From Startup to Exit
Welcome to the Startup to Exit podcast where we bring you world-class entrepreneurs and VCs to share their hard-earned success stories and secrets. This podcast has been brought to you by TiE Seattle. TiE is a global non-profit that focuses on fostering entrepreneurship. TiE Seattle offers a range of programs including the GoVertical Startup Creation Weekend, TiE Entrepreneur Institute, and the TiE Seattle Angel Network. We encourage you to become a TiE member so you can gain access to these great programs. To become a member, please visit www.Seattle.tie.org.
From Startup to Exit
2025 Investment Outlook Panel - In the age of AI
Listen to our 2025 Annual Investment Outlook Panel with four top VCs based in Seattle and the Bay area. Get their perspective on the investment climate in 2025 and their plans to invest specifically in AI-based startups. See how they did with their 2024 forecast and what their big predictions are for 2025.
Tim Porter, Managing Director, Madrona Ventures
Tim joined Madrona in 2006 and invests in B2B software companies in the Pacific Northwest. He is currently interested in intelligent applications and SaaS, cloud-native software infrastructure, the modern data stack, machine learning, DevOps, and cybersecurity. He is also a board member of numerous Madrona portfolio companies.
Sudip Chakrabarti, Partner, Pruven Capital
Sudip is a partner at Pruven Capital where he works with companies that are transforming the enterprise, cloud and infrastructure markets. Sudip's past investments include Exabeam, Heptio, Serverless.com, Datacoral, Spotnana, Temporal, Tesorio and Polly. He had also been a board observer at Databricks, Digital Ocean, Actifio, Mesosphere and Samsara.
Kellan Carter, Founding Partner, FUSE Ventures
Kellan Carter is a founding partner of FUSE where he focuses on early-stage investments in intelligent software in both horizontal and vertical categories. He currently sits on the Boards of Zuper, Pictory, Avante, Owl, Xemelgo and PDM Automotive.
Suman Natarajan, Senior Principal, B Capital
Suman Natarajan is a Senior Principal at B Capital’s San Francisco office, where she invests in startups powering the enterprise, across domains such as cybersecurity, cloud infrastructure, DevOps, data and ML/AI, and B2B SaaS.
Brought to you by TiE Seattle
Hosts: Shirish Nadkarni and Gowri Shankar
Producers: Minee Verma and Eesha Jain
YouTube Channel: https://www.youtube.com/@fromstartuptoexitpodcast
But now over the last two funds of ours, to call it the last five years or so, it's it's been about 15% of our dollars have gone into this space, kind of life science. And we we started out talking about it as the intersection of innovation where cloud and AI meet uh life science and biotech. And so those are the the most exciting, most competitive deals. Both of those things, I think, drive the growth stage market. And I think we'll see more of that in 25.
SPEAKER_00:Welcome to the Startup to Exit podcast, where we will bring you world-class entrepreneurs and VCs to share their hard-earned success stories and secrets. This podcast has been brought to you by TIE Seattle. Thai is a global nonprofit that focuses on fostering entrepreneurship. TIE Seattle offers a range of programs including the Go Vertical Startup Operation Weekend, the Thai Entrepreneur Institute, and the Thai Seattle Angel Network. We encourage you to become a Thai member so you can gain access to these great programs. To become a member, please visit www.seattle.tai.org.
SPEAKER_04:Hello everybody. Welcome to another episode of From Startup to Exit. My name is Gary Shankar, here along with my co-host Shirish Natkarni. The podcast is produced by the Thai Seattle Chapter, a non-profit focused on fostering entrepreneurship. Shirish is an author and serial entrepreneur. His latest book, Winner Takes All, is available everywhere books are sold. We both would like to thank all of you for supporting us on this journey. For those here for the first time, please like and subscribe to our podcast, and both of us will be grateful. It is a brand new year. And at the beginning of each year, we talk to four leading venture capitalists to get their prediction and see how they did on their previous year predictions. As we enter 2025, we are talking to four of them again. Shirish, please introduce our guests.
SPEAKER_06:Thank you, Gauri. I'm really excited about our podcast today. We have an exceptional panel of VCs from Seattle and the Bay Area. From Seattle, we have Tim Porter, managing director at Madonna Ventures, and Kellen Carter, partner at Fuse Ventures. And from the Bay Area, we have Suman Natrajan, who's a senior principal at B Capital, and Sudeep Chakrabarthi, partner at Decibel Ventures. Welcome everyone. Now let's get on to our questions. So Tim, um, now that we are in 2025, uh, would like to understand from you what was 2024 like from an investment perspective compared to the previous year? And was AI the primary focus of the investments?
SPEAKER_01:Yeah, AI certainly was. Um it is amazing. We're in 25. Thanks so much for having me. Um, you know, in reflecting back on Madronas 2024 and mine personally, there was a lot of investment in AI, and there was just a it was a very active year. Um we were uh our pace was really kind of consistent on early stage companies between 2023 and 24, and our new investments in some of the what we call acceleration stage actually ticked up a bit. Um and I think those two um metrics are consistent with the overall VC market. So just to give you some specifics, so in 2024, Madrona invested um in 16 new early stage companies. Um there were five pre-seed, five seed, and six series A. Um in 23, we did 13 companies. So 13 companies to 16 companies for both years, uh the split between you know seed series A was about the same. You know, we're kind of roughly, you know, a third, a third, a third between pre-seed seed and series A. In 2023, across those deals, we invested 56 million. Those were initial checks, and in 2024 it was 55 million. I was actually surprised when I went and looked at that that it was that close. Um, so a few more companies generally pace was similar. We also invest in these kind of series B and B plus companies. We call it acceleration stage because we're a little a click earlier than you know firms that consider themselves growth stage. And there have been 23 deals. One of them was a new company and two of them were into existing companies, which is unusual. The fund exists to invest in new companies. So that sort of shows the overall market was still pretty slow for those later stage deals. And then in 2024, we invested in uh five companies and they were all new investments. Um, and uh mostly were AI related. Um, in terms of um, you know, kind of sectors, you asked about AI on the early stage companies, and I'll mostly talk about those. That's where I spend most of my time, and that's you could tell that's the bulk of our focus. Of those 16 deals, I'd say six of them were AI intelligent applications, so app level, seven of them were sort of data and infrastructure. And all of those data and infrastructure companies had some AI platform, you know, kind of flavor, one cybersecurity company. And then we do some, we call it AI bio or digital biotech. So we did two of those. And they're so those are kind of AI deals also, but in the biotech realm. So that gives you a bit of a flavor. I think that our biggest theme for the last couple of years has been applied AI, intelligent applications, and the software infrastructure that powers those intelligent applications. And um, we expect that to continue into 2025. And I know we'll uh we'll pick that apart and talk about that more looking forward here as we get into the conversation. But hopefully that gives you a sense. So overall, it was a busy year, pretty consistent with where we've been the last the last couple of years.
