FULL TRANSCRIPT
Slava Rubin (00:00)
Welcome everybody, thank you for joining. This is the latest in our pre-IPO briefing series from Vincent. We're gonna be covering Perplexity today, the AI company, super exciting. My name is Slava Rubin, I'm one of the founders here at Vincent. We are your platform for all education and access to alternative investments. You can obviously check out our alternative investment report.
our daily newsletter, as well as our podcast, Smart Humans. With me is none other than the famous Jan-Erik. Jan-Erik say hi.
Jan-Erik Asplund (00:34)
Thanks Slava, great to be here.
Slava Rubin (00:36)
You know, we have these conversations about pre-IPO companies. Very excited for today's on Perplexity. We have a lot of folks who have asked about Perplexity. This conversation is brought to you with our partner from Augment. Augment's a platform that can help you get access to these pre-IPO companies so you can trade to be able to buy some of these amazing companies, including potentially Perplexity. So let's dive in.
So first from the audience, who are you listening? Quite a few of you are credit investors. As a matter of fact, almost two out of three of you are credit investors. And we actually have a decent amount of pre-IPO experience in the room. Well over 50 % is intermediate or advanced. And it's interesting in terms of, you plan to invest in the next 12 months? Again, over 50 % of you
plan on allocating capital into the pre IPO market, which I can't blame you. It's a very, very exciting market, especially at this moment. AI is one of those spaces. And again, over 50 % of you are suggesting that you would be investing. So you all know yourselves, but now you know a little bit more of who else is listening in. So now a word from our compliance department. Nothing in this presentation should be construed as an offer to sell securities or solicitation of an offer by securities.
All investments involve risk and the possibility of loss, including loss of principle, and neither past performance nor forward-looking information is a guarantee of future results. All right, we're now going to dive in and let Jan-Erik start with an overview on Perplexity.
Before we get started, how many of you are thinking to buy Perplexity shares in the next 12 months?
So are you interested to invest in Perplexity, binary? Yes or no? We'll give everybody 20 seconds to fill that out.
All right, do we have results? There they are. About half the market says they are thinking to buy. Okay, quite a few unsure and not that many nos Interesting, so there's quite a bit unsure, so that's why they're on the call. Let's move on. Let's see where, this changes by the end of this chat. All right, Jan-Erik, take it away, give us an overview.
Jan-Erik Asplund (02:58)
Great. Yeah. So Perplexity, know, probably a lot of people know Perplexity was founded in, in 2022, sort of just before the launch of, of chat GPT. So this was really one of the kind of, you know, OG companies in, the modern kind of AI space. So it was released again, prior to chat GPT built on, you know, what at the time was GPT 3.5. And the core idea was what if you basically took a large language model that you could chat with?
and hooked it up to the internet. And at the time and for quite a while afterwards, Perplexity was more or less the only company doing this. And so if you wanted to, for example, talk to Chat GPT or Claude or Gemini and have sort of the ability for those LLMs to access the web, because if you remember late 2022 when Chat GPT came out, the training cutoff was back in like 2020, right? was...
really not up to date. so Perplexity was for a time, the only place where you could talk to an LLM and have access to real time data from the web. And this was, you know, obviously really big, great product market fit. The product has expanded a lot since then. You know, if you could sort of look at the top right, the founder, you know, is now talking about Perplexity doing everything for you, right? Making transactions, booking your travel, getting you deals, shopping, all this kind of stuff has become very,
to the product, which has expanded a lot, but the origin story and the key is really about search.
And revenue wise, I think it's been reflected that it's had a great pattern of product market fit going from roughly 7 million at the end of 2023, almost 10X into 63 million in 2024 and halfway through 2025, which is the last time that we have comfortable numbers. We feel confident about numbers for Perplexity. They were already more than double their end of 2024 value. The sort of scuttlebutt is
roughly 200 million ARR as of this last September. But yeah, 30 million monthly actives, 780 million queries per month as of last year around that midway point. so yeah, the revenue ramp has obviously been really big. The main sort of revenue streams are these prosumer and to a lesser extent enterprise subscriptions, which is roughly 20 to $40 a month per seat.
And they also have sort of an API business, which is smaller. The main business is the sort prosumer subscription.
Slava Rubin (05:29)
So you have here 148 for June. What do you think, if you had to guess, is the revenue for 2025?
