth rate is still so impressive so if you want to compare it directly to to those i mean any anybody that's moving from what did you say 70 million to 200 you know by year end that's alright for a company that's already raised a couple billion bucks yeah it's true you're right okay let's talk let's wrap us up here today with N8N so this is like what I would call layman's terms AI agents for dummies right so if you're familiar with like Zapier is maybe the closest kind of parallel to this business where you can set up workflows right or zaps, they call them zaps.

12:33Like you connect, seamlessly connect

12:35Clint Sorenson:your different solutions that you use and you can kind of kick off different tasks automatically. And that's effectively what innate and does, right? But using AI, right? So you can connect it to different, your email, you can connect it to different solutions. You have QuickBooks, whatever, right? And that's like certain events happen, it will kick off these things. The AI will come in and do something and then execute a result. So super intriguing, I think. Right. But this company is quite a lot. Right. So Excel is is is leading this new round. They're up from three. So the round is going to be a two point three billion pre money there.

13:15Clint Sorenson:That's up from three hundred and fifty million dollar valuation four months ago. OK, so these guys have 50 ,000 customers. They got some big ones like Siemens. Also, they're doing over 40 million ARR as of Q2 2025, right? So the reason I brought these guys up, one, the valuations increases, I think, super interesting. But also, too, I think I'm in the constant search of finding out how people, small businesses, medium-sized businesses, or even large organizations are actually going to implement AI. and this kind of an app, like this is an AI app, what I would call an AI app, right? But it's focused on AI agents.

13:58Clint Sorenson:This like helps a guy like me who doesn't, who's not, I'm not like very proficient in, you know, coding. I know enough to be dangerous, but it's certainly not enough for these types of applications. Right. You know, I can now like leverage AI in a more a bespoke way and create these effective agents. And it can really add value, a lot of value to me, right? So I guess my thought, Nick, is you guys are a technology company. Are you guys standing up AI agents, per se? Or do you think that using some type of technology like this would be helpful to you guys in any way? Or maybe just maybe commenting more broadly about the broader business community.

14:42So we are testing agents generally, right? You have to wonder, is an agent for everything going to work fresh off the line? Is it like if Amazon came out and started selling everything on day one, would they have been as successful as starting with books and then evolving to become what they eventually became? So, for instance, just an example on this is I'll soon be investing in an AI-focused company that produces websites, marketing material, and pretty much your bring the company branding set to market. That's one of the investments I'm putting out there on a personal level. and would they, if I were to put that same task to N8N, are they going to do as good of a job?

15:43And I'm guessing not because you want to have a lot of robust specialization and training for something that something is going to be best at. It's kind of like having a great athlete and forcing them to play every sport instead of specializing, I think. So maybe that's not going to be the case and they're going to get enough training material and knowledge built that it's going to be awesome out the gate. And because I just look at a sale coming in as almost social proof and then we can jack the valuation a little bit. But there have to be better players out there because there are probably a thousand startups doing the same thing.

16:21So I'm not sold on this is kind of where I'm headed. If it's going to be the agent for everything as opposed to a specialized agent in a certain vertical and then bit by bit they expand yeah for instance i know how to value a bond but i'm not going to do that for my business right now because i don't want to compete with folks that have been doing that as their core business for three decades yeah yeah it's a fair point it's a fair point a bit small small bit of a rant but what do you think i mean i mean look i want to go with the every the everything guy or you want to yeah

16:54Clint Sorenson:Well, look, so I think what I've like kind of reset my expectations of what I would expect out of an AI agent. Right. And so like I kind of find these types of tools might be interesting in that context. Right. So like the way that I'm starting to think about AI, because I've I've literally in the last, I would say, 60 days, I'm like, man, I got to figure out how to actually like really properly use AI in my business and kind of like take it up a level. right so i was struggling frankly to think about how i could do that without spending like gobs of money or hiring developers to like go bananas on my business which i could sound like i think very expensive and probably what most small business owners would be thinking as well right like i'm gonna pay some developer come in here and pay them like thousands and thousands of dollars to do something and it probably won't even work or in six months time there'll be some better solution and then i'll like have you know like right jump the gun right yeah which which might be i think kind of your point like there'll be a better app for x and then you know that will be the right solution to use like down the road the way that i'm i guess i'm thinking about this stuff now and where i see applications for like in in in eight in or there's another company called like make.com or even zapier right for that for that matter too because zapier's got connectivity into like claw or anthropic and others is, is I think about AI now, like a smart 25 year old.