SPEAKER_02:Yeah, we we definitely saw a pickup between 24 and 23, which was really exciting uh as an investor. I think the the market started to ease up on uh valuations. And I think we just also saw enterprise buyers also starting to ease their budgets as well. So that bode well for the broader market in general. I think uh we're hoping to see uh that continue to unfold uh in 25, which is really exciting. And absolutely, I mean, AI uh and just core tech core enterprise software in general has become almost one with uh AI. You do have kind of the core AI platforms, but we're seeing AI getting sprinkled in or embedded into almost every existing solution, even if it isn't front and center. So, like it or not, it is the focus, even if uh we're trying to have some kind of Contrian uh strategy there. So, yeah, absolutely.
SPEAKER_03:So 24 actually was busy, but in a good way. I think in 23 in particular, things were crazy, but also with crazy deals going down, like in terms of valuation, um, just you know, kind of nature of some of the companies that were getting funded uh astronomical sums. In 24, I think uh the market normalized in a positive direction. I wouldn't call it fully normalized, uh particularly for any company that could spell AI, but still it was much better than 23. And I think for us in particular, we did uh multiple investments. I would say not every one of them was related to AI, but I would say many of them had AI component. And it was actually refreshing to see some of those companies actually showing real customers, showing real you know revenue. So I think you know some of those companies obviously uh were able to comment premium valuations naturally, but both founders and investors were more clear-eyed in 24 for good reasons because last thing I would say is you guys know this, which is the valuation of the current round does matters because you have to really uh hit it out of the part for the next round, right? So if you're raising at a crazy valuation this round, the next round becomes much harder to raise. So that's why I think um the normalization was needed and uh actually appreciated.
SPEAKER_07:You know, I think 2024 was probably one of the most rapid changing environments that I've seen as being a VC in the last 12 years. You know, historically, when I started in this game in 2013, it was multiple years of SaaS. And you fast forward to 2023 and then into 2024, the rate of change and pivoting to AI or agent-based services happens so fast that the whiplash that we're all experiencing is really feels like once in a lifetime. So it's almost like Moore's law and how fast things are changing, not necessarily accelerating like compute. And so it's one of the funnest environments that I've ever experienced, but it's not slowing down. It doesn't feel like it's slowing down at all right now.
SPEAKER_06:Excellent. That's an amazing amount of data. That's great. Thank you so much. Um, so uh curious to understand how valuations were impacted in 2024. Um, and I understand it's a mix of you know, the valuations have become normalized, but at the same time, the AI market has been frothy. So I'm just wondering what you have seen from a valuation perspective.
SPEAKER_01:Yeah. You know, the overall data, you know, that you get from from Pitchbook on the you know, their venture monitor overall 2024 data is that valuations ticked up a bit um in 2024. And the percent of all deals that were AI deals uh ticked up a lot. So in 2023, overall cross-all venture was like a third of the deals were AI, and in 2024, like half, you know, said they're and of course, how do you define, you know, I was on a I was speaking at one of our company's sales kickoffs earlier today, and um, there was another uh board member on the panel with me, and he had a good, you know, he's like AI is like electricity, you know, it's in all of our companies. They all um so it's almost everything is AI. To your point about valuations, though, I would say, you know, my sense anecdotally is that uh overall there's been a reversion to a little bit more historical ways of thinking about valuations, you know, a kind of more rational, reasonable, not that it's founder unfriendly, but also that it's you know not so high that it seems like irrational. And how's this company ever going to grow into these valuations? So I think overall there was a continuing like rationalization on valuations. That being said, with some especially pure AI deals with really strong teams that just kind of go nonlinear and get super, you know, super frothy. So it was sort of an interesting, you know, market in that respect. Um, and but I felt overall that the valuation regime kind of stabilized, um, still kind of coming down in some ways from where it was in 21, 22. But there's certainly a lot of um a lot of interest in competition, particularly around certain AI companies. And I think where that netted out for the industry as a whole across all deals is that valuations ticked up a little bit. Here in Seattle, I still feel like we were where most of our deals are, not all of our deals. You know, we're up and down the West Coast now. We have a team in California, but still, you know, two-thirds of our deals are kind of Pacific Northwest. And for myself personally, I've been super focused on Seattle Market, which is a great market. Lots of talent, lots of companies getting for them.
SPEAKER_02:Yeah, I mean, you still did see a much frothier environment for AI companies specifically. We also saw that I think for, you know, cyber as well, which makes a lot of sense. I think the adversarial automated attacks and those environments are continue to increase and that space remains really relevant for enterprises. And then we saw, you know, more. We saw kind of those companies that may have struggled a bit more in uh 23. One either see some kind of soft landing or exit the market, or in their next rounds see, you know, a little bit more kind of uh favorable pricing as well, which I think is is good because a number of these companies have already built a tremendous amount of customer love and uh growth. And uh it's good to see that they can continue along to hopefully find uh an exit themselves or find that second wind uh going forward.
SPEAKER_03:I would say for the non-AIDLs, it was obviously much more metrics driven. I mean, the market has cooled off since 22-ish, right? So people are um funding companies based on again multiples, um ARR multiples, forward uh looking ARR multiples, revenue multiples, those kind of things. I mean there's a there's a range always because if it's a great team, you know, great space, even if it's a non-AI company, they still got premium multiples. But in general, I think enthusiasm was guided by multiples for companies that actually had real revenues and were not in the AI space. I think in the AI space in particular, if it's a really good team, it did not matter that you didn't have a single line of code. Still, you still were able to raise. Like I said, the valuations came down for sure, but still, if it were not an AI company, those companies probably would not be able to fundraise. But if it were an AI company with a really good team, not a single line of code, still were able to raise. And raised from you know, great firms, raised credible rounds and everything. For AI companies that were in revenue, I would say very frankly. So most of the companies we saw that had some kind of traction were mostly like I would call them POC traction, design partnership traction, some early paid uh traction for sure, but still very early. We did not look at any company that had astronomical revenue traction and would be called a POAI company. At least that wasn't in our set for 24. I'm sure there are like a couple of them out there, but that wasn't us.
SPEAKER_07:You know, there's really a tale of two camps right now. Sharish, you really have things falling under the category of halves, people that have been there, done that, been, you know, elite product leaders or business leaders at really important companies or return founders that are raising$10 million without a single line of code. Sometimes more. I mean, look at Elon on X.ai, although he has a huge advantage in C-Log. And then you have the other type bucket of founders where it might be a little less experienced or the pitch deck isn't rock solid, and they might be doing the exact same thing, but it are scraping by to get nickels. And so the camp of have and have nots, there's a huge bifurcation in the market. And I think that's a reflex of the rate of change that I just referred to, where things are happening so quickly. So if there is a movement happening, we are going to capitalize what we think is the winning team to go take advantage because you now can move so quickly. You don't need to hire a giant sales team or onboarding team. This agent world that we're now living in, change and adoption can happen with happen with way fewer people and happen way faster.
