Ep. 88 - Shahin Farshchi

Hey everyone, Cassius Felicelli here and welcome to another episode of the Homeroom podcast. And today we're speaking with Lux Capital investor Shaheen Farshtey. So Shaheen is an early stage hardware investor who has partnered with several incredible companies including relativity space, Zoox, Epsilon 3 and MosaicML, which was recently acquired by Databricks, just to name a few. A couple weeks ago he was also on Capitol Hill testifying in front of Congress about the future of AI. I was so excited he was able to join us on the program. And I'm also joined by my co host today, Reza Azizian, the co founder and CEO of Farveret. So Farvarat develops liquid cooling solutions for data centers inspired by nuclear power plant cooling. Reza was previously at Cruise and Nvidia where working as the thermal engineer and was also one of the key thermal architects in designing a cooling solution for the Microsoft Hololens. So without further ado, let's get to the episode. So this episode is super special. We're joined by Shaheem Fartshi. We're going to jump into the question. So I thought a fun icebreaker to start would be Reza's from a Persian background. You are as well. I think it would be really cool for you to talk about maybe what it was like growing up in that kind of culture, that kind of household, all that kind of stuff.

Born in the East Bay, little town called Albany, other side of the tracks from a Silicon Valley perspective. Grew up with a huge emphasis on education, a huge emphasis on physical fitness, a lot of food, a lot of family, a lot of friends. So it was a very warm upbringing.

That's awesome. So I talked to Adam Goldbern and he said your nickname was Muscles. Can you share some of the workout routine? I've been trying to work out, but not working out.

So. So I probably am not the right person to ask because my routine is very, very much improvised and it's around kind of my. My work routine. So I like getting my workouts done in the morning. Okay. Because as you know, in venture meetings go over time, there are social events, so you can't really do evening workouts.

I got you.

You're always behind on email. And so I try to do my workout first thing in the morning and now that I have a baby that wakes me up at 5:30, it makes it a lot easier. And so I alternate between running and resistance. And so I go to the gym one day and I run the other and I try to swim as often as I can. So if I Have time I jump in the pool after I do a run or a workout. I'm fortunate that my home here in Palo Alto and my home in LA has access to a pool. And so I do a few laps before getting on the calls and the emails.

Do you have a favorite sport?

I do not do a lot of competitive sports. I played tennis growing up, played basketball, played baseball, but don't really do it very often anymore.

Actually, I'll ask you this. Were you more of a team sports guy or a single person sport guy?

Just given how busy I've been, very much single. Just from a coordination perspective. When I was younger, when I was a kid, I loved playing basketball and baseball with my friends. But unfortunately, you know, as I got busier when I got into Berkeley, I had absolutely no time. And so I just kind of built a routine around these individual activities so I could get it in when I had. Whenever I had a window. I got you.

That's awesome. Okay, well, jump into that next question. After completing your PhD, what made you choose to go the route you went on? Like going to industry as opposed to, let's say, academia? Maybe you can talk about some of the inflection points as well.

