BUILDSUCCEED
David Goldman, Celesta Capital — Investing in Deep Tech That Shapes the Future
In this episode, Celesta Capital Partner David Goldman shares his perspective on deep tech investing, the long game of venture capital, and how AI can drive meaningful impact in sectors like healthcare, housing, and education, when used with purpose and conviction.
David DeRemer: Hi David. Welcome to Build to Succeed.
David Goldman: Thanks for having me.
David DeRemer: Yeah, thanks so much for coming on. I'd like to start with a little icebreaker. So, uh,what's something wild you've seen in tech in the last seven days? Names like constrained to seven days.
David Goldman: I'm gonna go with something in robotics. There's this company called mec. Who has sort of changed my conception of what I think robots are going to look like, I'd recommend going to check out. Just their whole concept is, there's no reason a robot should look like a human, just 'cause humans look like humans in the same way that planes don't look like birds.
and their robot is designed to look [00:01:00] like a thing that is for the space it goes into. And it's very different from how a human looks. And I have been thinking about that a lot lately and just like, yeah, okay. A lot of money's going into humanoid robotics right now. Is that all silly? Should we even not be using the human form factor at all?
David DeRemer: Yeah, I love it. I was just talking to my wife. Um,we have a little pinhole leak that started and I was like, oh, at some point we'll just have like a little AI bot that will just climb through the walls and just like some sort of little spider that can fit through your power outlet and just climb into your wall and get there.
And it doesn't have to look like a human at all. So it makes sense, but we, it's a good point right now because I feel like we're creating a lot of things in our own image and maybe there's a lot of innovation when we start thinking about, well, what if we eliminate the constraint of like, human form and human ways of thinking and approach things differently.
So, well, a lot of deep stuff to get into on that one. Um, but, let's take a step back and say, uh, like, actually introduce yourself to us. Maybe tell us a little bit about yourself, um, what you do and your journey to today.
David Goldman: Yeah, absolutely. So [00:02:00] I'm a partner at Celesta Capital. We're a early stage deep tech venture capital firm. We invested in VGV at Series A, which is, I think the only reason I'm invited on this podcast.
David DeRemer: Well, and you're really smart and have a lot of good ideas.
David Goldman: so I started my career in investment banking, working on mostly semiconductor deals. At that time, the company that I was working for was AMD that was our client, and I was a young analyst at the bank and the head of strategy at AMD figured out he could just call me and ask me to do stuff for him.
And I couldn't say no because they were a big client. And so I ended up doing a bunch of strategy work for him, helping them map out the startup landscape within the areas of semiconductors they were looking at. And that was really my first introduction into the industry, was working within the banking side on the sell side and became really fascinated by all of these startups.
And then went to go work for a diversified investing group where we were an LP in [00:03:00] Celesta. And so I started looking at the deals that they were doing and some of the co-investment opportunities, which had a lot of crossover with where I had been looking when I was in investment banking. Went to work for a year at a startup in the building construction space.
And then about four and a half years ago became full-time at Celesta. And so at Celesta we look for early stage. So, that's kind of like seed to series B. The letters aren't as important as it is the stage of the company and the capability for us to make an impact. And then we invest in deep tech companies, so companies that are built around a technology innovation.
So that can be a systems level innovation, it can be an invention. In the case of VGV, it's really that differentiated knowledge of flutter. And then it can be monetized in a variety of ways. Sometimes that invention is monetized via hardware, via IP software or in the case of VGV in services.
David DeRemer: So, yeah, deep tech, it's the sort of foundations of technology innovation, and you're trying to find those seeds that will like blossom into the next wave. is that the general idea?
David Goldman: Yeah, it is. I mean, it [00:04:00] means different things to different people. for us it's generally, when there's a core technology innovation that the company has built around. So within the startup world, you can have lots of different types of innovation, right? You can have a business model innovation, you can have some sort of regulatory arbitrage that you've figured out where you're able to operate, but maybe someone else can't because of some way that you've designed your company.
For us, that core kernel is the technology. If there isn't a differentiated piece of technology there, we're not interested, even if, you know, perhaps in other ways, that it could be very successful. 'cause that's just not how we invest and not how we look for companies.
David DeRemer: Hmm. Well, and VC is like a, such an important part of the technology ecosystem and I think a lot of people, there's a lot of mythology around vc. There's a lot sort of. Looking at it with,you know, rose colored glasses and excitement, if you're building a startup, you wanna find your VC partner, uh, someone who's been in VC and also come up through other ways, uh, other types of career like companies, right.
And then landed in vc. What's something surprising about [00:05:00] working in technology VC that you think most people expect?
David Goldman: I mean, I think technology VC is like a very general term that describes a bunch of different jobs, and it can be everything from people who sort of like meet with the founder one time and are like, yeah, it seems smart, has good chemistry, you know, here's 250 K and I've made 300 such investments, and I'm just gonna see like if one of them works out and then be like, I'm a genius because I, you know, wrote the first check into X, Y, Z thing.
the way that we do it tends to be a lot more involved and a little bit more, driven around conviction in individual companies and individual markets. And so it seems like a very, Glamorous, like, you know, you're talking about big sums of money and these companies are all rocket ships and everyone's posting all their successes on LinkedIn, but I think you end up spending as much or more time with companies that are struggling than companies that are doing well.