SPEAKER_06:Right, right. So looking now into 2025, what's your expectation of how the economy is going to do and how that's going to impact investments? I mean, you we now have a new regime and it talks about tariffs and deportations, et cetera, which could uh you know significantly affect inflation and interest rates and so forth. So I'm curious to understand what your forecast is for the economy overall.
SPEAKER_01:Yeah. It's a weird time. Um we in general are not trying to prognosticate on macro, you know, and kind of year-in-year-out basis. And our fundamental strategy is invest early, plan on working with companies for eight, 10 years, you know, you s live beyond, you know, up cycles and down cycles. Um, and if you keep doing that, you don't have to be so sensitive to, you know, the that being said, of course, we're all wondering. I I think that one thing that we'll see is that the that MA will open up some more, that it there was just a lot of restrictions in the previous administration, especially for the bigger companies. You know, the number of companies that Microsoft, Amazon, Facebook, Google were buying just had plummeted. So I uh I think broadly you'll see MA pick up. It picked up a little bit uh last year. I think you'll see that pick up more. My sense is that sort of enterprise and consumer spending, especially when we talk about enterprise, has stabilized and it's not, you know, roaring, but it's you know, people, companies need to invest in AI, they need to invest in security. We've seen that pick up and stabilize. Companies have gotten through their multiple rounds of rifts. We talked about that last year, you know, on this podcast. So I think those are all positive. Yes, there's you know, uncertainty around just geopolitical inflation, et cetera. Hard to say. I'd say overall, we're kind of going into the year as, you know, as business as usual from uh, you know, from how we're thinking about investing and how our companies are setting their operating plans for the year. It continues to be about efficient growth. You know, how do we push growth as far as we can while staying uh efficient and low burn multiples, thinking about when we have to raise our next round, you know, kind of all of those things around balancing efficiency and growth continues to be top of mind for companies doing their operating plans for 2025.
SPEAKER_02:Yeah, just from a market perspective, I think we're I think the broader market in general, this is not uh me alone, uh, is anticipating uh, you know, an improved uh environment for IPOs. And I do believe also it will also increase just because there is a considerable amount of cash on the balance sheet of some of these large tech platforms. And as their market values have continued to rise with the advent of AI, I think there's appetite to you know make those kind of cash acquisitions to accelerate their roadmap and continue that focus of priority. The broader macro, I think there remain a lot of uh questions and uncertainties, as I think we're expecting a number of different changes on that front. So I'll I'll just say that from the market perspective in terms of how it's impacting investments. Look, I think the opportunity with uh, you know, a healthier market, even on the public market side, is that that filters down to enterprises and how they spend. And that really is, I think, a leading indicator for some of the startups that are starting to see, you know, post-product development, uh, starting to sell in the market, their ability to acquire customers and uh their ability to sell into customers who are willing to purchase best-of-grade solutions versus platforms, it becomes much easier. And so uh we I'm hoping to see that acceleration really be a positive tailwind for a number of enterprise uh startups. Um and so from that standpoint, um, you know, the the challenge then becomes how do we make sure that we're investing in those that are mission critical and will endure uh given the number of innovations that are kind of uh behind them or that they're building up on the other thing.
SPEAKER_03:So that's that's kind of the particular so I mean I'm I don't know if I'm um if I'd be the right person to even answer that, but I'll tell you like a little bit of uh I'll I'll cheat there a little bit uh because I was recently on another panel where there were a lot of public investors in the audience and they actually did a poll uh on like what they thought where the economy was going to uh go into 2025. I think the consensus was moderate growth. And when I say moderate growth, the range was 10 to 15 percent growth. And if you really look at last couple of years, I think the growth uh consensus was a little lower than you know last couple of years. It wasn't like oh super bearish, uh, but it wasn't super bullish either. From my standpoint, because we uh at Proven Capital, the way we work is we we are an independent fund, but we the value we bring to the table for founders is really helping them sell into large corporates. So as a result, we are constantly talking to the CIOs, CSOs, CTOs, CEOs of large Fortune 500, 2000 organizations. If I kind of summarize what we are hearing from them, is it's not like uh, oh, we are going to slash our IT budget, you know, 20%. That's not what they're saying at all. But what they are talking about is efficiency. So for them, efficiency is a big part of 2025. So uh everyone that I can think of within our network, and we have hundreds of such corporates, everyone is investing in some kind of automation technology. Everyone is uh thinking that um, you know, a lot of the processes they have that use humans uh could probably be personally automated or fully automated. But at the same time, they are not thinking of like cutting their human uh workforce either. They are thinking of you know finding other more valuable work for their humans. So I would say by and large, the corporates we work with, they the message from them has been we have cut deeply enough to get efficiency out of our human capital. Now is the time to get efficiency out of automation. So that's really the focus of 25.4, I would say, in general, from what we see.
SPEAKER_07:You know, I think the jury is still out because there's a lot of things happening on what the inflation print will be 12 months from now. You know, some some argue that hey, the tariffs are gonna be highly inflationary, others argue that it's just writ reciprocal for you know other countries applying tariffs to us, et cetera. And so I think the jury's still out. I think there's a lot of enthusiasm from the business community that the deregulation is gonna spur uh a lot of economic growth, that there's going to be easier to conduct transactions. You know, you have companies. That want to sell, but don't want to sell right now because of the current administration and the current FTC. And so you saw 2024 as a waiting game to see what would happen with LenaCon, with the existing administration. Well, now that we're entering 2025, there's a lot of enthusiasm that deal making will happen, that capital deployment will really happen, that the US, albeit with a very mismanaged budget deficit right now, is still the best house in a bad neighborhood in terms of the best place to invest in the world. So I think that will continue, but I think the deal making will really ramp up, particularly in MA, where there's just a lot of companies that are on a clock that need to find outcomes because every time we raise a fund, you're on a timeline. So a lot of these companies are now on a clock. You have a deregulatory environment. And I think hopefully rates will go down. Hopefully, D. The confluence of all those things should make for a pretty exciting 2025.
SPEAKER_06:So again, looking into 2025, uh, what are going to be the focus areas for your investments? Is it going to be primarily AI again? AI data infrastructure?