Sure, sure, sure. So I decided not to become a professor pretty early on in my graduate career. What I learned, first of all, stepping back, my dad is a professor or was a professor. He's retired now, and I thought of his job as being very glamorous, being on the cutting edge of science, having the privilege to spend his time trying to answer questions that were interesting, spending time with people that he wanted to spend time with who were brilliant students and researchers. And so that was a. It felt like a profession that was right for me. However, once I became a graduate student, I realized that the vast majority of time that professors spend is either performing administrative duties at their departments, writing grants, traveling, trying to market their research, get funding for their research, and just dealing with a lot of overhead as opposed to actually sitting down and trying to solve problems. In fact, as a graduate student myself, I spent very little time with my advisor on actually trying to solve problems that I had with my research. I spent a lot more time with my peers as opposed to solving problems with my PI because my PI was so busy. He had a big lab, he had to find funding, he had to deal with his administrative responsibilities at the university. So I realized that being a professor is not for me. I felt like as if he was doing or professors do a lot of the work that entrepreneurs do but don't have the flexibility or have the financial upside that entrepreneurs have. An entrepreneur can choose to take some time off. It's very difficult for a professor to take time off. For example, there's always grants and conferences and deadlines that are out of their control. And so I decided early on to become an entrepreneur. And like many naive young graduate students, I started a company around my research. So Basically my, my PhD research was on these miniature wireless semiconductor chips that would be implanted into animals brains to wirelessly transmit their neural signals. And the reason why this was interesting was because physiologists want to understand how the brain works, wanted to see brain activity when animals were interacting. They wanted to be able to measure electrical signals that are created by the brain when there is fear, when there is aggression, when there is other kind of interactions taking place between the animals. Not only did I have to build the chip itself, but also this whole kind of back end system that would capture the signals, archive them, perform kind of the grandfather of machine learning to try to categorize the cells, identify and categorize the cells that certain signals came from. When I was talking about my research with a member of the IT team at the UCLA hospital, he said, hey, I'd love if you could just have a version of this with the software without the hardware. So we could take a system that has vital signs that come from our patients as opposed to brain signals, just basic vital signs, and be able to do this kind of analysis remotely on our computers with this data. And so I started a company that built the, basically the predecessor to the smartphone, like a miniature device that had all this wireless capability, Bluetooth and WI fi, that had off the shelf biological transducers, so temperature monitor, a temperature probe, a ECG probe, a pulse oximeter, and basically take these signals into this computer that had all the wireless capability attached to it, it was a Linux based machine, and then transmit that to a remote server and get all these signals archived. And so I did that for a short period of time. It cost me probably 700 bucks to build these machines with all the sensors. And the hospital didn't want to pay more than 500 bucks for each one. So I learned the hard way that it's one thing to have a really cool technology, another to have someone willing to pay for it. And so it turned out to be a cool project, but not a great business. And so I got a great piece of advice from a friend which was get to know investors when you're not looking for money. And so I reached out to a bunch of VCs saying, hey, I'm an electrical engineer by training. I'm postdocing to pay my bills. I'm not looking for a job, not looking to raise money. I just want to build a relationship. I'd be happy to do diligence for you on companies you like. I'd be happy to send companies your way because I have friends that are joining startups or, you know, starting companies. And so think of me as a resource. And a bunch of VCs said, yeah, sure, you know, like, we'll send you deals, send us deals. And Lux at that time was raising his first institutional fund. And they said, hey, you know, would love to have you join us and, and basically be our kind of scout. So I was sending deals to Lux probably for a year and a half while I was postdocing. And then when they closed that fund, they said, hey, do you want to join us full time? And I was just so impressed by this team during the course of that time that I knew that the team was going to be extremely successful and I'd love to be part of it. So I was honored and I decided to join them. And that was, you know, 16 years ago. That's cool.

So on that point of understanding the technology as well as the market itself, I remember in an interview you said, I joined Venture at a time when companies were valued exclusively for their technology. I'm curious to know how valuable has changed over time. And then maybe when it comes to the intersection of software and hardware, how is that kind of contributed? Like how the forces in the market changed and how have they played out.

To where they are now? Sure, sure. So I joined Venture after the dot com bust, right. And it was an interesting time when there was still interest in companies that were de risking technology. There was somewhat of an arbitrage opportunity in taking a technology out of a lab, licensing that technology from the university or the institution where that technology was invented, putting a few engineers around it, somewhat maturing it, and packaging it as a company and selling it to a larger company. And there are many examples of this, of companies that probably invested a few million bucks in technology licensing and early headcount and were able to sell for tens of millions of dollars, probably low hundreds of millions of dollars, to larger companies. And that was pretty popular in biotech, namely microfluidics and semiconductors. Fast forward to say, the early 2010s, and it became apparent that technology is getting a lot more complex. Semiconductors, these systems, medical devices, they're becoming more complex, they cost more Money to mature and to commercialize. And so these acquiring companies are not willing to pay as much for them. Knowing that there's still a lot of risk involved and there's still a lot of development required. That arbitrage opportunity somewhat no longer existed, let's say, going into the mid 2010s, and there's a lot more emphasis on building a real business on top of a technology as opposed to simply just maturing a technology. So these companies required more capital investments, took longer to bring to fruition, and the whole model of investing changed again, from taking something cool out of a lab to building a real business.

Yeah, Reza, you're up.

Yeah, thank you.

And thank you Shain, for joining us. Yeah, I mean, you're going to have some technical questions. I am personally curious to know what part of your prior experience helped you the most to join locks.