Jan-Erik Asplund (05:38)
Yeah, would guess probably like two, if I had to guess, would guess like 225, like in that range 225, roughly.
Slava Rubin (05:46)
So it went from seven to 63 to 225.
Jan-Erik Asplund (05:49)
Yeah, I think so.
Slava Rubin (05:51)
Got it. Okay, so it's like a 9x followed by a almost not quite 4x.
Jan-Erik Asplund (06:00)
Yeah. Okay. Yeah, roughly.
Yeah. And fundraising wise, they've been, you know, one of the sort of biggest, most hyped names in the space. You know, part of it is sort of emerging at the same time as ChatGPT and Claude and all these. So it's had a very, you know, easy time on the markets. They recently raised a $20 billion valuation, you know, which on that June revenue would be roughly 135X revenue. Now, if we said they were at
you know, maybe they were at 200, you know, 200 million at that point, then the revenue multiple would be closer to a hundred X roughly, but still a hundred to 135 X, whatever you want. It's still a fair amount higher than the multiples of OpenAI and even Anthropic So those guys are at 20 to 40 X and Perplexity about triple that. And then I threw in for fun, this sort of quote here from an early investor.
Jeff Bezos, who gave Perplexity this advice that, you your success doesn't rely on Google failing at search, which I think is a key theme around Perplexity and around AI in general. And that retention would be the most important thing, which I think is also true and valuable to think about with Perplexity. But yeah, that's the sort of long and short of
Slava Rubin (07:16)
So ⁓ that's an interesting quote. So your success does not rely on Google's failure. It's just the customer experience and retaining them. So it's said implicitly there, it's not about taking away their search. You just need to be awesome. A lot of people have mentioned, and we even have a question in the audience already about, this is a transition of search to AI. Is that happening? And is that the requirement for Perplexity? Or can Perplexity win while search continues to win?
Jan-Erik Asplund (07:45)
Yeah, that's a good question. think that, well, a lot of people are conflating kind of AI chat with being search, but I do think it's worth thinking about how, you know, it's really kind of a new thing, a new phenomenon, like similar to how vibe coding is a new thing to traditional developers, traditional development, because now as a developer, can, for example, you can contribute to any code base. can build any kind of project in any language.
you're not really restrained or limited by what you know personally. And I think a search almost similar where, know, search typically is, you know, I'm trying to find a movie theater near me or, you know, some kind of other information, but sort of having an AI chatbot that can talk back to you and is, you know, hyper intelligent, I think changes the nature of search a lot in ways that aren't immediately clear, but, you know, I think will change the sort of TAM, you know, the larger market.
It's not necessarily right to just comp it to directly to search and say it's a fixed pie and Google will have some and Perplexity will have some. We're already sort of seeing that not to be true with the amount of usage of search and search queries going up as people use AI tools. So essentially I'm thinking that, know, Google and Perplexity can both win. And I think they can both win as AI companies. I think the key thing is
how much is demand for information via AI going to grow? And I think we've only scratched the surface in terms of public adoption of these tools. So I think by a lot. And I think both Google and Perplexity are positioned to benefit from that.
Slava Rubin (09:19)
So I just want to double click on what you just said there. Is it possible that Google as a company, which is a loaded concept, do you think Google as a company grows and Perplexity as a company grows at the same time? Is that possible?
Jan-Erik Asplund (09:36)
Yeah, absolutely. think the, you know, again, sort of scratching the surface of how these tools, you know, the adoption of these tools, very few people know about Perplexity in the grand scheme of things. So I think as we continue to get, you know, public mass adoption behind these kinds of tools, I think we will see both succeed. And I think we will see usage of both grow by a lot, just as people's behaviors start to change around AI and in a way that they haven't Yeah.
I think so.
Slava Rubin (10:03)
We usually wait till the end to talk more about price, you already brought it up. So I just want to dive in on that. know, Anthropic, I think, just raised recently like a $350 billion valuation, and they just announced that they're increasing their revenue by 20 % for their target for 26, which means that they're going to hit 18 billion, hopefully, with a valuation of 350, which is below a 20x revenue valuation. I know that sounds nuts because that Anthropic is quote unquote so expensive.
but it's below a 20X revenue valuation. Now here with Perplexity, we're talking over 100 or potentially 135X. So why is Perplexity in this different stratosphere of a number? How do you explain that?