Probably much more time with companies that are struggling than doing well . And [00:06:00] so at times it can feel like in your day to day like. Wow, everything's kind of struggling because that's where my time is needed the most. Those are the people who have the most questions. They're going through some situation that's really thorny and they're trying to figure out, do I trust my lawyer on this one?
Like, what are the other considerations, oh, this investor is saying this thing. Can I trust them? Do you know them? Do they have a good reputation? but the reality is that there's lots of good things that are happening, and that's mostly when they don't need to pick up the phone and call us, except to tell us.
By the way, it's going really well.
David DeRemer: Hmm. Makes a lot of sense. I mean, VC or investments in general, right. There's definitely that portfolio theory where, there's huge numbers around startups, like how many don't make it. And so, you know, in a portfolio theory that does mean that there's a lot that are gonna be struggling.
And you're right, I've actually never really thought about it that way. Like the other side of that coin is like. For every one company who's gonna IPO and make the huge return? There's a ton that don't have a great outcome, and that's really difficult and I would imagine that's a hard part of the job.
so kind of tracks with sort of leadership in general too, is [00:07:00] right, like the hard things, the problems tend to boil up that you need to,address and deal with, not necessarily the celebrations and great things to high five about. So.
David Goldman: Yeah, well, the really challenging thing is you build personal relationships with the leaders of these companies, and you've, in many cases, been with them for years and you've seen the ups and downs, and so if things don't work out a lot of the time, it's not anyone's fault. Like the market shifted.
Something happened that was outside of their control. They had a big customer who walked away because of a thing that was outside of their control, whatever it is. And now you have this emotional moment of this thing that they've been working on for a long time that they've poured their heart and soul into is ending.
And you're a part of that ending and you've built this personal relationship and Yeah. it's just tough.
David DeRemer: Yeah. Well that's a, that's a good transition to something I wanna talk to you about today, which is, obviously right now the market's in a bizarre spot. Right? and,uh, you know, we've got tariffs, trade wars, uh, economy's been going down. and I thought maybe like, rather than talking about like what's going on today, but may, and maybe to try to make our conversation a [00:08:00] little bit more evergreen.
What are some strategies,and sort of recommendations,we can follow as entrepreneurs, as leaders of companies about how to handle these types of crazy chaotic markets and moments. What do you recommend to some of the firms you're working with right now around how to endure these tough times?
David Goldman: Yeah, I think the important thing is understanding the trade-offs, so. Immediately when a big event happens, you're gonna get questions from investors, from customers, partners, whoever. Like how does tariffs affect your company? Because that's, the thing that's happening right now. and people tend to have this like knee jerk of, I need to immediately have an answer.
And that answer needs to immediately result in something like, oh, don't worry, we've moved manufacturing away from China to Vietnam, which may be the right answer. But a lot of the time people aren't thinking about it in a structured way about the various trade-offs and what you're giving up if you do that knee [00:09:00] jerk reaction.
And so having the time to just take a beat, talk to the right people, think through it in a structured way, consider the trade-offs, and then have your answer, and you know, in the first 24 hours, it's appropriate to say, I think the impact is this. But I'm doing this, this, and this to find out more and I may adjust my view.
So gimme 72 hours.
David DeRemer: It does often seem that the pace at which we need to react as companies or as a sort of global tech environment or whatever it might be, is, is really fast, but. I think you're right. Like these things do take time to unravel and even things, you know, even things take time to actually sort out what the impact even really is.
But regardless of what the news reports or anything like that, how do you as an investor or as an investment team, navigate these environments? 'cause clearly, I would imagine it has an impact on how you're thinking about the types of things you'd wanna pursue or go after, or, how to, deploy your capital differently, across, through your portfolio.[00:10:00]
David Goldman: So we have like kind of an interesting style of investment. We venture capitalists relative to other types of investors. So if you work at a public market hedge fund or something, you can go in and outta positions really quickly. You can change your mind on things. Everything needs to be like really liquid and fast moving.
We're making an investment at the early stage. We're signing up for potentially a decade plus investment in a company, and we want it to become massive over that timeframe and usually over that timeframe. Underlying things are going to change. In a decade, you're gonna have macro changes, you're gonna have different governments, you're gonna have different policies in place.
And so unless the underlying technology creates some sort of favorable situation or differentiation for the company that's really persistent across those things, it's unlikely to be able to be successful over a time period like that. Or even over five years, you know, five year timeframe you're guaranteeing there's gonna be a presidential change.
So. You can't just make a decision that's based [00:11:00] around something that changes, at the drop of a hat based on, one comment from someone who may not be in power in five years.
David DeRemer: Yeah. And actually when things are chaotic or being disrupted, I think sometimes that's the right time to be making bold investments or,big choices. a lot of the times we work with big companies and one thing that tends to happen in the marketplace is people become risk averse 'cause they're uncertain.