So the key to Venture is not having the answers. It is knowing what questions to ask and who to ask. As you know, as an academic, you know a lot about a very, very narrow space. And once you go outside of that space, you become like any other layperson. So what you learn as a venture capitalist is the ability to really simplify problems into their individual parts or unknowns and then identify the right people that can help you better understand those questions and even potentially give you a sense as to what to expect. And so venture is less of a game of knowing all the unknowns because once you know the unknowns, then there is no risk and there is likely no opportunity. But once you know what the unknowns are and have a decent grasp on the risk and feel confident that if those unknowns are addressed, then there is a reward to be had. That is when you are doing your job well as an investor and most times you are wrong in most cases, most startups are not successful. And what we try to solve for is to feel comfortable that in the event that we are right, that the outcome will be meaningful. And so for us, for example, Lux is now investing out of a $1.15 billion fund. We are making sure that if we are right that this outcome could generate something on the order of the fund. For Lux, if we own 10% of the company, if we expect to own 10% of the company, IF we're right, this becomes a 10 plus billion dollar company that can return something on the order of the fund. So it's identifying the unknowns, feeling comfortable with the unknowns by being able to find the people that can help you and ask the right questions.

Right. So you said you mentioned that your background was electrical, right?

That's right.

So it was. Obviously that was your comfort zone, right?

Like initially. Yes.

Projects and stuff. And then if there was other hardware companies that you wanted to scout, you would basically consult with other experts in other areas, right?

Well, I actually found it necessary to consult with folks even in semiconductors. So I was a MEMS analog, you know, expert. And the first company, in fact, that I invested in at Lux was a MEMS company. But there was a mixed signal component to it and I had to ask the mixed signal experts about the mixed signal part of it. So no matter how much of an expert you are in a given space, there's a good chance that you don't know everything about a given investment. And there hasn't been a company that I've seen in the past 16 years as a VC where I felt like as if I have a grasp of everything that's involved here. I've always found myself having to pick up the phone and call people that know a lot more about these things than I do. Yeah.

So the other question that we have is how has your expertise impacted portfolio management? Basically, can you provide some insight into how portfolio construction works in venture?

So there's portfolio construction and then there's portfolio management. These are two different things. So portfolio construction has to do with the sizes of investments you take make, the ownership stakes that you take, the stage at which you invest in, and that is all a function of your fund strategy. So for example, there are seed funds. Seed funds aimed, for example, put $20 million to work over the course of three years with the expectation of those companies raising Series A's, you know, 12 to 18 months after the initial investments, they plan to cut 500,000 to $2 million checks per company. So they invest in 15 to 20 companies and in some cases they don't do any follow ons, they just let their winners ride and they don't put any money into their companies after they made the initial investment. And others would set aside a quarter, you know, half of their funds for follow on investments in the companies that are able to raise, for example, series A. So this whole conversation that I just had right now is all about strategy. It's all about portfolio construction. Okay. Okay. Then there is portfolio management. Portfolio management relates to how you oversee and let's just say, manage your investments. Okay, so for example, do you take board seats? How active are you with the companies? What kind of information, whether it's financial information, whether it's information regarding their business metrics, what kind of information do you demand from your companies and how engaged are you with your company? So for example, okay, if I cut a check going back to that seed fund example, if I cut a check over $500,000, then I expect to have a certain level of transparency into the company's financials, into their strategy, perhaps even demand a board observer seat. So that's the portfolio management piece. And doing follow on investments is also kind of a portion of that is governed by your portfolio construction strategy. So at Lux, for example, to get specific, at Lux, we're investing out of a billion plus dollar fund. We plan to invest anywhere between say 30 to 40 companies out of this fund. Check sizes can be $100,000 to $100,000,000. So we can have million dollar positions, we can have 150 million dollar positions in the fund. And again, the expectation is that every single core investment that we make has the ability to at least turn the fund or generate something on the order of the fund in the success case. And then as far as. And then another aspect of portfolio construction is that we also have a strategy called eno, which is experimentation and optionality, where we take the equivalent of one core investment, which is, let's just say for round numbers, say $20 million, and then we kind of sprinkle that into small companies that could potentially convert into a core in the future. So let's say we have 20 or 30 ish of these small kind of EO investments that would be in a basket that would comprise one core investment and say 30 or 40 or so core investments as part of our kind of portfolio construction. And then again seed and series A. Typically in some cases we would do a series B later stage or a growth round at a company that we missed in the earlier stage. But the core remains seed in series A, which is true to our roots. In addition to company formation. We can talk about later. Talk about that later. As far as portfolio management is concerned, for all of our core positions, we typically take board seats. We typically have reserves set aside for doing follow on financings. We perform a quarterly review of the financials of our companies and every quarter we also review and revisit the reserves that we have set aside for those companies. So that goes to the portfolio management piece.