Jan-Erik Asplund (10:51)
Yeah, it's definitely high. I think there's a few things. One is sort of Perplexity being, you know, and I don't think anyone knows all these numbers for sure, but you know, early on there was a bunch of coverage and reporting about their gross margin, you know, gross margin being a bit higher than most kind of AI companies that are responsible for training their own models and compute all that inference.
And Perplexity doesn't do that, right? Perplexity doesn't have a lot of those massive costs that you need to continuously raise from sovereign wealth funds to support. They're just a layer on top of the models. And so in a way they have this very kind of nice, less costly business model where they can charge, they charge you $20, $40 a month. And all they have to do is link you to the Anthropic API or link you to the OpenAI API.
So the business model has that sort of nice quality. I think the other thing is, you know, they are been around the longest in terms of these AI upstarts and they are a sort of index on all the models, right? Because if you use chat GPT, you're locked into OpenAI. If you use Google search, you're locked into Gemini, but with Perplexity, you get access to all of them. it's sort of, I think it's like the play where you can, if you are a person who wants to use, be able to use different models, Perplexity is the sort of biggest.
⁓ company in that space and they have sort of owned it. So I think in a way they become that, the face of that kind of investment in like multiplexing models. Yeah, those are the two things I can think of that helped me explain it to myself.
Slava Rubin (12:28)
Got it. Can you walk through their tech stack a little bit?
Jan-Erik Asplund (12:32)
Yeah, perfect. yeah, they really started out, like I mentioned, with GPT 3.5 hooked up to the Bing search API. GPT 3.5 would come up with basically a web query, and then they would run it through Bing, get the results, synthesize them with GPT 3.5 again, and present them back to the user. So this let them get started really quickly, although it created this dependency on both Bing and OpenAI.
And they've since then transitioned using more open source. they've also, they've fine tuned their own model to sort of, to sort of do some of the orchestration work around search. And they still are using, you know, external web crawlers like from Brave and they are still using, then, but now the core of the product is really these other models, right? So you can get access to, to every other foundation labs models in Perplexity. The other sort of underlying thing, you know,
that folks found impressive early on about Perplexity that I haven't mentioned is speed. so latency has become this huge issue with all generative AI products. And you're starting to see companies going harder on speed, low latency, like Cerebras Groq and others are really focusing cursor, are really focusing on how do we increase speed by a lot? Because the idea is people can be more productive if they're faster.
And Perplexity has been long sort of obsessed with latency and early on the product being as fast as it was, was really the key thing that made it successful.
Slava Rubin (14:02)
Got it. We have a question. How would you compare it or contrast with Exa with our semantic search? Do you have a comparison there?
Jan-Erik Asplund (14:09)
Yeah, so Exa is more, you can use it as a consumer, but Exa is more B2B, B2B search API. And so you would use it if you wanted to build an agent that could, for example, do search across the web. If you were a salesperson, you might want an agent that can search for people that fit the right profile based on their LinkedIn bios or whatever for outreach tool. So you would use Exa most likely.
there's an overlap in that you could use Perplexity's API also to do search, but their API is not a hugely emphasized part of the product. I would say it's been sort of superseded in some ways by OpenAI having web search in their API, which we'll talk about. And so yeah, think the main split is Perplexity is more consumer now, more consumer focused, and X is more B2B focused.
Slava Rubin (14:55)
Got it. Is there any other context you want to set as it relates to search or do feel like we've covered that?
Jan-Erik Asplund (15:00)
I think we've covered a lot of it. I do want to say, know, part of the initial thesis for Perplexity was that Google would never sort of kill their golden goose of ads, right? And so when you search in Google, you get ads at the top. You know, the first few links are always ads and that's been the core of their business model for since, you know, 2001. And it's made them, you know, a gigantic company that hasn't really proven to be true. So they've adopted AI overviews.
you know, seemingly quite readily, know, seemingly quite eager to cannibalize their existing search business. What's happened so far, at least what we have from earnings calls is that, you know, AI overviews are definitely big for Google. And to refresh this, when you type in something on Google and you get the AI answer at the top, which you can click into for more information or ask it a follow-up question. So...