SPEAKER_01:Yeah. Yeah, as we we touched on earlier, from a macro theme sort of a megatrend standpoint, um, applied AI and intelligent applications. You know, there's been a lot of discussion around uh AI agents and agentic applications. You know, nobody even sort of knew that agentic was a word two years ago. And I think that's real. You know, of course, the big Salesforce and others are really banging the drum and turning into kind of a marketing, you know, food fight around agents, but we think that's very real. We are seeing using agent-based architectures to rewrite applications and write and envision new ones. I think this won't play out fully in 2025, but over the next, you know, two to three years, you know, agent-based interfaces really, I think, reimagining a lot of the way we interact with software. You're hearing, and we we wrote a post recently, my colleague Soma and Sabrina wrote a post about SaaS is dead. And I, you know, that's provocative. And people, what do you mean? I think embedded in that, there's a couple of concepts that you'll see play out further this year is that the again, the user experience, the way you interact with software, you you go to a web page, you go to a forum, you type in a bunch of things, you navigate screens. I think that will more and more be, you know, instruct your agent through some sort of natural language to accomplish A, B, and C. And I think the way you pay for software is going to evolve. This won't change overnight. But instead of, you know, based on seats and things like that, you'll see more how can companies be based on outcomes, based on some version of ROI. So we're going to invest heavily around these agent-based applications, as well as we've really thought through, you know, what does it mean for an agent-based architecture? Where in that stack uh do you need uh new companies and new tools? Where will you use existing companies and tools? Where will those tools come from, existing providers, where will the new companies need to be created? Um, so I think that will be a big theme uh going forward. Uh last thing, so I do think there will be more. Um, we already we talked about this last year and we saw it in 2024, the the fragmentation at the model layer. You know, it's not just open AI, open AI is amazing, but you see a set of other, you know, mega model providers from Anthropic to Llama to uh Mistral, et cetera. Um I think we we will see more and more open source, more and more domain-specific models where companies or organizations they have their own data set. They can start with something in open source, they can build and train or fine-tune a model to fit their specific need with their specific data, and then insert that into whatever their business process or insight is. I think that will continue to become more and more prevalent and allows you to make investments in, for say, let's say vertical markets, you know, where here's a new application, new model, it's allowing a new set of insights or workflows around a vertical market.
SPEAKER_02:Yes. In the short answer, yes. But more specifically, the things that I'm personally really excited about. One is really as AI is getting adopted, I think there were a number of day two solutions that were emerging in the past couple of years that were anticipating uh, you know, a certain level of enterprise adoption. I think those are gonna start to mature. And I think it's really exciting to see the companies that go are able to go after um, you know, the monitoring space, the security space, and so forth. I think there's gonna be an evolution of a lot of those companies that emerged initially to tackle those problems, and then now new companies that are going to take some of the existing assumptions and build upon that. Next, I think there are like some spaces that are super uh ripe for disruption within the cyber and DevOps spaces as well. And that's really with the opportunity to combine LLMs, use some kind of uh deterministic modeling to ensure that they are going to be viable in critical workflows, as well as leveraging agenda capabilities to actually complete tasks versus only assisting. So I think that's the opportunity, and I'm excited to see that be really disruptive in the infrastructure layer. And then lastly, an area that I'm super excited about is just how AI can be used. Again, leveraging some of those similar capabilities to be disruptive with some of the system integrator platforms as well. I think the services market generally has been a core focus area for AI investors because we see the opportunity for LLMs to be highly disruptive in that regard. But specifically in thinking about uh the kind of core technology tools of Oracle, SAP, a lot of these that have that are embedded within uh traditional enterprises, the services that surround the services ecosystem that surrounds those types of tools is massive. And I think we hopefully can start to see some of those spaces uh start to get disrupted as well. And I think that's incredibly exciting uh to see that change and fold.
SPEAKER_03:Well, if I didn't say agents, you guys will throw me out of this project. But I'll go I'll I'll start with something that is a little out there because it is actually something uh we are pretty excited about. And I'll say in many ways, it's actually beyond agents. And what we what I'm really referring to is more like service as software. And we have a very strong thesis, and we have been working on this part as a firm for the last year plus and made some investments in the last few months as well. But we kind of look at every corporate partner that we work with, we look at their budgets, IT spends, and typically every company is spending more than half a billion dollars every year on cognizance, TCS, Deloitte's, Infosys, you name it. And it feels like you know that business model is stuck, you know, way um, you know, at a very different era, which is like very people-driven. The efficiencies are obviously annoying, but I think much more annoying to many of these corporates is the time it takes to turn around a project. If you're building an integration between system A and System B, it takes like two, three months just to get it back through these SIs, you know, take all the design, building, testing, all of that. And then maintaining that is also a pain. So, long story short, I think there is a fair amount of opportunity to disrupt some of their businesses. Not all of it. I wouldn't say that. I mean, they have very deep relationships with all of these corporates. It will, it's a multi-decade journey, but I think technology has gotten to a point where it's actually not unreasonable to think that a new services firm can be built that will still look like a services firm to the outside world. They will still have people delivering the service. But under the hood, it will be powered probably 80-90% by software. And we are seeing that trend coming up in different areas that traditionally has been served by these SIs. So we call it service or software. We have an inside joke saying we are going from the SaaS world to another SaaS world. But I think that's actually something is a pretty strong focus for us as a firm. And then, of course, everything else that is going on with you know, will SaaS be dead? You know, will agents take over the world? And there's definitely time we are spending there. But I'll pause there for any further question. But again, like I said, I mean I wanted to throw out something I haven't heard as many investors talk about or being focused on, which is like, okay, instead of thinking about agent tech stuff or copilot, which is even the previous generation, can you deliver the whole thing as a service?
SPEAKER_07:The pace of investment, so the focus, I'll address the first question. We're playing the game on the field. Traditionally, we are software investors through and through. Historically, like I mentioned, that was SaaS and seat-based pricing. Now there's just a different way of pricing and delivery to some extent. And we're really enthused about the vertical applications of AI. You know, you're seeing scenarios that could never have been solved with software now being completely unlocked. Things like claims management. You used to work, have claims agents and insurance companies have to review all these documents, hundreds and hundreds of pages. Now AI can enable you to review the claim and all the appropriate data and important data in minutes. Things like freight forwarding, where a freight forwarder has to deal with shipment documents from 15 different counterparties. AI can now tell you where there's errors in your shipment documents and where you might have an issue with customs. So things where there's a ton of paper and a lot of, I would say, cost center people are now seeing huge unlocks with AI doing better, faster, and cheaper work than an actual human can do. So very vertical domain-specific solutions. Where we're also seeing a lot of really interesting opportunities is actually in space. And this is a new fuse development, really, over the last year. And a huge shout out to my partners, Brendan, and then Nolan, then Nortwick, who have really dedicated a ton of energy and output into finding really interesting space companies. And we're really excited about the next generation of applications that Sprache will bring beyond connectivity and more around compute, like data centers in space with Lumen Orbit, and then around management of satellites to provide services to the terrestrial world. Got it. Thank you.
SPEAKER_06:Over to you, Gabri.
SPEAKER_04:Okay, great. Tim, so you talked about quite eloquently about agentich applications, right? Let's just kind of pick on that theme because that's going to be a key focus of your 2025 investment. So you talked about uh architecture, interfaces, how we interact with them, and then business model, SaaS is debt, right? So in the previous versions of those software, either the business model drove the architecture, the architecture drove the business model. Now you have the business model dependency on bigger model providers, right? There's an underlying you have agents with the architecture on top, then you have the interfaces that is uh that you're making it easy. So your investment thesis, getting back to that, right? Are you focused on an agent application uh which is here now in 2025 that you see will take advantage of one or the other? That is, either the business model is somehow clearly measurable, so you can extract the value out of it, or the architecture is so flexible that you can build a business out of it. Where in the Madrona Tim Porter mindset is this sitting?