So the other question is, is there a vertical within Frontier Tech that isn't receiving enough money in your opinion?

It feels like every single vertical in Frontier Tech is getting a lot of attention these days. I feel like Frontier Tech has become somewhat of a, I don't want to Use the word fad, but it's something that a lot of investors aspire to do. And it feels like everything from quantum to hydrogen transportation, to vertical takeoff and landing planes to hyperloop like projects are getting funded. So I feel like there is a lot of capital going into all of these areas. And I feel like there is a shortage of great entrepreneurs and founding teams who really understand how to take these assets, which are these technologies, and turn them into valuable businesses. And when I say a valuable business, I mean a business where there is demand for the shares in the company, which is driven obviously by great unit economics, great growth, and some kind of competitive advantage in the market that they're going after. So technology in itself can't get you this. In fact, technology is the ultimate equalizer. If a technology is successful, then everything becomes cheap and everything becomes free. If you look at agriculture, technology made food abundant. We went from a scarcity of food and famines in the 19th century to the abundance of food and perhaps even the opposite effect we're having. Metabolic disease become an issue in the 21st century. So technology becoming the basis of a profitable, valuable company is the exception. It's not the norm. Technology typically would make everything cheap, everything abundant. Everyone benefits, nobody really profits. That is the dream of technology. And we are looking for those unique cases where technology can truly be the underpinning of a company that becomes a once in a generation company.

So you think that in general, hardware companies these days get a lot more funding compared to the past?

Oh yeah, 100%. I feel like traditional investors tend to be wary of hardware companies as they were before. But there is a lot of interest in getting hardware companies off the ground. Now that said, even though there is access to early capital for a lot of these frontier tech companies, in many cases growth capital is hard to come by unless they're able to generate true unit economics that are attractive.

Oh, I think, yeah, that makes sense. So you also mentioned building a new technology alone, even if it's magical, is not valued ultimately by the markets.

That's right.

And based on that, what does the product market fit look like to you? And how has valuing companies change over time?

Yeah, yeah. So there's multiple questions in that. So the first question is technology. Keep in mind that underwriting a company for its technology is the tail wagging the dog. One should be excited about a company that has some kind of unfair competitive advantage in a market that is exciting in a market that could support a valuable business. And in most cases, that competitive advantage is rooted in and some kind of trade secret or some kind of, in the case of pharmaceuticals, for example, some kind of monopoly on a chemical or a process that would result in that competitive advantage. Going back to the earlier conversation, in many cases technology does not support a competitive advantage and there isn't a business. So again, it's an exception here that, okay, you have great technology. Check, next step. Okay, does this technology in itself or coupled to other things lead to an unfair competitive advantage in an attractive market? And if the answer is yes, then it's up to the team to execute and be able to leverage that technology to create some kind of unfair competitive advantage. Okay, so that's the question on magical technology and how you would ultimately get valued. Then there's a question on product market fit. So I don't feel like as if the way companies are valued has particularly changed. I feel like value is always derived from perception. So if there is perceived demand for the shares in the company, whether that perceived demand is a result of fundamental unit economics, oh, this company is printing cash as profitable, therefore there will be demand for the shares or because of some kind of promise in the future, Elon Musk says, oh, I'm going to launch this network of telecom satellites. I'm going to colonize Mars. Okay, that's vision of the future is what's creating value in SpaceX or all of these AI companies saying that we're building general AI, we're going to put the white collar workforce out of business. That creates value in the company, so long as there's an expectation that there will be demand for the company's shares, whether it's private or public. That is where quote, unquote, ultimately value comes from. And then product market fit is a component in generating metrics and unit economics, which itself could be a driver for demand for the shares. So ultimately the driver of value is demand for the shares. And what would create that? That that demand is from metrics is from a great storytelling CEO Elon Musk, or from a future that again comes from a evangelizing CEO that people would be buying into and creating demand for those shares.