So like making, maybe making tech investments or things start to slow down because people aren't sure there's risk. but it feels like maybe we should be doing more in these moments, right? Because these are opportunities where there's disruption where you could kind of do something interesting.
And I guess if you're investing in early stage companies that have a long time horizon, those ideas, those things could be materializing any day regardless of what's happening in the macro situation.
David Goldman: Yeah, I think you touched on a good point. So when things are going well, you tend to have a higher degree of perceived success from startups, and so more people who do not have strong conviction in an idea are likely to [00:12:00] pursue a startup because there is either a perceived higher likelihood of success or some social value attached to that that they think is good.
And in some cases that's fine. There's more people pursuing ideas is good. There's a marketplace of ideas the best one should win out. But it can be challenging as a venture investor to differentiate between those who have real conviction and those who are perhaps doing it for other reasons. And in times of uncertainty or in downturns.
The only people who are really trying to, you know, leave a cushy job in big tech and go do their own startup are because they really, truly are passionate about it, and they're going to have the grit and stick to it.
David DeRemer: Is that conviction kind of the key, one of the key elements of success when you evaluate opportunities? One thing just in general, I was curious about is when you look at. All the various companies out there that are, interested in raising money, they've got a cool idea. They're chasing whatever the gold rush of the day might be.
How, like how do you evaluate, companies for [00:13:00] success, um, and identify what has the highest potential for good outcomes?
David Goldman: Yeah, I mean. If there's a lot that goes into it, maybe not enough for a pithy answer right now. In general, I think on the founder's side, we're looking for people who have, like I said, that grit to stick by an idea because any startup is going to go through harmonics. It's never just up into the right, and so those who don't have the stomach for the downturns are unlikely to survive it.
And if you want your startup to be successful over a five or 10 year period, it's going to have to survive downturns . On the technology side, it's kind of different. I mean, we invest in a lot of different categories, so the way that you look at a software company is gonna be different from a semiconductor company is gonna be different from a sensor , bio convergence company, et cetera.
So I don't necessarily wanna go into the, each one of those.
David DeRemer: Yeah. Yeah. Well, I do wanna, um, tap into AI maybe, but, um,it's, it's funny, you're that grit and conviction, it's obvious when things are like a downturn or you gotta navigate a difficult thing. It's also not easy when things are going [00:14:00] great. Like, I think one thing that people underestimate is the complexity and difficulty of scaling too, right?
Like how you, how teams change. How, more money, more problems, right? It's like the bigger you get, the more things you, expose yourself to and the more just challenges there are to face. And so I think having that grit and that conviction, like you have to be able to see it through the good times too, kind of in a way.
David Goldman: Yeah, it, it's also, I mean, to use the term broadly, like a question of integrity. and I don't necessarily mean integrity in the sense of, are they gonna return someone's wallet they found on the ground, although that is also important. but.
more do they have like a, an internalized sense of self.
Where in the downturns, they're not going to become so despondent that they can't survive through it. But also in the upturns, when things are going well, they're able to see when people are maybe giving them too much shine and they need to be focusing on actually running the business instead of themselves.
Or, you know, they're don't need to, not overhire because. Lots of interested, [00:15:00] lots of people are suddenly interested in their company and they're like, oh my God, that's so amazing. I wanna hire all of these people. They're like, no. Okay. I understand why I am hiring what I'm doing it for and just 'cause I have money right now, it doesn't mean I have to spend it because I have these specific goals and I want to go achieve them.
And so it, it's a little bit about that, just like internalized sense of self
David DeRemer: Hmm.I like that. And you know, obviously, uh, one of the technologies, so yeah, I know you guys invest in a lot of different types of things and,you know, you don't have a strategy around a particular type of technology, but deep tech in general. The big one of course, uh, macro right now is AI machine learning in general gen ai.
so just with respect to that, given that it's the big theme, what is your current view and thinking, about what's going on in AI these days?
David Goldman: I mean. can I ask in what sense? Like with, from the investing lens, from my personal life, GDP impact.
David DeRemer: Let's do the personal life first. Like what's, when you look at all this stuff, there's a, um, you know, we, we live in this really interesting moment right now where you have sort of some economic turmoil, new administration, tariffs, all these things. You've got [00:16:00] AI coming and it's got this like insane promise and like massive productivity.
And you can do things like that. You can do a week ago. The pace is really fast, but it's also terrifying for a lot of people. I think for me, what's been interesting about AI is. We've in sci-fi, we've been talking about AI for a long time, and so we have this like existential crisis kind of baked into AI that other types of technologies haven't really quite had, because we kind of have this like terminator like vision of the future that's been given to us by sci-fi.
So you have this like insane technology progress. Coupled with kind of a tricky economy and then fears of like, oh my God, the AI's gonna come take all of our jobs. and I think that's the current state of fear, maybe not less the terminator fear and more the what does this do for the future of work and for the future of what we all do every day.
and so I'm just curious, like as you kind of have, have been filtering in all the information and seeing things from an investor side, just being in tech, um, being in business in general, what's your current kind of. Vibe [00:17:00] on it all.
David Goldman: I am broadly very optimistic about it. I think in general. There will be subsections. I mean, it's not gonna be evenly distributed, both in terms of the impact and the sort of distribution of the actual technology usage. Right? We have lots of technologies that exist today that are still not widely distributed for a variety of reasons, right ?