SPEAKER_01:Thanks, Gary. Um, it's interesting, uh really interesting question. I our mindset is always to try to work back from the customer. And uh in my our partner Matt did a post last year called Um, you know, something like the most important model is the business model. Um and I do think sort of working backwards from the customer need, which may mean business model innovation, is the right, is the way we're gonna evaluate new opportunities, not working forward from like, hey, this is a super flexible architecture, or this is I think you get into a risk around being too um, you know, tech driven that you've got to be customer need driven. But I that being said, I don't think in all cases it will be, hey, this is some radical new business model, or that everything is going to flip to paper for performance. I think in certain areas you'll you'll see that uh or in certain cases. Um I don't mean to make that a little bit of a non-answer. We're we're we're always trying to look for the intersection of technology enabling some new customer problem. And the business model is usually more important than the than the technology architecture there. So that's that's where our thinking starts. Sometimes the way you go find those opportunities, though, is to try to follow the technology innovation, right? And then think about like, okay, what's the business model implication? What's the customer implication for this? Whether that means, you know, going to conferences like Neurups and reading the papers and hearing what the new things that are coming out. And that's part of the role what investors are supposed to be trying to help with, right? Is help, you know, founders think about like, okay, how do I apply this, you know, this new breakthrough that I'm working on? But if they had to pick the answer to your question, we would work from the business model and the customer, not from the technology first.
SPEAKER_02:Yeah, I think what we saw in 24 was uh a number of different um modalities come out. Uh we have like pure play, you know, agentic development platforms uh that um have been you know starting to uh penetrate the enterprises. Um, but we also have seen exit like large SaaS companies and other more vertical or end user-facing applications leverage agentic uh capabilities as well, which I think now is going to be an interesting moment to really see which of these types of forms of kind of agentic capability is going to win in market. My uh early hypothesis there is that having those uh frameworks embedded into existing tools that provide a system of record are going to be have an advantage over those pure play, just because they're providing their time to value is going to be more immediate. But in the long term, the pure play agentic platforms should have an advantage because they should be very sticky, because they can be customized uh for exactly the uh enterprise need. But we'll see. I think this is gonna be an interesting year where we should hopefully see some of those that battle play out. Um, I also think that the creator economy and the developer ecosystem has is like continuing to is starting to come back in full force. I think we saw as the market kind of was more quiet, a lot of the developer swipes of credit cards had been closed off. And I think that's gonna start to emerge because experimentation um, you know, is not just top-down, it's also bottom up. And I think enterprises are trying to figure out what those kinds of opportunities are as well. So excited on all those fronts.
SPEAKER_03:Yes, it's a really good question. It is, I would say a services software is the next evolution after agents. If you really uh I actually saw this presented by some company which I thought was really good. I think if you really think about evolution, we started with more like data retrieval search kind of capabilities when it comes to AI. Then came the co-pilots. Okay. So it's basically a tool that works, basically you can do work with, right? Then came the agents, which is AI that actually works for you. Basically, you tell it to do some task and it goes and does it for you. And then the service as software pieces like that AI that completely gets the end-to-end work done for you, right? So that's kind of the ultimate evolution. Now, agentix solutions are very much core of any service as software offering. So like I said, it's a very much an evolution. So our thesis is uh there will be a ton of use cases where agentix solutions will disrupt whatever SaaS or multitude of SaaS SaaS tools you are using, right? But then if you take it to the extreme, you could say, well, those agency solutions could be part of the entire service offering that can be delivered to you as a work product. So to answer your question, it's really an evolution. And then we are barely getting started with agent solutions, frankly. Like uh it will be uh again, if agents work out. I think there is an if, by the way. I mean, I don't think it's a given that okay, all of this will work out seamlessly and it'll be so easy to displace, you know, a decades old habit of using SaaS and so on. But uh and also you need to figure out the UI for agent solutions, big, big you know, part of it, right? But if they work out, I think enterprises will be locked into Agentique Solutions for the next three to five years.
SPEAKER_04:Um you do feel they will make an appearance in 2025 in reality, right?
SPEAKER_03:In reality, yes. Again, I would say uh we are already seeing some of that actually. It's limited in terms of actual production use cases. I think the hype always proceeds, right, with all of those things, right? So um every corporate we interact with has some kind of effort going on around agent TI, all the way from the CEO, CTO level. So we ourselves have spoken to multiple such what I would call corporate-wide, like innovative ends or brainstorming think tanks, you name it, right? Small, large, small kinds of efforts. But agent tI, everybody is sitting up and thinking about okay, could I get a lot more efficiency out of like spending you know on 500 different SaaS tools, which I may not be utilizing as much as I think, or as much as I'm paying for. So that's what I think a lot of the agentic AI uh interest is coming in, which is like efficiency and speed and cost of those.
SPEAKER_07:I I think a lot of them are already here, to be honest, Gary. And I think what you've seen in some of the early portfolio companies from fund one have already moved towards agentic workflows and applications, like I mentioned with the example with claims management or fraud detection with OWL. And so a lot of these companies are really refactoring their software and their pricing with with more AI for very vertical specific scenarios. Where you're not seeing us spend a ton of time is in the foundational model game. It requires so much capital, so much compute, so much money to train these models. And they're very, very necessary to unlock these applications. But with a smaller fund like Fuse, it's really hard to play that game where you see, I don't know, Anthropic's gonna raise another 2 billion. You know, that's X the size of our fund just in that round alone right now. And so we've stayed a little bit away from the foundational model game, but we see them similar to that of the web browser to enable these new apps. You know, the web browser enables SaaS, agentic applications or the foundational models are really the web browser to enable these vertical agentic applications. So I always like doing analogies or comparing the new generation versus the generation of the 2010s, and that's how we're we're we're really thinking through some of the investments that we're making.
SPEAKER_04:That's a great analogy of uh web browsers because all the applications that were built based on the browser being stable and available, right? I mean, they it's just everybody said, Oh, go to your browser and do it. It took a while to train people what what that was, but once they got it, it got there. That's actually an excellent analogy.
SPEAKER_07:Uh yeah, look at look at look at our our uh video conference right now. It's enabled by the house.
SPEAKER_04:Exactly. Exactly. Got it. So one uh follow-up, which is uh you mentioned you you have opened up AI bio, right? Historically, Madrona has not necessarily wavered from its technology focus. This also is technology, but it has a very different uh you know underlying science to it, so to speak, right? What's the history a little bit and uh as to why you guys chose it? Second is how do you apply the 30-year learning to a new completely new sector, which eight years seems like tomorrow in a drug discovery game, right? So it's like a 10-year, 12 years like these are long cycles with a complete one-zero failure at the end of 10 years. So uh if you could just elaborate on that, because that's so unique at this juncture in your history.