I think that's a great point. That's the power of a storytelling.

That's right.

Some of the great CEOs have.

That's right.

So on the other hand, what are some of the false flags around product market fit that funders should be mindful of?

So there is a misconception that early market traction is an indicator of future product traction or customer traction. That typically is not the case. Typically it's the inverse. And there are so many books that have been written about this. But people still tend to forget. Getting your first few customers and generating your first few million dollars of revenue is actually more straightforward than taking that first few and turning into a, you know, into tens of millions of dollars, for example, of revenue. So scaling a business is not trivial. And the economics of scaling a business are not initially obvious. You may find out that taking your business from zero, for example, to four or five million dollars of ARR could be extremely efficient. But then going from 5 to $50 million of ARR could be extremely dreadful. Could be a trench warfare, you know, very, very expensive. And a lot of founders tend to not totally appreciate and realize that.

I think that's a, that's a good point. And as you know, we have a hardware company as well, Right. We are, we have fervor at that. We develop liquid cooling technology for data centers.

Sure.

And we believe that's. That's the future.

That's right.

Because with all these AI and generative AI, I mean, everyone tried to get all the performance that they can from the chip, but that's.

Thermal management's a problem.

Right. The limiting factor is thermal. And that's what we are trying to solve. And we believe, you know, with the experience that we have, you know, in the nuclear reactor, the only way they can extract energy efficiently from the core is by heat transfer.

That's right.

And we have some background in that. And that's how we are trying to implement that technology into the data centers. And we have been running some paid pilots and we were able to generate some revenue. So we'll see how the scaling up of that would work. But I agree with you that the scaling up the company is extremely challenging task that we have to attack. So moving on to the next question. So what are the current trends and challenges you feel hardware funders should be aware of? I mean, it's a very general question, but I mean, your opinion.

I can tell you where a lot of hardware companies struggle, and that is the challenge around iteration. So a lot of hardware companies have failed because the first product was 90% but not 100% there. And doing another iteration of the product would have required raising another round of financing. And as a hardware company, get into this vicious cycle where you have to spend more money to be able to accelerate development. And if that first product doesn't strike gold, then you're already spending a lot of money. Your burn is already very high, and it makes it very difficult to attract more investment dollars. This Is in contrast to software companies where they can keep their burn very low until they find product market fit, until they start generating attractive metrics and then raise their spend to continue scaling those metrics, proving out the scalability of those metrics and then have access to capital from there. So the high non recurring engineering components of hardware companies relative to software companies makes it very difficult for them to iterate. So getting it, getting the product wrong can ultimately cost you the company. As opposed to software companies that have the benefit of patches of updates, of pivots, of even deciding to do something completely different because their cash spend as it relates to bringing out product is relatively low.

Yeah, that's why I think it's important even for the hardware company to keep the band rate low, as low as you can. Yeah, that's what we have been trying to do.

And there's a lot of money that unfortunately goes out the door. To pay vendors you have to buy physical things as opposed to software companies that put everything up on the cloud and build on top of open source. So until they have a product they don't have to spend a lot of money. It's the other way around for hardware companies.

What's currently missing in the ecosystem to support more technical funders to take their research projects, to launch companies?

I feel like it is a very special time right now. I feel like there is actually for those technical founders that have the ambition to go out and start companies. I feel like there is a lot of investors out there that are seeking out these companies. I feel like unlike software developers that participate in hackathons and meetups, that social where a lot of investors are present, in fact we even host these things. There isn't as much of that in other disciplines. So in semiconductors and chemistry and oceanic sciences and engineering, those communities are rather isolated, unfortunately from the venture community, from the investment community. So how do you create kind of a social ecosystem like what you have in software for those of the disciplines so they can get more interaction with the investment community.

And then the other question is how do you navigate chicken and egg problem of hardware, which is basically to get customers you need to have some sort of minimum viable product. But at the same time to kind of develop that mvp, you need to have a target of the customer. So how do you, how do you address that?