We have the capability to have clean water everywhere in the world. We don't have clean water everywhere in the world true of a lot of other things as well, but where I'm very excited about it is in making. More productivity in industries that are important to us. So if you look at where a lot of the sort of like economic malaise in America comes from, it's things like healthcare, housing, education, and the unequal distribution of those within the populace.
The inability or high cost, you know, inability to access or high cost of them relative to what [00:18:00] many people think is what they can afford. And so anything that we can do to improve productivity and make those areas more accessible to people is going to make people's lives a lot better. And so the extent to which you can make homes easier to build quicker, to build cheaper to build, you can make healthcare more broadly available at lower costs.
Particularly like expert level healthcare available to people who can't go to, you know, Stanford or Harvard's, hospital system and, access the best doctor in the world. But you know, if anyone in America can get access to that, quality of that doctor through ai, that's a huge unlock in terms of outcomes.
and then in education, like just the incredible things that one can learn using these in a self-directed way. It's not too difficult to see how they can be imbued into the educational system in a way that allows children to learn much more. Lots of challenges to that stuff. There's reasons that Ed Tech hasn't permeated more in the United States, um, but the promise is there, at least.[00:19:00]
David DeRemer: I love it. I love you. How you're kind of spinning it into like, or turning it in, in the focus onto the things where we can actually like real, really create value for a lot of people. Like one thing that's driving me nuts about how we talk about AI right now is, you know, I was just at Google Cloud next, right?
And there's all these booths and they're all saying things like. build this thing in like an hour, or replace your engineering team with this like AI product. And I'm kind of like it, it kind of blows my mind 'cause I'm like, here you have a tech conference, or you're selling stuff to technical people and you're basically trying to sell them something that's gonna replace their job.
Is the promise. And I'm like, this doesn't seem like good marketing to me. You know, it seems like you're missing something about how to really sell this stuff. But then, um, I dunno if you saw the most re this, uh, Harvard Business Review thing where they were talking about the changes in usage of gen ai and the top three were therapy, companionship, organizing my life and finding purpose.
Which is really interesting. And of course it's been making the rounds 'cause people are like, oh, like we're already offloading our like deep philosophical needs to ai. But in the article they actually talk about how well [00:20:00] hold up, you know, people in Africa don't have access to therapists or psychiatrists.
And so if you could kind of emulate if. That sort of experience with a gen AI product, maybe it's not as good as the real thing, but if it's better and actually now accessible to a lot more people. That's really interesting. Right, and so kind of focusing on how these things, like you said, build houses faster, like get the cost of that down so more people can have affordable housing or like do things with our food supply or whatever.
That's the stuff that's really exciting and I feel like we're in this like marketing bubble right now where all you have all these tech companies racing to figure out how to do something interesting. And so they're finding like really like quick, like dopamine hit things that like really hit the, the balance sheet or the, or the p and l statement.
and they're focusing on those things. I think we need better vision for ai, more aspirational use cases for humanity.
David Goldman: Yeah, I mean. There's some stuff that's just nice and like it, I don't think it really adds anything to GDP, but I like it. Like there either there's an app that's really good at meeting notes, summarization, like I.
used to have to take notes in meetings. Now I don't, my life is [00:21:00] better. I, I'm not paying for it.
I don't know what that adds to the GDP, probably nothing. but then on the subject of things like therapy, like, it's not just A question of access in terms of can people in Africa get access to therapists? Can everyone in America afford to pay one to two $50 per session to talk to a therapist? But there's also this subsection of people, and I don't know what size of the population it is who don't feel comfortable divulging certain things to another person.
and like the example I would give is, I don't think there's anyone in America who would be willing to have all of their Google searches. I. Publicly posted for everyone to see, right? But you're very comfortable putting weird stuff into the search bar on Google because you know it's just between you.
And so to the extent you can have that interface that is private with an AI therapist where you know there's no human, and as long as you can trust the AI therapist company, which let's just assume you can. I don't know if you, obvi obviously always could, but lots of people will. then you've created a whole section of society who is now accessing therapy, [00:22:00] who possibly were the people who needed it the most.
Like people who couldn't feel comfortable around another human can now get that help.
David DeRemer: A hundred percent. Yeah. we talked to, uh, a doctor at Mount Sinai Hospital. He is the, like foremost prostate surgeon in the world. It's a thing that men don't really like to talk about, you know? and there's a lot of people who like die and have major health outcomes just because they're socially awkward about talking about it, but maybe you can talk to the bot.
And turns out a lot of people like to do that and they're comfortable with that. So. Totally agree. Where's,as we are covering a lot of ground here. this is,changing so fast and even within what's happening here, I mean, there's such a huge spectrum of where dollars could be funded or where you could invest in AI broadly.
It's a very big space. I've two maybe kind of two lenses on that. Like where is in, in your view of like observing what's going on in investment in this area right now? Like where are the most dollars going right now? More broadly, not just from Celesta, but like the community and then. Where do you think there's like sort of latent or emerging opportunities that are maybe being [00:23:00] overlooked by the broad investment trends?