SPEAKER_01:Yeah, thanks. You you we have the benefit of being friends for a long time, don't we, Gary? We were so you you know the history and and kind of where where where we started and and where Madrona started. This is a great topic, you know. And in fact, uh, if if you and your audience are interested in in biotech and AI bio, you should have on one of my colleagues, uh Chris Picardo or Joe Joe Horseman, who can go really deep in this area. So I I don't spend as much time here. On the other hand, we're all one team. You know, we have one investment committee, and so um I'll I'll tell you our my thoughts and our strategy on this. So our first investment in in the AI biospace or the applied biotech space was actually a number of years ago. It was not Which uh the co-founder and CEO is Sujal Patel, who was the founder and CEO of Icelon, who is one of our big early winners in the storage space. And you probably know that company well. And that was that was all the way back in 2018 when we made that investment. So this isn't like a brand new strategy, but now over the last two funds of ours, so call it the last five years or so, it's been about 15% of our dollars have gone into this space, kind of life science. And we we started out talking about it as the intersection of innovation where cloud and AI meet uh life science and biotech. So, how do we learn our way into the space? First, is trying to pick opportunities that were really IT breakthroughs, ML breakthroughs, using the cloud, using data, et cetera, as opposed to laboratory science breakthroughs. We have partnered closely with institutions here in, especially in Seattle, that we had long-standing relationships with and had a real trust-built relationship. So, you know, Matt was the chair of Fred Hutch, and we've done some spin-outs out of Fred Hutch. We've done a number of spin-outs out of Institute of Protein Design at UW. We've done other things with UW. We've done other things with the the two Allen Institutes who've done some things in biotech and brain science, et cetera. And then we've worked with co-investors who are, you know, live, breathe, and are wholly focused in in the biotech area. Uh and we've worked with founders who we've came to know and trust. And so we didn't just start sort of plying the globe, you know, and trying to go bid on, you know, you know, other technologies and companies that we knew cold. We wanted to get in early, partner closely, and stick to areas that we felt like we understood some of the key success factors around. And we've really liked that strategy so far. So we have a number of companies, a company like Terray just raised a big new round from a growth investor and an NVIDIA. They're using AI to do new new types of drug discovery. We've done, as I mentioned, a number of things out of IPD. We had David Baker who won the Nobel Prize this last year. He spoke at our annual meeting a few years ago. So that's been the way we've got into it. And then last when I mentioned have Joe come out. I mean, we have a people on our team now that this is all they do, you know, and and Joe has a PhD, you know, in in this area and worked in biotech. And then there's some others of us on the team who had worked in this area but also worked in other areas of investing. So we try to bring the right, you know, team members to bear to make good decisions in this area.
SPEAKER_04:Right. Looks like the fundamentals have been chained. Learn the work with the founder, know about the founder, and grow with the founder. That's same whether it's bio or technology.
SPEAKER_01:Absolutely right. But to your original point, though, that the the milestones and the timing is different for these biotech companies, right? And sort of um we've all we've tried to be, you know, milestone-based around, you know, and I don't mean specifically like structuring and tranching investments, but just thinking like, all right, we're the pre-seed investor. What can this company actually solve to raise the next round? What do you have to solve to raise the next round? Being cognizant of, you know, at some point the exits for these companies are licensing or selling to larger pharma companies who then actually take it all the way to market. So there is a different set of factors in thinking about the company lifecycle, but the fundamentals around differentiated technology, founders that you trust with high integrity, attacking big markets, important problems, those are all exactly analogous to what we've always been doing.
SPEAKER_04:Got it. Switching topics to another sort of a bigger trend, the IPO market. There were some, you know, openings in 2024, but not, there wasn't a gush, right? And some triple. Did you guys uh did any of yours go public in from your portfolio last year?
SPEAKER_01:No, we didn't have any IPOs. We had a few very nice MA exits. Um, the biggest ones for us were Rover, which had been a public company and then was taken private by Blackstone. That was well over$2 billion. It was a company called Jama Software that was bought by Francisco Partners for$1.2 billion. Lexion was an AI deal that was bought by DocuSign. So we had some good liquidity and good MA, but we didn't have any IPOs. And you're spot on. The IPO market opened a little bit in terms of the number of new issuances. There was some good IPOs that happened, but there wasn't a high volume of them. You know, Service Titan here, right at the end of the year was a good one. You know, there was there was a chip company, a semiconductor company a little bit earlier in the year called Estero Labs, but there were there weren't a ton.
SPEAKER_04:So just to follow up on that, the are these private equity deals the modern IPO? Do you see more of that in 2020? I mean, the exits are enormous. I mean, they're not, I mean, two, three, four billion, too, right? They're they're capable of assembling any amount of capital. So without the regulatory hurdles, you get to enjoy the exits, so to speak, right? Not commenting on whether the inventory is bad or good. Just it's the truth of what it is.
SPEAKER_01:Yeah, it's a yes and no. There's certainly more of them, and the scale has gotten bigger. Um, and I think you're gonna see more of that activity in 2024 and going forward. A different related dynamic is private-to-private mergers, which I think you're gonna continue to see more of for a few reasons. One is that you have a bunch of companies that are at sort of medium scale that raised a bunch of money and now are trying to figure out, you know, what to do. And back to the IPO market, it's not that the IPO market is closed. It's just that the scale you have to get to to have an IPO has gone up so much. You know, back, you know, 10 years ago, it could be you're at 100 million, you know, 100-150 million, you can go. Now it's 300, 400 million. It's just, you know, big scale. Um, and then, you know, the the companies don't like the valuations. The multiple. So, you know, you could go public. You just, you know, if you're Databricks, you could go public, but you like the valuation, you just got in the private market and you can, you know, raise, you know, an enormous amount of capital in the private markets at better prices and then not have to report every quarter. So, you know, when you have companies like Stripe and Databricks, they could certainly be public companies. They just choose not to be. So I think in this year, I think you will see some more IPOs. Um, I'm not sure that you're gonna see a gusher, you know, of IPOs coming back, but there are definitely a bunch of interesting companies that are, you know, in some some degree of filing, and you'll see some of those come public in in 2025, whether it's, you know, more people um trying to cash in on the, you know, on the NVIDIA semiconductor, all of things happening around, you know, AI chips and semiconductors like uh Core Weave or a Cerebrus. There's companies in categories like cybersecurity, like Sneak, who's, you know, I think they're at the right scale and profile to be an interesting public company. You really wonder the Databricks, Canva, Discords, I don't know. I I think you'll you'll not see, maybe you'll see one out of that type of group come public this year. Stripes, Klarnas, uh, etc. So I think there's a great backlog, and we'll see incrementally more, but I don't think it's gonna be like a a run for the a run on the IPO market.
SPEAKER_02:Yeah, I think there is a broader market sentiment that it is um a more favorable environment for exits and IPOs. So at this moment in time on uh January 9th, hoping that uh you know that remains true, but uh we'll we'll see.
