Why Regenerative Tech Needs to be the Future with Tom Chi, Google X Founder & author of Climate Capital

(Intro Narrator at 00:00:00) Today, we're talking to Tom Chi, Google X founder and founding partner of At One Ventures, about his latest book, Climate Capital. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:17) Why did you write this book?

(Tom Chi at 00:00:20) Yeah, I wrote the book to be able to speak to the three big crises that we are likely to have to face in the 21st century. So that is climate destabilization, economic upheaval because of the widespread adoption of robotics and AI, and geopolitical instability. And I wanted to both talk about how one might be able to step up to those challenges and also practically teach a bunch of skills to step up to those challenges, so folks are ready for the coming century.

(Joel Beasley at 00:01:01) I'm interested. So you're currently backing technology that cleans up industrial waste. Can you explain more of that?

(Tom Chi at 00:01:10) Yeah, I'll tell you exactly how it works. So the purpose of our firm is to help humanity become a net positive to nature. Nature to us has four physical subcomponents: air, water, soil, biodiversity. So what we do concretely is we stack rank the industries that are currently doing the most damage to each of those categories. And some of the categories have subcategories. So within air, we have four major greenhouse gases, and each one of them has got a different set of industries that generate that greenhouse gas. But luckily, across all of these industries, it's like a Pareto distribution where the top two to five industries are 80, 90-plus percent of the damage. So when you account for all of it, it's about 30 industries drive more than 90% of the damage to the planet across air, water, soil, biodiversity. And it's actually too late to try to go clean it up after the fact. What you want to do is you actually want to change the core industrial production so it doesn't create the damage in the first place. And that's actually the cleanest way to be able to solve these problems.

(Tom Chi at 00:02:01) We have a related concept, which is called entering a problem at the right point in the entropy curve. I'm formally trained in physics and electrical engineering, so I oftentimes think about things through a physics lens or an engineering lens. And entropy is basically—it has a technical definition, but for colloquial conversations, it's close enough to the word disorder. And basically, all systems, if you give them enough time, will tend to more disorder over time. So if you enter late in the entropy curve, then problems are oftentimes basically overwhelming and unassailable. If you enter early in the entropy curve, the disorder is not high yet, and you can solve problems way simpler.

(Tom Chi at 00:02:36) So very concretely, just to land what this concept looks like in practice: if you're trying to deal with microplastics in the ocean at the point that billions of plastic bottles have gone downstream and gone into the ocean and smashed into micron-sized pieces of things, then the entropy is very high. The energy and effort it takes to go and recover all those little bits of material, incredibly high. But if you could intercept this problem at the, you know, order 100 extruder nozzles that produce all the plastics in the world, it's honestly very low entropy. You basically say, well, what if we could change the material that's going into the extruder? What if we could change the extrusion process or the way that the material is produced to make it more biocompatible? And you can solve it at 100 extruder nozzles as opposed to every cubic centimeter of the ocean.

(Joel Beasley at 00:03:50) Yeah. And how do you do it without making the straws suck? Those paper straws are horrible.

(Tom Chi at 00:03:57) Right. We went from a polymer-type approach to a pulp approach, and that doesn't totally work. I sometimes call our firm the most boring, exciting firm on the planet because we work on breakthrough technology. And we get to go and displace really damaging production in a way that's quite inspiring. But the way we do it is the most bread-and-butter building of businesses, succeeding at the manufacturing derisk, really in the details of how one actually physically produces something that has got better performance and better unit economics and is dramatically cleaner for the planet.

(Joel Beasley at 00:04:44) You said a couple of times, you know, when you said stack rank the top industries, all of that, you kept mentioning damage to the planet. How do you define damage to the planet?

(Tom Chi at 00:04:55) Well, the reason that we break our goal into four subcategories is if you use the word nature, the word nature is amorphous to a lot of people. And some people will even say stuff like, well, humans are part of nature, so all the cities we build, that's part of nature too. And I was like, well, this is getting too fuzzy to be useful. So the reason we break it into air, water, soil, biodiversity is it's actually relatively clear when we're doing damage to water. It's relatively clear when we are damaging soil function. And there's very measurable outputs that one can do in terms of the amount of pollution that exists in the water, the reduction in practical soil function across biological function, hydrological function. By being able to break it into these four subcategories, it actually becomes very clear when damage is happening. Around air, then we've got four major greenhouse gases. So the industries that basically need to use or produce those molecules are the obvious spot to go to to come with some radical innovations that can improve how that production is done. But yeah, this is the reason that we break it into those subcategories, because it makes it very clear which direction is damage and which direction is repair.