David Goldman: I think most of the dollars today are going into two things, and they're interrelated, which is basically creating, for lack of a better term, supply of ai. So you have lots of dollars going into foundation model companies. So this includes the open AI Andros of the world, but also the new ones like, you know.
There's been $4 billion raised from two new foundation model companies. I mean, at least I think that's what thinking machines is going to do. I don't totally know their business model, but in, you know, 18 months ago, I would've told you that those were more or less set that I was gonna be open AI and philanthropic and X and mytral and there weren't gonna be too many more who are gonna raise a lot of money.
And now we have two new ones. So that's been one big bucket of money is at the foundation model layer. And then the second one is in. Hardware and infrastructure that enables that. This is where we tend to do a lot of our investment and even before ai, where we were doing investment. So this includes on the public market side, stuff like [00:24:00] Nvidia and AMD, but then also on the private market side, investments in new AI accelerators.
So, semiconductor processors that are specifically tuned for AI workloads. things like networking, which allows you to move data around, better, which is particularly important for training these really large language models. And then software tooling on the infrastructure side. So stuff that helps you use your infrastructure more efficiently.
Stuff that helps you train your models better, stuff that helps you access them, et cetera, et cetera. So all that stuff's in a broad sense, like creating this supply of ai. and a lot of the growth so far has been linked to the foundation models and to companies who are building on top of them. so then you have this next layer of.
maybe the top layer up, which is like these vertical apps. So that's everything from like Cursor or Windsurf in the coating space, who are building on top of other people's foundational models to some vertical apps that are doing their own models, either because they don't need to have a super general [00:25:00] thing.
They need something very specific, like their type of text. The type of input for their model is something other than text. It's some different type of data that isn't contained within chat. GPT. or because they think they have a really good knowledge graph and so they're just gonna take an open source model and build on top of that.
David DeRemer: Nice. So supply of AI and then sort of the, the interface interaction layer on top of that building on those foundational models.
David Goldman: Yeah. and it's really in that last piece where I think a lot of the growth is going to come from in the next two years. and so let's call it basically like supply and big demand drivers, right? Like, There's lots and lots of really interesting things being done in different verticals, and not all of them will be successful, but some of them will.
And so in areas, I mean like some of the things we talked about, healthcare, construction, coding is like the most obvious one because it's an easy, easily verifiable domain. so it's easier for the AI to see if its answer is correct or [00:26:00] not because it's like, does it compile Yes no. which is very different from construction of.
Did I actually speed this project up by two days? Like kind of difficult to prove the counterfactual.
David DeRemer: Yeah, it's exciting to hear that there's more investment and opportunity. 'cause walking around, like take Google Cloud, NEX is an example. You know, there's thousands of vendors there, 33,000 people, and all these people kind of pitching their, their own unique story and the thing that they do.
But then you look down on the floor and there's a giant Google logo, and you realize like, we are all just guests in Google's house, right? And so is all of this stuff, just really for the Googles and the Microsofts open ai, the, and Nvidia of the world, right? All of. What we're really doing is creating value for them, by being this like service layer on top.
But I wonder, you know what, I've talked to a bunch of, uh, machine learning kind of agencies, development companies and stuff, and, uh, one interesting nuance that I found there was like, there's some like degree of like scoffing, around people who've been doing ML for 15 years when it comes to gen ai.
'cause all of a sudden gen AI put AI on the map for so many people and they're [00:27:00] like, Ugh. You know, like there's so many things people are asking us to use, LLMs and Gen AI tooling for that would be better just to do it the old traditional ML approach. And so I wonder if, there's like a boomerang that will happen where, LLMs and gen AI will kind of be like the safe onboarding pathway for a lot of companies to get into.
Machine learning and AI solutions, and then they realize like, oh, actually we have all this data. Maybe we should train and build our own ML model, build our own things. And whether or not that becomes the flywheel that kind of gets new types of models and foundation models and things like that sort of started.
so it's not just everything built on Gemini or, uh, chat GBT. I think there's an interesting flywheel there potentially.
David Goldman: Yeah, I think that's broadly correct and it's, I think also a good thing. So the extent to which you have people awakened to the idea that value can be created by. Investing in r and d around better using their data, better understanding AI tools. They may start with saying, okay, [00:28:00] use chat GPT, because that's the only one they've heard of.
But ultimately, people care about outcomes, right? They want a six inch hole, not a drill bit. And so the people who are building these things, if they can deliver those outcomes in a different way, will get the great credit that they need. And this is the. Investment and executive attention needed for those people to be able to have, prioritization within their organizations given to them.
And whether that's an outside group getting money or it's an inside group, you know, getting resources and getting people and being able to present to the board meeting, whatever it is, it, it gives the opportunity for this value to be created. And we see this in our portfolio with some of the semiconductor companies, for example, where.
If you were doing something that, let's say for example, on the networking side, so you're creating a hardware product to help move data more efficiently within hardware systems. Even before ai, people [00:29:00] needed to do that, and there were lots of reasons that you needed to accelerate it, and there was a whole roadmap of them doing it.