What they're seeing currently is that these are obviously getting a lot of traction, a lot of engagement as they're rolled out and it's actually increasing revenue so far. So the ads are being placed in the AI overview summaries and those ads are apparently getting higher click through rates than the traditional search ads. So at least so far, and we just have Google sort of results show, it seems like more AI is driving more search, which is driving
more ad revenue, which is good for Google, but also potentially good for complexity. So again, not a win-lose, but a win-win.
Slava Rubin (16:28)
Why do you think, so I'll counter that for a second. If I'm hearing originally that Perplexity is helping to replace Google search, but then you're telling me that Google is bringing AI to the forefront of the search, which is driving more ads, which is quite positive. Why is that not net negative for Perplexity?
Jan-Erik Asplund (16:50)
Yeah, I think that specifically. So, and just to bring into like OpenAI, know, launch their own web search. And it was very bad at first. You know, this was like 2023, 2024. Now it's very good. I would say, you know, personally, it's comparable to Perplexity. It's comparable to Google. And so I think there's definitely been cannibalization away of some of the more basic, maybe casual Perplexity.
usage, maybe some of the free usage, because now instead of, you if you're a very occasional user, yes, Google is going to be comparable. ChatGPT is going to be comparable. And you might not need to go pay for Perplexity. But at the same time, you know, we've seen Perplexity's revenue continue to grow despite this. And a lot of that is kind of their other, the other parts of the product. So they expanded very quickly outside of search, probably expecting that this would be,
This would not be their forever moat. And so the product is pretty extensive now with multiple models, deep research, they've begun very hard on the finance offering. And there's sort of a bunch of these sort of products inside Perplexity now that I think have picked up the slack for the most part. Although I think you're right that some of the core basic usage has been commoditized.
Slava Rubin (18:10)
You mentioned that they're more consumer. What's their business model where they're making most of their money?
Jan-Erik Asplund (18:16)
Yeah, most of the money is from the $20 a month subscriptions where you get access to a limited usage of the models and you can access deep research, stuff like that. You also can upgrade to the $200 a month plan on Perplexity, which is gonna give you unlimited usage of all the models, or unlimited with an asterisk, and then give you access to
their web browser, Comet, which we haven't talked about, that, sort of, yeah, unlimited access to create things like dashboards, spreadsheets, presentations. Not clear right now, sort of what the split is, how many folks are on Macs, but yeah, and then there's enterprise, which is sort of, you know, all of what Pro has, but just with the admin controls and single sign-on and stuff like that for the enterprise. And then sort of a more emerging stuff, stuff like ads.
Being able to give Perplexity a take rate when you buy something through Perplexity, stuff like that. They're doing some of these partnerships with telecoms, trying to get sort of mass distribution onto cell phones and that kind of thing.
Slava Rubin (19:24)
As it relates to their business, somebody mentioned, are they profitable? I think the answer is no, right?
Jan-Erik Asplund (19:30)
I don't think so, I don't think so.
Slava Rubin (19:33)
Okay, great. Moving on, talk to me about competition. We've kind of skirted around that. But can you just give me more directly, you know, what is the competitive situation?
Jan-Erik Asplund (19:43)
Yeah. So like I said, a lot of the thesis or, you know, marketing around Perplexity early on was that Google could never implement AI search because, you know, people, if they just give you the answer, which is what Perplexity does, then people wouldn't click on ads and Google wouldn't make money. And Google makes so much of their money from search ads that this was seen as something they couldn't do. Well, they have added AI overviews ⁓ with ads.
And so Google has sort of definitely come very strongly into the space and is probably the number one sort of, the number one kind of competition or Perplexity on search. But OpenAI is also right up there. They sort of folded it in very organically into chat GPT where a web search is triggered anytime you ask a question that relies on live real-time information. And they've added web search into the API, including sort of their deep research models.
So if you're a developer building something that requires web search, whether you need a very quick web search to get a piece of information or a more deep, multiple minute long research deep dive, then you can get that all through the API and similar with Gemini. So those are the main two competitors. You also do have sort of Anthropic a little bit, although their search, their sort of web access and deep research stuff is not really
in the same league. So yeah, I would say that those are the the main competitors.
Slava Rubin (21:04)
If you were gonna say Perplexity is unique because of A, or C, what is that unique thing that Perplexity has? And I think I can answer that for the others on this list on the right. What would be the unique thing that Perplexity has? Just one thing.