SPEAKER_03:I would like to think so, yes, especially the second half uh of 25. Yes, short answer. Um I was really expecting uh, I mean, we know all the unusual suspects that should go IPO, could go IPO, um, like the stripes and the data breaks uh of the world. Uh I was expecting actually Databricks to go IPO, probably. Uh I have no insider knowledge, but I was expecting them to go IPO in the first half of 25. Could still happen, but then of course, you know, they just raised a giant uh funding round. 10 billion all said said and done. So I I don't know anything better, but it feels like a lot of these companies that have been very successful, they might go IPO for a different reason, which is to give liquidity to their employees. And of course, if I'm in the right time in terms of interest in AI, interest in you know just interest in like really next generation companies from an investor, public investor standpoint, I might actually take that opportunity when the volume is light, as opposed to you know having to compete with five other similar companies. So my sense is companies that are really ready probably will uh target second half of this year. Will it be a frenzy? Absolutely not. I don't think so.
SPEAKER_04:Okay.
SPEAKER_03:That's that's so about the same, you're saying, as uh, some of these iconic companies I can see them go public late this year, I think. That's my you know, like very foggy crystal ball prediction.
SPEAKER_04:Yeah, my opinion is that the IPO was viewed as an liquidity event, but the market's uh sentiments are lost in value creation, right? I mean, look, there's a whole lot of movement in the top, you know, five or six companies that garnered all the value because the money going to public markets is still available. I mean, I understand PEs are giving the valuation, but there's something said to be gaining it from uh other investors too. But it'd be good to watch, I think.
SPEAKER_03:Correct. And you know, um, some of these companies that have the luxury of going to the private market to raise you know enormous amounts of capital. If I were running one of those companies, it's uh it's probably slightly more comforting from a strategic standpoint, right? I can make much longer-term bets without having scrutiny of quarter by quarter running skulls, right?
SPEAKER_07:You know, I made the mistake of telling my wife that I certainly was going to go public in 2021, and so I've lost all credibility. It's like where where where are the dollars at? Well hopefully one day. Yeah, okay. But I do think that the there's a lot of again, it kind of goes back to the clock that VCs are on, right? Where uh and and companies are on, or what's the maturity, what are the what's the profile of these companies? What do they need to look like to be IP already, where they get tier one investment bank coverage to where they're gonna rise in the markets and the markets are stable enough to buy and hold? And I think, you know, and I'm hopeful that we're entering a year where there's a lot of stuff gunked up that's gonna get ungunked up in the second half of the year. Now, the thing that concerns me a little bit is there's too much supply relative to the demand of the public market investors, and that's all got to get sorted out.
SPEAKER_04:Is that the reason why you see private capital buying up companies at a fairly large valuation? And it seems like they're going public, right? You get a$2 billion buyout from a PE firm of, you know, or Databricks not wanting to go public because their valuation is better than the IPO market might have given them.
SPEAKER_07:Yeah, and and there's a lot more comfort from Thrive, which you recently did that with both Databricks, I think, and OpenAI. And I think there's a lot more comfort in these transactions that are not necessarily on balance sheet, but they're to relieve stock options or provide liquidity for employees. And you know, there's I think the public markets make companies operate better. You know, that's obviously a very debatable point of view where, you know, some founders say, well, it's harder to innovate if you're a public company because people are short-term focused versus, you know, the ability to invest in the long term. You can see both sides of the coin and debate it six ways till Sunday. But I think public markets make companies operate better. Maybe a little harder to innovate, but certainly operate better. And so I think there's a natural um, you know, if you're a founder and you're in the public market fishbowl where you got six billion eyes on you every single day, yeah, that's a little daunting versus here's my board of directors and my inside investors and employees are the ones I need to appease. And so it's really getting over the psychology of getting there and going public versus taking these later stage rounds. Um, you know, historically, and I think in 2020 and 2021, of course, with the crossover funds, there was just way too many late-stage deals that got done. And it just kicks the can down the road before you go public. Well, then when the water went out or the tide went out, you know, you got to grow into that valuation that's anchored in your mind. So a lot of founders won't want to go public until they hit their last private valuation high water mark. And I think three years later, four years later now, going into this year, three to four years later, a lot of these companies are finally probably getting there. So I think the psychology is now more acceptable for founders and boards as you're getting close to the most recent high water mark.
SPEAKER_04:It could it could also help that the, at least at the moment, seems like the tech community may have a seat at the table of the new administration. So it could help in the regular regulation of the of uh how tech um advances it takes itself, hopefully. Okay, in in your true fashion, in 2024, the prediction was we see investment dollars, number of deals, MA dollars will increase in 2024. How do you feel about that prediction now that you're in January 2025?
SPEAKER_01:I was sort of right, but not like wildly right. You know, I I think you'd have to look at the numbers and say it was mostly consistent. They were up slightly. You know, I think technically I would right, but I wouldn't claim victory on my predictions. Yeah, right. I I think the overall, you know, capital raise was up by about 20%. The MA dollars were up by about 20%. The deal counts were pretty consistent between years. And so the thing that increased a lot was the percent of deals that were AI driven. And you started to see some of these series B C kind of later and growth stages start to come back, you know, and that's why the overall dollars went up and the deal count stayed about the same. You know, that kind of growth market, the series B C D market from, you know, middle of 22 through the middle of last year. There's basically almost nothing happening. And, you know, even in our portfolio, we had we saw more interesting companies in that range, you know, in the second half of last year than we had seen before. And I think there are a few different dynamics around that, which are going to continue next year. One is that you've got companies that were like born in this AI generation, born kind of in the chat GPT plot that are like getting to scale. Like they're growing super fast, and now they're rising series Bs and things. And so those are the the most exciting, most competitive deals. And then you've got a set of other companies who maybe they live through, you know, the the the downturn in in 22 and they got efficient. You know, they got they got efficient, they got their business, and so those companies are raising, and it's also far enough away from probably silly valuations and multiples that they either grew into those valuations or they're willing to take some sort of reset valuation. So you're seeing both of those things, I think, drive the the growth stage market. And I think we'll see more of that in 25. I don't know that the early stage is going to get like even more busy. It's been busy, it's gonna stay busy. I think it's kind of more the same in the early stage side, and I think you'll see an increase in activity in the the growth and early growth stage.
SPEAKER_04:Right. So in 2024, just to uh uh for remind our audience, uh you said there'll be a couple of enterprise IPOs by the end of the year. That was your 2024 production. So was there?
SPEAKER_02:Uh yeah, we we saw uh Rubrik and uh Service Titan um uh go out to market. I think I uh and I anticipated uh Databricks would go out, uh maybe in 25. That's now uh what you know is expecting they're one of the many in the backlog. Uh so we'll see. But uh yes, I was uh I'm glad to see that we actually started to see some uh enterprise companies come to market.
SPEAKER_04:Okay, there's the moment of truth. Okay, this was your prediction in 2024. There will be I told you we'll bring it back. Uh there will be an open source model that'll give open AI a run for the money. Has this come true? And what do you think uh the status of open source models are today?
SPEAKER_03:I think I did pretty well. I'll say that.
SPEAKER_04:You did, you did, I would agree with you.
SPEAKER_03:Well, I think back in 23, I was actually looking at some data on this, but back in 23, I don't think there was any model that was in the GPT-4 class.