So that goes back to your previous question on the inability to pivot and, and, and to experiment with, with products, to iterate on products and hardware. In hardware you have to get it right the first time, unfortunately. And so that's why experience, that's why knowledge of a market comes in with hardware companies. Because of the cost, you can spend 10, 15, $50 million to get a hardware product out there. Let's just say for example, these mics, if you don't spec out this mic correctly, if you don't come up with the right channel partners, you don't come up with the right distribution, if you don't come up with the right branding, then and this product is not successful, then you're hosed as a company to go out and either start all over again or raise money under onerous terms. So you really need to get it right the first time you and you know, that's always been the problem with hardware companies.

Well, you've given me an idea.

Yeah.

What I can do for the podcast, get a little bit of a social dynamic going for the hardware side. But let's talk about Lux specifically and the fund that recently raised. It's the fund. Sure, company, but tell us what makes this fund unique? Where do you see it going?

It's our largest fund. We are very excited to continue to both create and fund great companies. Again. We expect to write checks ranging from 100,000 to get a company off the ground to 100/million dollars to be part of a growth financing, either doubling down on one of our existing companies or entering a company that we missed since earlier around. And the goal is to continue to identify great teams that are solving big problems. That could be the Jason that could be the basis of giant, giant companies.

Awesome. And then Lux Labs as well.

Lux Labs is a. Is a continuation of an effort that we've had all along at Lux. So at Lux Individual Partners, Adam, for example, founded Kaliope. I was a founding investor in Ava. Josh, my partner, founded Curion in our own offices. And so we as a team always had an interest in creating alongside funding companies. With luxlab, we formalized that effort. We have a team now that focuses exclusively on going after great talent, attaching great assets, great technology to that talent, and then putting in money and helping them get off the ground running.

Right. Okay. And are there any companies that you're looking to start? Obviously we've mentioned in this podcast how hardware investing has become more of a trend, if you will.

Yeah.

So are there any ideas, things that you're keeping your eye on?

There's areas that I'm excited about. Broadly, I'm excited about the infrastructure piece associated with this new wave of AI. I feel like as if there are challenges. For example, you mentioned thermal management. There are challenges associated with connectivity within the data center. That I think is interesting as it relates to these training. These large neural networks, Data centers today are designed around doing arbitrary compute. When we start talking about training large language models, there is a very clear workload that's deterministic and that kind of changes the way you can build and optimize your data center. So I'm looking at a couple of ways to kind of advance that. I'm excited about the whole kind of design cycle of hardware semiconductors to electronics broadly. So how do you make to the earlier conversation that we had around the big NRE piece associated with building hardware? Can we make building hardware as cheap and easy as building software? I think that would be a huge shot in the arm for that entire space. If you look at companies like SpaceX, like Relativity Space, their innovation was not necessarily in. They didn't invent rockets, but what they did invent was how to make rockets. And that was a huge turning point for the industry. Just like how Henry Ford, for example, he didn't invent the car. He invented a particular design of a vehicle and the way it was manufactured so that the person in the plant building the vehicle could drive home. Right in one of those versus prior where you had to be a extremely wealthy person to be able to afford an automobile. Yeah. Okay.

What books would you recommend people read? Resources?

I just finished reading Civilization by Neil Ferguson. Okay, love that book, highly recommend that. And I would like to read his other books as well. I think Empire is another one of his books that I want to read. It's about the British Empire. But I feel like he's a fantastic author and highly recommend Civilization. And I've been reading other books on kind of. I, you know, I'm a big fan of history and so, you know, different parts of kind of our history, whether it's American history, I'm just interested in reading.

So that's awesome.

Yeah.

Okay, let me ask you this question. If you could ask the venture industry to start and stop doing one thing, what would it be?

I would recommend the industry. I would recommend the industry stop looking to one another. Because the more they look to one another, the more they tend to over capitalize some things, some areas at the expense of others. So rather than looking over their shoulders to see what their peers are doing, I think they should focus on the entrepreneurs instead. I understand that every investor wants to fund a company going back to the earlier conversation around demand for shares. They ultimately want to invest in a company that if it succeeds at its interim milestones, its near term milestones. There will be demand for the shares by their peers. That's how the venture ecosystem works. But I think there is too much emphasis right now on that and less emphasis on finding new opportunities. You hear about few opportunities that many investors are pursuing, whereas I think it should be the other way around, where it's many invest, where it's many opportunities that fewer investors are pursuing.

That's awesome. Okay, well, we're at 4:33. We'll leave it here. But man, thank you so much.

No problem.

It was great to have you here.

Thank you. Great to be here.