Jan-Erik Asplund (21:27)
Yeah, I think it's kind of the, you know, they have a very, they have a very fast product cadence. So they ship very quickly and they ship a lot of things, which I think is big. Like Gemini, know, OpenAI, know, they'll ship, you know, stuff like Pulse, you know, and that'll be, you know, a big thing for like a whole quarter. And part of that is because these places have to take care of their, ⁓
they're building around their own model. Whereas Perplexity is sort of very, they integrate with every single model. They don't have a sort of foundation model that they need to protect and maintain. So they can sort of be very aggressive in trying to expand across the surface with different products. So they have Discover, they have Finance, they have Travel, Shopping, Academics, Sports, Patents. Everything's plugged into Perplexity Search.
And so they're just trying to build aggressively for all the different use cases and verticals they can. This is very much not a one word answer, but I would say yeah, it's sort of aggressive product cadence, shipping velocity is really their key thing that I've seen.
Slava Rubin (22:28)
Got it. So if you included XAI and maybe some others on this chart on the right, it'd be fair to say that each of those are these LLMs that have their own kind of infrastructure. Perplexity sits directly on top of them. Is it fair to say that any one of these competitors can try to move aggressively to commoditize Perplexity's, interface differentiation, or copy their product on the front end to kind of
take them out of the picture or make them harder to grow? What do you think about that?
Jan-Erik Asplund (22:59)
Yeah, so I think the core thing Perplexity has is sort of multimodal and actually none of them can really do that. And none of them have. So they all have chat, they all have chat and they all have search, but none of them allow you to access models from the other companies, which is, you know, really is one of the main things about Perplexity that's valuable. But also XAI, I'm glad you mentioned because they are sort of interestingly similar in that they have this real-time access to
you know, data from X, which is very useful for especially some of these finance use cases. But XAI, again, we're actually seeing B2B enterprise is kind of the biggest driver of revenue there. So I would say, yeah, I would say they can't really emulate the core theme of Perplexity, which is access to many different models.
Slava Rubin (23:49)
So, you know, in the home mortgage area, what happened in the early days of the internet was somebody said, you should come onto our site. We could give you a great price for a home mortgage. And then there was another company that said, I can compete with you. I can give you a good price for a home mortgage. And eventually there were some sites that said, you know what? You guys actually like looking at multiple home mortgage prices. We're going to aggregate it and let you look across many home mortgages.
And eventually some of these individual players, what they said is, ⁓ we're going to give you our price for our home mortgage and just give you the option to see what all the other other prices are as a feature. I guess my question is, is the indexing across other models a feature that can be added? And maybe the Chat GPTs or Claudes of the world figure out that just adding that feature will make a more sticky customer, whether or not your customers actually use that feature or not.
Jan-Erik Asplund (24:44)
Maybe, I would be surprised, maybe mainly because I'm seeing there's already this kind of aggressive, the dynamics between them are pretty aggressive on that front in a way that you don't really see anywhere else in tech. Like for example, have, Anthropic has built a kind of model that's more focused around writing code and
Now, if you are trying to use that model to power a competing coding tool to cloud code, Anthropic has reserved the right to basically cut off your access entirely. And they have done so. And they have also done so to OpenAI itself. So if you work at OpenAI, can't use your emails cut off from using cloud models. You're seeing this kind of
this very aggressive competitive dynamic. And so it's hard to see them kind of, I guess, reconciling or offering up that kind of access in the long run. And yeah, I think they're all sort of competing to be the one that you pay for.
Slava Rubin (25:48)
If there's always competition between A, B, C, and D, it seems like then there's going to stay an opportunity for an aggregator. Is that kind of the takeaway for maybe Perplexity's value add?
Jan-Erik Asplund (26:00)
I would think so, yeah. And I think early on there was also the fear with complexity that all the models were converging. And I think GPT4 Claude's on it 3.5. There was kind of a sense that that was right because they were starting to get more more similar, the models. But I actually think the last year or so I've felt at least like the models are diverging more where I feel like they're being sort of, you know.
post-trained and fine-tuned in different directions and for different, for to be good at different things. And so I actually think that that sort of narrative makes more sense to me now. And I agree that that's probably with Perplexity, that's probably, you know, one of the key kind of points in their favor is this idea of, of there always being competition, especially with open source, you know, for example, which was a big one for them last year with DeepSeek, you know, OpenAI is not going to integrate. ⁓
with DeepSeek, no one is, no foundation lab will do that, but Perplexity can because they're sort of, you know, the neutral ground.