SPEAKER_06:Uh right.
SPEAKER_03:In uh if you take the leaderboard data that you know publishes all the ranks all the models, I think end of 24, we have 18 other models that can give OpenAI a really good run, and some of them actually have exceeded. So take whatever the Lama's you know latest iteration is, and then the model that really uh shocked a lot of people was uh Deep Seek's uh model that they dropped uh I think the day after Christmas actually. And that one uh was as good, and I think in a couple of benchmarks, uh it was slightly better than GPT-4. I think the most shocking piece was it was trained using less than six million dollars. Um so yes, I was very correct. Like that's the world we are headed in, right? It's like proprietary closed source model, uh, advantages basically. I wouldn't say going away, but it's leveling up.
SPEAKER_04:Yeah, yeah, yeah, yeah, yeah. Totally true. Because nobody wants to continue to give all the money to NVIDIA. So, uh moment of truth. You made the uh following prediction for 2024.
SPEAKER_05:Oh no, what did I say?
SPEAKER_04:Uh I said, beware of the incumbents. Now, what did you mean by it? And do you think it has come true?
SPEAKER_07:I think it uh yes, and probably more so than I probably would have anticipated with Microsoft with Amazon's Nova announcement recently. You know, if I was if I was anthropic, I'd be like, um, excuse me, uh now we're competing with an internally built tool that everyone is probably mandated to try both. The whole purpose of that is to improve the the the Nova model to outperform the anthropic model. I'm sure of it. No one's told me that those are Kellen's words, but I'm sure that's the point of view is choose both, but make sure you try both so we can improve Nova. And so I think you look at where Microsoft's positioning, where Amazon's positioning with their data centers, with their compute power, with their big advantage, I think their market position is excellent and improving fast.
SPEAKER_04:Yeah. That's a that's an excellent uh way to look at it is how are they tackling it as incumbents, how are they tackling it?
SPEAKER_07:So even with Salesforce, with you know, it's no longer Dream Force, it's Agent Force. I know. And and then with you think about the huge advantage that they have, you know, with CRM software and all the amount of text and semi-structured data in a CRM. And it's a really interesting uh I was having a drink with a really close friend um a couple months ago, and I won't say the name, but you know, I asked him, I said, what's been the biggest surprise? You know, your firm got acquired, what's been the biggest surprise to the buyer in a positive way? And and you, and he says, the quality of our CRM in today's AI world. And what he meant by that is when they moved completely remote in 2020 when COVID hit, then they came back to the office. If a meeting was not entered in the CRM, it did not exist, and you were not confident. You know what was competent meetings that were not included in the CRM. So the discipline of the CRM hygiene ended up helping them build one of the most important data assets of what the strategic buyers and the private equity buyers are most interested in in terms of assets. And now that they have that huge database of meetings logged, that's really, really important that they can now query and surface up insights for an MA deal. So just, you know, it was a really interesting insight, really interesting insight in a non-tech company.
SPEAKER_04:So the reason we are here to hear from Tim Porter, the prediction of 2025. What's your prediction, Tim, for 2025?
SPEAKER_01:We to we've touched on you know a few things. So I'll summarize a few things. You know, I think you're gonna continue to see uh heavy pace of AI investing, especially around AI agent-based applications, domain-specific models. I think those that will be those will be dominant trends. I think you'll see a pickup in the the series B C D you know, investments. I think, you know, one thing that I I'll maybe go out on a limb here a little bit, and I think you'll see a billion dollar plus exit in Seattle. Um, I'm not saying that because I have any inside information on some company that's in a process, I'd swear on that. I just think that you know they're maybe that's part of you know more MA happening. We'll we'll see. But I think I think there'll be another big exit in the Seattle area uh this year, which would be which would be great to continue to see.
SPEAKER_03:You guys, you guys are pretty, I don't know. I don't know, I don't use the word publicly I was going to use it, but anyway. Um see here's the thing. I I do think um and I'm copying out a little bit, there's a little bit of a cop out answer, but I'll say this. I do think that you wouldn't see any uh new model that has like a huge number of uh like let me put it this way if there was a model out there today with 750 billion. parameter, you don't see one with you know three times that this year. And the reason I'm saying that is I feel like, and again, uh obviously it's coming from uh uh a lot of the AI experts. I think that the advantage and also the data that you need to train uh another gigantic model that is two, three times the size of what you have, that that slope is plateauing. So you wouldn't see a gigantically bigger model in 25. What you will see is a lot of this AI compute and use cases moving to the inference side of the house. And the reason I'm saying that is within our portfolio companies, when we ask around like, hey, what LLMs you're using for your application, more and more the answers are coming back as 01,03 inference reasoning kind of models because they are building, they are basically solving enterprise use cases. And for them, the bigger is not always the best answer. It is much more about accuracy and of course cost and all of that. So inference the way it delivers accuracy or delivers much better accuracy than one big model delivering a one-shot answer. For them, that is important and I think you will see more and more compute moving to inference. It's already happening by the way. But you will see the plateauing of training on large models, building large models and you will see a lot more AI compute moving to inference by the end of this year.
SPEAKER_07:Okay, a couple. How about two? And they're gonna be way more specific than beware of the incumbents because that's vague that you can you can spin that to be right no matter what. Yeah Amazon buys Instacart. DoorDash is too expensive. Amazon needs to have a a better captive network flexi's worked out okay but I just don't know anyone I just don't know much about it. It doesn't have a ton of market awareness. Instacart does it's trading at what$11 billion it's not$75 billion like DoorDash. So it fits in kind of that strike zone like Whole Foods where it's going to get them further extended into their delivery network. That's one wow gold I like it there's both it's gonna be it's binary it's either it happens or it doesn't. So you could say you were wrong. Here's one that's slightly less binary Microsoft detaches begins its detachment from open AI because they don't need open AI near as much as open AI needs them anymore.
SPEAKER_04:I think Microsoft is going to advance their models I think with their var reseller SI network and I think with where they're positioned in this AI world I think if you wear a Microsoft and hey open AI they need Microsoft more than Microsoft needs open AI fantastic thank you Tim with that note we'll uh we will record it and come back to you in 2026. All right thank you so much Tim thanks thank you so much appreciate it appreciate it thank you so much Shudeep thank thanks a lot your insights are always great appreciate you taking the time always a fun true to your incumbent strategy from last year you built on the incumbent two predictions both are bold I love it both are both are local companies that could make make a big change thank you for those and we'll come back to you in 2026 to see which of those happen more both are going to happen it's a question of which one's gonna happen.
SPEAKER_07:Yeah yeah the startup stuff we spend all day working with startups that stuff's a little bit easier but I like to say okay let's put yourself out on a put yourself out there and be okay being wrong.
SPEAKER_04:Thank you very much for that thank you very much I appreciate it thank you for listening to our podcast from Startup Exit brought to you by DI Seattle assisting in production today are Isha J and Mini Verba. Please subscribe to our podcast and rate our podcast wherever you listen to them. Hope you enjoyed it
.jpeg)