18:19Clint Sorenson:Right. So like, if I'm doing my research that I publish every week and I want to highlight, go and look at the 30, you know, pre-IP of stock companies that I would cover, if I had a really smart 25 year old research analyst, I'd say like, Hey man, every week I need you to deliver to me on like Friday morning, eight interesting things, highlights for the companies that we cover. And I need those to be like the best, like the best highlights. And usually I think for our, the people that follow our research, these are the types of things that seem to do what most well and people, what people are interested in.

18:52Clint Sorenson:Like, I need you to deliver that, come up with those eight and then give it back to me. And I want it on my desk every Friday at 8am. Right. So like you can use these tools to use AI to go do that for you. Right. And it will automatically kick it off every Friday. And then, you know, email me the results and like, you can connect it to Gmail and it will email the results to you. So like that, that's the kind of thing. It's, it's like very specific, right? It's like, as if you were talking with a 25 year old that didn't really know your business, right? You literally have to tell them what to do.

19:26Clint Sorenson:They're very smart. So they'll go out and they'll do it and they'll come back to you and you'll be like, you know, these six are great, but these two are no good. So come back and give me two more. and then they'll come back. But the six that they gave you were great. And then in time, because of memory and everything else with these AI large language models, like in time, like in six months, nine months time, they'll have figured out like what you like. And then the eight will be perfect every time they give it to you, right? Right. You know what I mean? And maybe even the most with our audience because over the last six months or 18 months, we've seen a growth for top eight list of this or top three, you know.

20:01Right. Exactly, man. That's pretty basic, right? Right. Well, but that's what I think you have bigger aspirations, I think, because you want to track everything.

20:08Clint Sorenson:Well, I think what's interesting is it's the combination of many of those things together. Right. And it's again, it's just like a 25 year old. So like you break down a task into multiple steps and you're checking it every step of the way because you just don't trust that 25 year old yet. Right. They haven't done. They haven't proved that they can do it yet. Right. So but over time, they get better, they get better, they get better. and then eventually you're nine months down the road or a year and you're like look well you just do this and publish it right like that's all just do this and publish it and then we're and we're good to go and uh and i think that's kind of like where i'm at with this stuff and i think the key is for me at least for me using tools like in in in a n or a make.com or zapier right is that like Define those workflows, define those different, you know, define those workflows, define the spots where I think this could add value or reduce time for me.

21:06Clint Sorenson:Right. And then and then take the time to build that out and then commit to it, just like you would commit to a 25 year old human. You can't expect them to be perfect right out of the gate. So you're going to have to invest into this thing for like three to six months and then it will be good enough where it's like really starting to incrementally add value to. to your business. But I've just come to the realization, Nick, like I can't just expect this thing to work out of the box. I have to like invest into it. You know what I mean? That's where I feel maybe this is part of that dynamic. You want it to work out of the box.

21:38You've got to grab one that's rather specialized for marketing or the next thing because then it learns from other people that are doing that same task, you know, as opposed to the holistic one where it actually learns most of the job from you.

21:54Clint Sorenson:That's right. And maybe some are better because then you've got an AGD copycat if you take the holistic one. That's right. As opposed to you're not the best marketer in the world. You're actually really freaking good at it. However, it will learn from so many other marketers for sure in this space. For sure. And it'll up your own game. Yeah, that's like super well said. That's like very well said. But like, it's like, so to me, to me, like you're a hundred percent right. Like if I want a great salesperson, it would be best to have a great sales focused large language model that's been trained specifically for that purpose.