(Joel Beasley at 00:06:13) Do you choose those attributes, or are you getting that data from other research facilities?

(Tom Chi at 00:06:19) Look, about 80% of our team has got technical degrees—and including multiple people that were former scientists, including myself. So we're very comfortable reading papers in terms of how different substances we use in industry have negative downstream ecological impacts, and then we can just use a pretty straightforward mechanism to stack rank it.

(Joel Beasley at 00:06:44) So that's first-party data then. You're deciding what the damage indicators are, and you're coming up with all of that. And then that's how you're doing it.

(Tom Chi at 00:06:52) That's not what I'm saying. I'm saying that at the end of the day, we get to choose where to put the investment dollars. But it is a concrete fact which industries produce the most carbon pollution versus the most methane versus the most nitrous oxide versus the most fluorinated gases. This is just an actual stack rank. So it's not like we are—because more fluorinated gases, you're higher on the stack rank. Less and you're lower on the stack rank. It's not some deep, tough to interpret sort of thing, and it's definitely not made up by us. These are what the industries already do.

(Joel Beasley at 00:07:30) Okay. So the industry is already watching this stuff, and then you're looking at the data.

(Tom Chi at 00:07:34) Yeah. And the industries are well aware that some of the stuff that they are doing could be classified as pollution. So under the Montreal Agreement, we honestly saved the ozone layer because we were able to practically ban CFCs. But then that is now expanded to some other refrigerants and a bunch of the folks that are using fluorinated gases. Because a lot of times, the property of these fluorine compounds is that a lot of them can be useful as refrigerants, just in terms of the boiling and condensation point. But long story short then, these folks have been well aware at least since 1986 or so that they have been a large emitter of a type of pollution, and it's slowly getting cleaned up, basically.

(Joel Beasley at 00:08:23) Can you tell me about an investment that you guys made that went all the way through production and actually has improved the world?

(Tom Chi at 00:08:32) Yeah. We got a couple things that are basically on the cusp of having their big impact. But for us, we consider it quite far along because we are seeing the new unit economic, and we are seeing our successful ability to produce products using that new unit economic. And before I jump into the specific examples, it's just worth noting that if you go to the physical production all around the world of different industries, you're not going to see 9,000 different ways this industry is looking to produce this good. It almost always gravitates down to just one or two approaches that have the best unit economic.

(Tom Chi at 00:09:14) So one thing I really like about hardware and physical investing is when you actually introduce the new winning unit economic to the industry, you're likely to win the whole pie. Because the entire industry ends up gravitating to the technical approach that has the best unit economic. This is very different than software investing where in software investing, most stuff is discretionary. Why do some people use Asana and other people use Slack and other people use Trello or Monday or whatever? Why do some people use ChatGPT versus Claude versus Gemini? This is all discretionary. It's like, oh, I like the results I get better from them. But there literally are other options that are about as good. And when you look at the technical work that happens on the physical production side, no. There's not dozens of options. There's literally a technical approach which has got the best unit economic. If you are the inventor of the next one, you win the whole game.

(Tom Chi at 00:10:08) So let me give you a couple examples. One that I think is pertinent to a lot of folks in the world and not just people that care about the future of air, water, soil, biodiversity is the built environment is about 40% of emissions. So about two-thirds of that is from the OPEX of running the buildings and all the stuff we have in the buildings, and one-third of it is in the embodied emissions that come from the core material production that goes into buildings. So we've backed a company that has been able to dramatically reduce both the cost of building and the practical emissions on both of those categories. And that company is called Cuby Technologies.

(Tom Chi at 00:10:46) Just to give you a little bit of a way to land the unit economic win: if you look in the US, then the average cost of building new structures in the US is about $230 a square foot. If you live in a bigger city or whatever, that sounds incredibly cheap. I live in San Francisco where it's a thousand to 1,100 a square foot, or when there's labor shortage, then, you know, 1,300 a square foot. But anyway, the average across the US is about 230. Cuby is already down to between 80 and $110, so this is not even close. Like, anybody that's in this sector would be dramatically improving their margins because their cost structure is down by more than half.