But now that it's so essential for ai, there's a lot more focus. There's a lot more dollars, there's a lot more buyer interest in these things. And so even though you're maybe making a similar technology as to what you would've made before, and that's not universally true in networking, but just as a hypothetical, because you now can say that it's ai, you can access a bigger, better pool of capital from investors.
You can get better employees to work on it because it's more interesting when they go tell people at cocktail parties what they work on. And then when you're selling into end customers, it's a higher priority for that CEO. Because he's talking to his networking team and saying, I need this stuff for AI and make it happen.
And now they have the budget and the focus to go out and work with vendors
David DeRemer: Yeah, I was drawing a parallel to the AI stuff happening in the market right now to what you were saying about like reactions to tariffs and market conditions, right? Where sometimes it's like you need that urgent reaction. Like, what are we doing about this? I feel like over the [00:30:00] last 12 to 24 months, that's kind of been happening in, in large enterprise and companies too.
Like, oh, we gotta do something about ai. same, kind of similar, would you recommend sort of that same approach still of like, Hey, take a minute, think about it. you know, figure out where you want to do, like, how, how does someone navigate that when there's so much change and the pace is so fast and you know, you need to react to it?
what kind of advice do you have for leaders or product creators to like smartly engage the wave that's happening?
David Goldman: I think having a really deep understanding of what makes your company valuable to your customers. Then seeing where AI or whatever the thing is, interacts with that. So, it's a little bit different in the two scenarios. 'cause I think in the case of tariffs, it's like, do we need to make changes to, because something is affecting our cost structure And so maybe we need to change pricing.
Maybe we need to change supply chain. But in a lot of cases, people don't even fully understand their supply chain, so. You probably need to do some [00:31:00] work to understand all of the inputs and trade offs before you just do a knee jerk reaction. Whereas in the case of ai, it's really more about thinking about what is the possible upside?
What is the threat if someone else recognizes that upside before us? And then where can we create the most value for our customers using AI or creating our own AI or whatever the case may be in a way that makes sense with our business. So, you know, it's a little bit like, you know, when Snapchat introduced stories and then Facebook introduced stories and then you saw every other app had stories on it, like LinkedIn had stories and you're like, well, why does LinkedIn have stories that has nothing to do with like, what I.
wanna see outta my professional network?
People were just fast following too fast. and so, you know, think for a second about why people are coming to you, what it is they want from you, and whether or not the decision you're making actually matches that, customer desire.
David DeRemer: Makes sense. Makes a lot of sense. Um, one thing I wanted to ask you about, switching gears a little bit around the team, you [00:32:00] mentioned, um, conviction, right? Being important to the teams and, in general, it's like you guys probably come across, I would imagine being in nbc, like a lot of companies selling a lot of things right?
and sort of like really engaging the, the fleeting trends. And you have to somehow distinguish like what is the real genuine. Like sort of real convict like founders of conviction, defensible technology, et cetera. but one of the things that I, I noticed in your profile, you have this thing about like, advising founders to focus on like 10 x capabilities and avoid getting bogged down.
And I feel like in a world where things are changing a lot and you have these new stimulus, you're like, oh my god, these emergencies or these new emerging things. Can you like, expand on that sentiment a little bit more in terms of like what you mean by that? Sort of the,the te like focusing intensely on the 10 X capabilities and what kind of, um,how a leader can operationalize that in their day-to-day when they're just inundated with chaos and change.
David Goldman: Yeah, I mean, I think it's about recognizing where you have the greatest impact. No, CEO has more than [00:33:00] 24 hours in a day. And so thinking about what you bring to the business that no one else can do. So there's some things that a founder just has to do, or a CEO just has to do. The decisions that must be made by the CEO, no one else could possibly make them.
But lots of decisions that get made by A CEO don't fit into that category. And so in many cases, those things can and should be delegated either to someone within the organization or someone outside the organization. So like an example I.
would give is like, accounting decisions. For the most part, those things can be delegated.
you wanna make sure that you're broadly within the rules of what is acceptable, and you're not in violation of anything with the law or with the IRS, but you know how you should be categorizing some expense between different GL accounts. That's not a good use of any CEO's time, but some CEOs do spend time thinking about that sort of thing because they just like to be in the details. and so instead, if you're a highly technical founder who's a incredible inventor, you should be spending that time working with your technical team [00:34:00] on the product roadmap or on driving forward innovation, not on, you know, is this expense in category A or category B?
David DeRemer: and what about like, leading teams? You, you recently wrote an article, about, uh, the, uh, sort of the importance of multidisciplinary teams, right? Getting, engineers and designers together in the same rooms. And sort of that idea of like sleeping on the factory floor, getting your, in the case of like a Tesla or something, getting your car designers and your engineers together to work on it.
can you expand on that idea too, and maybe tie it to that theme of like, getting the right people in the right place and getting the right types of collaboration? Because the opposite of the thing, like you're saying with the CEO who likes to get in the, in the details is that that could actually maybe get in the way of collaboration.
or if you're not doing it for the purpose of collaboration, maybe it's not the most productive thing to do, but maybe dig into that a little bit. I'm curious, your point of view on, on these multidisciplinary teams and the importance of that for innovation.