22:31Clint Sorenson:You know, and I would still argue though, that like the personal context is important. So even though it's like the world's greatest salesperson, you're coming in and now you're selling my product. So you got to get smart on any salesperson would need to get smart on my product. they would need to understand the you know the profile of the customer that is the highest propensity to buy like all of that stuff right and and then and you know in probably short order because to your point that model's been trained on sales they'll have a much better impact faster than if it was something that was being trained kind of from scratch right to go out into the market and deliver value but i gotta tell you man like that this is the stuff that i'm this this nuance, in my opinion, is where the AI app stuff is really going to start.

23:17Clint Sorenson:We've been talking about AI apps for a long time and where the revenue investment opportunities will be. This dynamic, like what you just said, is going to start, I think, where AI apps present itself. And I'll give you one example, which we talked about a couple of weeks ago. Remember Cluely? I brought up Cluely, which was that AI app that watches your screen, and everything that you do on your computer, it listens to everything and everything you say here and everything you do on your computer, it records that and then it's available for the model. This to me is, I could go in and start to tell this AI agent, research agent, what I'm trying to do and use N8N and make.com or Zapier to kick off these workflows.

24:02Clint Sorenson:Or I could just stand up an instance in Grok or whatever, turn on Cluely. and say, hey, just watch me do this for the next six months. And then after you've watched me do it, literally stroke by stroke, every conversation, the whole thing, the output, all of it, I just want you to take over and start doing it for me. That to me is the easy button. It's a lot less work. So it's like post-training. It's like post-model training specific to your personal context. And that to me is so much more interesting that's a really interesting ai app like clue in that context right and the cool part about it is you know i'm a developed listener to the founders podcast yeah and the way he describes some of these um you know some of the companies and the founders are you know apple is steve jobs with a thousand lives yeah or you know that company is this this founder with a thousand lives eventually all of these AI apps are going to allow us to actually arrive to a single person company worth a billion dollars.

25:15Clint Sorenson:I believe that. The way you describe it is actually that way because it will exude the inner essence of you or me or whoever the person is to do it. That will be a very, very cool moment. Not to go tangential, but I wonder if ever we should be concerned that now you've gone head over heels and uh ag dylan and co is wildly successful we realized the market is way bigger than you or i clint or evan ever thought it could be right and whichever ai apps were used in the creation of of yours you didn't hit the t's and c's right or you agreed to something you shouldn't and they got not agd but it's bgd comes in 100 you know a person stands up your business as a direct competition.

26:06Clint Sorenson:Or the large language model. I use Grok. So Grok decides, oh, this guy's got it dialed in on this. I'm going to take that and now sell it to everybody else. And it's not really a competitive advantage anymore. I forget what Elon Musk said about IP being or should be dead or IP law or something. But yeah, there is a world where it becomes extremely dangerous. and then that gets in the way of progress because if you can't protect your IP, you're not going to take that risk in the first place. So yeah, there's so much unknown with this given the power of these AI agents.

26:50Clint Sorenson:Yeah, it's true. I will just say this. Yeah, for sure. I will just say this. If folks haven't checked out like an N8N or a make.com or a Zapier and kind of like done a little research around workflows, like I would venture to guess that everybody has some kind of like repetitive task that happens on like a daily or weekly basis. And you should look at automating that with one of these like workflow solutions and AI plus AI, because it's like pretty, it's pretty amazing. So I'm not saying you can do a hundred percent of a task but i bet it could do 80 percent automatically so you walk in if something takes you you know like when i do research every i do it every friday it takes me six and a half hours or seven hours to do it right for the week and um i'm now looking at implementing some of these workflows where i think i probably could cut that in half to maybe a third and you're not But as your first port of call, I think this says a lot.

27:55I mean, you're pretty with it with this stuff, but you're not trying to hire someone right now. Your first port of call is going to be this.