Slava Rubin (27:00)
Yeah, I'm in the camp of differentiation as well. I don't think there's gonna be one model that rules the world. I mean, everybody knows what a knife is and it's easy to say I need a knife. But once you wanna get into the subtlety of it, there's lots of knives, whether it's the butcher knife, the steak knife, the butter knife, they're all a knife. But once you wanna get into more expertise, there probably will be something that's better at finance. There will be something better at images. There will be something better.
at research, there will be something better at coding, blah, blah, blah. Will one model be the best across every single one of these things? That's hard to believe, right? So it's interesting how that's all going to play out. I mean, I guess the counter to that to argue with myself is Google is pretty good at search across the board, right?
Jan-Erik Asplund (27:44)
Yeah, definitely. I think Google on search has surprised a lot of people with how willing they've been to cannibalize their existing search business. Although I'm not sure if it's cannibalization if revenue increases as a result. to really reinvent the search experience, I think, and it's still rolling out. I would say, they're definitely the biggest risk factor there, I would say. Yeah.
Slava Rubin (28:09)
So give me the, what's the upside, what's the risks for this business in the next few years?
Jan-Erik Asplund (28:17)
Yeah, I mean, think we've talked about it a bit, talked around it. think like there's going to be continued competition that appears between all these different model companies and between open source models, which come in and every once in a while you got a drop from China that is a model that does what American frontier models do or almost as good and for 10 less, 10X cheaper, which obviously helps Perplexity's gross margin if they can serve you that model.
So I think that is a huge upside factor that there's always gonna be competition and they're always gonna be good at different things. And people will always wanna access all of them at once without having to go and pay $20 a month to each one. So I that's right. I think the other part is what I talked about at the beginning, which is that we're only just scratching the surface of adoption. If you look at Google trends and see Perplexity right now, it's a smidge on the bottom of the chart compared to Google. And as know,
years continue to go by and people become, know, the behaviors change and start going from search to AI. I think Perplexity is really well positioned to, to, you know, win a lot of that traffic. And at the same time, you know, they're really ⁓ doing a lot around vertical, sort of different verticals to try and stay relevant as search itself is sort of commoditized. And, know, that includes this sort of finance and shopping being the two biggest ones right now where they're trying to drive, you know, this is more and more
you know, ⁓ habitual usage, not even daily usage, but more like hourly usage, trying to make sure that you're using Perplexity to do all these kinds of tasks that OpenAI and Anthropic haven't built ⁓ specific kind of features or products around at this point. So I'd say that's the big thing on sort of, you know, sort secular growth drivers. And then I think the big thing, you know, that I'm interested in on the market expansion side is really these telecom partnerships that they're doing.
And they've done a few big ones in Europe and in South Korea and basically getting Perplexity installed, pre-installed on Android phones that are going out to folks in those countries, which I think is interesting, you know, seeing that telecoms could make deals with OpenAI, could make deals with Anthropic, are choosing to, you know, bring Perplexity on. And, you know, most likely I would think Perplexity is sort of an easier partner to negotiate with.
and to work with there. They are not a 10, 20 billion a year foundation model lab with a lot of leverage and Perplexity needs the distribution. yeah, those are kind of the key things I think are interesting upside-wise.
Slava Rubin (30:44)
Do think there is a specific vertical, if you have to choose just one vertical, that you think Perplexity is going to dominate in the next couple few years?
Jan-Erik Asplund (30:54)
Yeah, that's a good one. Yeah, fine.
Slava Rubin (30:57)
real estate, manufacturing, does it have any specific access to data or the way it's doing things that it's gonna be known to, yeah, go to that, go to Perplexity for that.
Jan-Erik Asplund (31:08)
I'm not sure. think finance is the big one that they have indexed on. And I think that they are sort of giving pride of really prominent place in the sidebar and the product experience. And I think that they're betting on finance being a key one and travel to a lesser extent. Yeah.
Slava Rubin (31:30)
to
finance. That's an interesting one.
Jan-Erik Asplund (31:32)
Yeah.