David Goldman: Yeah, I mean, Ithink one of the biggest advantages startups have is being small. and what I mean by that [00:35:00] is if you have a large enough organization, you almost by definition have to break it up into some form of grouping. And oftentimes that's like some form of functional grouping. So you might say, okay, all of the designers work in this office, and our head designer lives in LA and so our design office is in LA and we really like this head designer, and he wasn't willing to move.
So all our designers now work in LA and we have a thousand of them. So they have their own office. You know, our head of manufacturing, our factories are in Mexico, so he lives in Texas, and so all our manufacturing people are now gonna be in Texas, and we have a thousand of them . And so inevitably, all of those people may have cohesion with each other, but they're not going to have cohesion between the two groups.
And so you're going to have things that are designed, which cannot be AC that had you had feedback from manufacturing into some of those decisions, you would've been able to produce it more easily at lower cost. At greater scale. All of these things, startups, by virtue of being small, have the [00:36:00] capability, although they don't always take advantage of it, to be tightly cohesive and to be, if not physically in the room together.
'cause you know, these days, remote work is very often, very often the case to be in close contact with each other, you know, virtually. And so because there's a limited number of people, you can just have that cross pollination that allows those decisions to freely flow from just normal collaboration. So instead of saying, Hey, we're gonna write up a, requirements document and then send it over, you guys are gonna mark it up and then send it over.
You're gonna be like, Hey, I'm thinking about putting this screw here. Is that gonna be okay? And you'll say, no, actually, our screw placing machine can only go six inches to the right and you've just placed it seven inches like I need. You need to not do that.
David DeRemer: Amazing. And so, it's kind of interesting because a lot of things we hear about ai, like in defense of like humans in the future relative to AI, is like we, we still have this communication, like humans aren't good at telling each other what they need. And so [00:37:00] if we can't do that, how are we gonna be able to like, direct the ais?
but I feel like there's kind ofa, an interesting maybe opportunity here for technology, right? To bring people together and increase that collaboration and if, the ability to share information and for us to kind of like. Refine it together, gets us closer. I wonder if we can get to that.
There's also this book, team Topologies. You ever heard of this? I think I mentioned it before on,this podcast, but it basically is the idea that, the shape of your, uh, product you create, mimics the, um, structure of the team. so like if you have teams that are feature-based, like over time your product will diverge in terms of how those features
David Goldman: Mm-hmm.
David DeRemer: If you do it stack based, like backend, front end, whatever, like you know, over time your systems will diverge and get harder to integrate. so it's just the top apologies of things I think are very interesting as you think about this. but yeah, coming from where I come from, like user-centered design, human-centered design, collaboration, getting design, engineering, everything, product people all together, yeah, I think it, it makes a lot of sense having the good, good, at least triads of people who understand things from different perspectives.
So you get aesthetics. Sort of the economics and the ability to make the thing [00:38:00] all sorted out at the same time.
David Goldman: Yeah, I mean one of the areas where I don't think AI is good yet, which I would love to see it improve at, is in being able to make these sort of like cross-functional connections. I. So in a theoretical future, you could maybe have the two big organizations in different places, but if they're doing a good job documenting their thought processes, so like maybe they're having their meetings transcripted and put into some knowledge base, then there is some central actor who is capable of looking at all of that stuff and actually pointing out to them that if they had been in each other's meetings, they would've realized that there was some conflict and they're actually on divergent paths and they don't realize it.
And so it can like raise a flag and be like, oh, actually there's an issue here. Now we're not at that point yet. At least not to my knowledge.
Um, but that's like a future point for ai That
David DeRemer: Yeah, the pace is so crazy that it feels like, uh, one of the things I've just been saying is like, once you really start exploring these things and using them and you see how [00:39:00] fast things are going, it's like something that doesn't work. It's like just a matter of like how long until you'll get there.
but, let's look at time for a minute, right, because we have, obviously we're in an interesting moment, but it's, it's always like this every year or, you know, every, whatever finite timeframe you look at, there's always a lot of change happening. The world is never static. Our lives are never static.
It's always perpetual change. So if you were to kind of like,think back on, on your career and maybe the investments the firm has made, or like others that you've seen, just like companies be really successful. Given the,the nature of this is called like, build to succeed with the idea of like, let's build stuff for the purposes of like, getting better and better and finding ways to like make the things that we do like more meaningful and more impactful.
What are your themes or sort of trends that you feel like you've observed across the most successful companies, startup teams that are, you know, not necessarily causation maybe, but at least correlation or at least, something that you feel like you could bet on as a probability of success.
David Goldman: So I think the first one is like continued [00:40:00] organizational cohesion. And what I mean by that is there is. A clear understanding that is throughout the organization of what you are building towards, and people are able to get behind that and stay behind that. And so where you can see companies diverge is when you have fracturing of organizations, you have politics.
You have people who are working towards things that are not necessarily company goal related, but individual goal related. And so that a lot of that falls on the CEO to create the right structure, to hire the right people and to make the right culture so that everyone is working with cohesion and there's not one culture that works.