Slava Rubin (31:33)
Do you
use Perplexity, Jan-Erik?
Jan-Erik Asplund (31:37)
I do not.
Slava Rubin (31:37)
You're a survey of one, which is fine, but if you were thinking to use Perplexity, what do you think you might use it for? What's the first thing you're thinking to use it for, if at all? Or you're like, I don't think about it at all, it's not relevant to me.
Jan-Erik Asplund (31:53)
Yeah, well, I would say I used to. I did use it quite a bit for a while until it felt like search became a default in Chat GPT and Gemini to the point that I didn't need Perplexity to do search. But I do think finance is probably the main one. I I spent a lot of time, you we mostly focused on private markets at Sacra but we do spend some time looking at public market comps. I think finance is probably what I would most use it for.
of the features on the platform. I think the other big thing is Comet. You they're sort of agentic browser products. If you remember, they offered to buy Chrome. They didn't buy Chrome. They sort of built their own Chromium browser, which is you chat with it and then it performs actions in your browser. And I've tried that out. think that could be, I think that could be bigger than finance. I think that could be really a key product for them in the future and one that I would be interested in using. And, but yeah.
Slava Rubin (32:45)
Here's an interesting comment from the audience, an anonymous attendee. So several months ago, I brought a product from Newegg via Perplexity with a $50 credit provided by PayPal. And the process went very smoothly. Automation of asking for the product advice and then comparing the recommended options prices, including discounts, was very valuable. So that's interesting. So somebody really liked it for their shopping experience. So that's interesting. What do you think about, you mentioned
comment, what do you think about the agents, Perplexity AI agents?
Jan-Erik Asplund (33:16)
⁓
Yeah, I think it's a really interesting emerging space that Google has released a product in, actually OpenAI has released a browser. I don't think any of them, and Claude has an extension. So I think they're all going after this, ⁓ but Perplexity seems to, they seem to be wanting to make this kind of like the second core product. And in my brief demos of it, I was pretty impressed by how well it was able to sort of...
you take instruction and manage in the browser and do sort of long running tasks autonomously, you without stopping or getting confused, which is often the case with these. I do think they're all roughly at the same stage right now in terms of capability level, but I think that will, you know, if we know anything from watching LLMs in general, they're going to get better and better at it.
Slava Rubin (33:58)
What are the risks?
Jan-Erik Asplund (33:58)
there's quite a few. think like we've talked a lot about OpenAI, Google, commoditizing search. They've sort of, you know, in a way kicked out the legs of Perplexity's business that they started with and made that sort of, you know, irrelevant. I think there's still people obviously, you know, judging from the revenue who want to be able to search across many different models and do other kinds of tasks that Perplexity offers. But
Yeah, it sort of remains to be seen how durable that is. Like all this extra product stuff, finance, shopping, there's definitely value in it. ⁓ That's hard for these other companies to replicate. Like shopping, like the last question, I think that's a great point. Like shopping on Perplexity is nice because they built a whole interface around shopping. You can't just really do it in a traditional chat interface. If you try it on chat GPT, it doesn't really work that well to just have it in the default chat window. You kind of need to have a separate
interface so you can have big visuals and compare multiple things. And they've done a good job with that. Just a testament to the product velocity and ability to really rapidly iterate on the sort of UI because they're not responsible for the models underneath. But that said, all that stuff is still very nascent in terms of contributing to revenue. So it remains to be seen if it'll be enough to sort of make up for search being commoditized and
I think there are other things you can point out around these legal battles that they've had over search. Every AI company has those. Some of these still being dependent on Brave's search API. I think really the key thing, the one big risk is really that search being commoditized means that the reason to use Perplexity gets weaker and weaker as you can do more and more in Gemini and OpenAI and so on.
Slava Rubin (35:41)
Yeah, that all makes a lot of sense. Give me the future projections. What's the bull bearer, you know, base case?