Uh, and that's why I didn't say, you know, hey, everyone has to be like this one company, but there does need to be one culture pretty much. That everyone understands and is working towards, and one broad set of goals and mission for the company that everyone agrees on and is [00:41:00] able to line up behind. So that's one.
I mean this one's not super helpful because it just is what it is, but like getting the technology wave right. so people who use famous examples, like there was music streaming before Spotify. It just wasn't the right timing. Being early is the same thing as being wrong, and so having the right technology at the right time for the market, and sometimes that means being slightly too early.
so. In our case, many technologies that we invest in take one or two years of intensive r and d and development before you're launching. And so if you start working on it, if you started working on a, hardware, AI product, the day that, you know, you heard about chat GBT and it had crossed, you know, a hundred million users or something, well, it's gonna be two years before you're actually in market with it.
So you really need to have the version that works in two years, not the version that works today. And similarly, if you had been working on it from two years ago because you understood how big this would be, and then you had it in [00:42:00] market the day that chat, GPT crossed a hundred million users, you'd be very well positioned.
So having the right timing and the right technology for the time.
David DeRemer: Hmm. So a little bit is, uh, the people and the mission, and the culture you create. Some of it's just timing and just somehow things work and sometimes they don't, you know? So,
David Goldman: And some of that timing is, uh, vision. You know, it's understanding the technology trends. It's being able to skate where the puck is going, a little bit of a it.
David DeRemer: Makes sense. Makes sense. Well wrap us up. So I wanted to say, uh, crystal ball time, right? Um, there's a lot of things going on. so I guess maybe to constrain the question, like let's maybe focus on AI and what it's doing. what kinds of like unexpected or transformative businesses or just broader social changes do you think might emerge?
big picture, like long run, you know, let's do a 10 year kind of thing. What's out there that you, you think might be unexpected for people in terms of where this all goes?
David Goldman: I think maybe, I don't [00:43:00] know what the right timeframe to say this is, but like, if you are, if you're having a child now when your child is an adult, they will potentially talk to AI as much as people. So they're gonna have all forms of personal I, whether that's therapist, it's life coach. It's dietician, it's just someone you know you're wanna be able to play video games with.
And instead of playing against the computer or playing against someone online, you wanna play against someone who's like perfectly matched to your capability. And then you also wanna talk smack to them at the same time. that like the summation of all of those things and whether they're embodied in one AI or multiple ais, is going to equal more than they talk to, human beings in an average day
day.
David DeRemer: Hmm. Yeah, I think you could be right. And, um, I, I think we don't really even fully grasp, that's my takeaway is like, I think we're just early stage scratch on the surface. It's like we have, um, the example I keep giving is I feel like we have like the Edge network right now and [00:44:00] the version of ai, right?
Like when the iPhone first came out, it had, I. Internet connectivity, but it was like that, like really kind of poor edge network on at t and you could kind of barely, you needed wifi essentially. And now we have like 5G ultra wideband where it's just like, you just expect to be able to like stream high definition video like anywhere on the planet, pretty much.
And I'm kind of like, what's that? Like, what is that version of ai? Like, we're in the Edge network ai, where we're like very constrained, right? In terms of what it can do. So it's pretty exciting. And I had this, um, I was on
David Goldman: By the way, I also don't think that that's necessarily dystopian. Uh, like some people may take it that way, but you know, for example, we use our phones a lot during the day and if we were to be just using our phone less and instead interacting in the world, but we were help, you know, had some, I dunno, wearable or embodied air or whatever it is that's collaborating with us and helping us with stuff.
And so now we don't need to be buried in a screen that I think that would be better.
David DeRemer: Yeah, I agree. I think, I would love to see the major [00:45:00] players in this space like Google and open AI and others like. Start to paint some better pictures like that are aspirational for humanity, right? Like, oh, let, let's just like replace the workforce with ai. Like that's not really exciting to anyone, I don't think, unless, you know, if you're managing the p and l maybe, right?
and certainly we'll probably drive like broader like income inequality and other things, right? I think that there's a lot of things that these tools can do for us that are actually truly extremely aspirational for like humanity. I would love to see a little bit more like vision put forward by the tech community around what those things are as opposed to sort of the more near term cost savings and sort of the more dystopian things that I think are like less exciting.
but like back in March, I, uh, I heard this talk by Eric Schmidt,you know, former CEO of Google or Alphabet or I dunno, whatever it is, and he had, someone asked him like,what's a prediction about AI or what's your a position on AI that, That you think is like controversial and he said that it's under hyped.
and I thought that was really interesting, position. 'cause o obviously it's about as over-hyped [00:46:00] as anything in the world right now. but just a idea that it can go so far. So, well thanks very much. Thank you so much for, for coming on sharing all these ideas. We covered quite a, abroad range of topics, today.
And so, um, if people wanna find out more about you or Celesta, where can they learn more?
David Goldman: Uh, you can go to Celeste's website, Celesta vc, and I write a substack called close to the metal.
David DeRemer: Close to the metal, like it. Awesome. Well thanks so much. Appreciate your time and uh, thanks for sharing your ideas.
David Goldman: Absolutely. Thanks David.