Jan-Erik Asplund (35:50)
Yeah, roughly, think, you know, so that was sort of a downer, that last slide. I think the upside is this whole space is still expanding so much that every foundation model lab and AI company that is decent will have usage and revenue growing for years to come, you know, unless something terrible happens. So, you know, I think even in the sort of more conservative, bear case of growth, they end up at, you know, a billion in revenue.
roughly in five years from now. And I think the base case is more like two, and maybe the bull is like the four or just under four. Now, that's pretty significant. Let's say, Anthropic got maybe nine as of this year, OpenAI at 20 as of this year. So they'd still be sort of dwarfed by the foundation model labs, but obviously great business. I think the question will be sort of valuation. So these multiples, it's hard to know if you had to...
guess what multiples will be used to value AI companies five years from now. But if you were to say, you know, value them at 6X on the bear case, you know, then you'd end up at a 6.6 billion valuation, which is about a third of their valuation today. So that would be problematic, obviously. And then the base case here, let's say it was a 19X multiple, similar to Anthropic today, you'd end up at 40 billion. I mean, that's double their current 20 billion.
And then the of the bull case is like a little over quadruple their current valuation. So yeah, all of it is very hard to say and hard to predict, but that's roughly how I'm thinking about it.
Slava Rubin (37:20)
I mean, today they're at 200 million directionally. So getting to a billion dollars of revenue by 2030, that's not nothing. I mean, that's a sum. But you're not painting a very rosy picture for getting to a billion dollars of revenue, right? Because it's not such a huge growth rate. So you didn't give us such a big multiple. And today it's valued at 20 billion if you could get in on that. So that's a majority of what we wanted to cover. I want to quickly do a survey again, which is.
Jan-Erik Asplund (37:29)
No.
Slava Rubin (37:48)
Did we move the needle on how many people want to invest? ⁓ do you invest or not? Let's give me that survey. Everybody gets 20 seconds. And then I wanna hear your final thoughts, Jan-Erik, and my final thoughts.
All right, so survey says.
All right, I think some unsures the yeses stayed pretty flat and some unsures moved to no. So the Perplexity marketing department should not hire us or be replaying this is the net takeaway. Now, obviously we're just two people doing whatever we can here. Jan-Erik, what's your final word on your thought on investing here?
Jan-Erik Asplund (38:30)
Yeah, it seems very expensive to me. mean, yeah. I'd say 20 billion. Apparently they fielded offers at 50 billion. I do think it's a great product in a lot of ways, very well designed. think it was hugely valuable really early on and insanely fast. I still like the product. Personally, I don't use it now that search is sort of a feature of every AI model.
And so, know, yeah, it's hard for me to get super invested when I'm sort of a churned user. But I think they do a lot of really interesting product stuff that they put out. And I think I wouldn't bet against them.
Slava Rubin (39:08)
Are you buy sell hold?
Jan-Erik Asplund (39:09)
hard to say, but probably, and maybe, maybe, maybe hold. I think I can't sell. Yeah.
Slava Rubin (39:16)
Okay, so my point of view is in line with what Jan-Erik is saying on price, I think it is obnoxiously expensive. I do not think it deserves that multiple. Now, does that mean I am a sell? It does not. I think it's got a lot of AI hype and a lot of halo around it. And I think AI search and AI answers is going to be such a big market. There is an opportunity for Perplexity to continue to grow.
So I think this is a question of opportunity cost. I think this is a question of where do you want to put your thousand dollars? I think that you can enter at 20 billion if you get that opportunity. And I do think it will go up. I think it might hit 25, it might hit 40, might hit 50. There's a chance it might hit 100 billion. I am not optimistic about 100, but I'm not anti. I think you can make money. But is your thousand dollars best spent putting it here versus Anthropic or somewhere else in the AI world? I think the answer is no.
I think there's more money to be made in other AI plays than there are here. So I am a hole to sell. I do think you can do a trade and maybe get an up round in the next year or two and make a bunch of money, but I don't see this being one of the breakout companies 10 years from now and that's how I like to invest. Jan-Erik, any other thoughts?
Jan-Erik Asplund (40:31)
No, think you said it all. think, yeah, there was a lot of news about Apple acquiring Perplexity which I thought would be very interesting. But I think that that's over now. There's some talk about deals with Samsung. I do think they're going to be one of the big names of AI for quite a while, but I share your trepidation.
Slava Rubin (40:49)
So many people have asked for the deck or for this video. As always, we will archive this and it will be available for all. So thank you everybody for joining. I'm sorry I wasn't able to get to the last questions that started after the survey, but we love everybody's engagement. Jan-Erik, thank you very much. And this completes our pre-IPO discussion about Perplexity. Have a good one.