(Tom Chi at 00:11:20) But in terms of the environmental wins, then this is something that has got way lower wastage, way lower operating costs—so way fewer emissions from actual operating of the building—and it has a very high material recovery capability because even though it speeds up assembly of buildings, it also speeds up disassembly of buildings with full material recovery. And that's an example where you get way better unit economics, but you are also getting way better environmental economics because we substantially bring down, almost by 50%, the operating emissions from buildings. So that two-thirds pie, bloop, shrink that by half. And then in terms of the embodied emissions and materials, then more than 90% of the building is recoverable. Now that said, will everybody fully recover everything in the building? They should actually because the materials are more valuable than trying to mine new materials and produce them to that point. But, you know, you can imagine that sometimes people recover it, sometimes you don't. But if we can improve substantially existing material recovery percentage, which is quite low right now—it's oftentimes single-digit percentages, 5% material recovery for a building they're knocking down as opposed to 90—if we could improve that from 5% to 50%, we're doing quite well as well. And these to us are the real meaty bites in terms of actually changing the overall footprint of the building sector.

(Tom Chi at 00:12:36) One thing that I forgot to mention, but I think it does kind of link this stuff together is we oftentimes look at our investments through this lens that we call the triad. The triad is disruptive deep tech which ushers in radically better unit economics paired with radically better environmental economics. So we're never needing to twist anybody's arm to do the more conscious thing. No. You are just making more money by adopting a better technology. We don't need to wait for regulation to force folks. We don't need to convince the consumers to believe a different thing and want to buy a different product and force you as a manufacturer to change. No. People adopt the technologies in our portfolio because it's literally just a more profitable way to make the thing. And if you're a home developer and you could take out half of your cost structure, amazing. Your margins are through the roof now.

(Joel Beasley at 00:13:57) That is brilliant. That's exactly what you want in a product. What is Cuby again? Can we have a picture of it? See what it is?

(Tom Chi at 00:14:06) Yeah. You can go to Cuby Technologies, the website, and they will show you a lot about it. It's cubytechnologies.com, C-U-B-Y. But to give you a sense of what it is: in one hand, people will look at this and say, oh, you guys make a bunch of homes. That's the product. But no, the product is not the homes itself. The product is actually the factory that makes homes. And basically, a single Cuby factory employs about 300 people, can make about 370 homes in a year, which is averaging more than one home per day in terms of output, obviously, because 365 days in a year, and is able to build at half the cost structure or less. So the ability to produce—the factory as product.

(Tom Chi at 00:14:58) Not only that, our production facilities are way cheaper to make than most production facilities. A lot of times when you do industrial production, you're talking about a new factory build, you might be talking about 100 million, 200, 500 million. These are all very standard amounts to be able to get that sort of facility and the CapEx inside of it off the ground. Cuby facilities, we can bring up an entire other factory employing 300 more people closer to $15 million. So basically 10 times cheaper than a standard factory and also relocatable. Because if you go through the website, you'll see that the superstructure of the factory is kind of inflatable the way that the Houston Astrodome is. And given that, then we actually can pack up the entire factory into a series of trucks and move it to anywhere in the world that we need. So as long as we've got a couple clear acres to set up the factory on, we can get that sort of scale of production going in any locale in the world within six months.

(Joel Beasley at 00:16:01) That's interesting. So you're doing it both in the industrial space and in the residential space?

(Tom Chi at 00:16:08) Yes. It turns out that there's a huge amount of commonality in building. Look, in the industrial space, then sure, you need to think more about MEP requirements, which is short for mechanical, electrical, plumbing. And yes, sometimes you will need to have more interesting handling of energy, water, materials that move through an industrial space. But at the end of the day, it still is four walls and a roof that has got standard fixtures inside of it. And the way that you design those things can be easily reconfigured in the Cuby build system. So they also have an offering for being able to speed up the construction of medium-sized data centers.

(Joel Beasley at 00:16:55) Interesting. So I'm looking at one of the homes right now, the exploding view of the home, and it looks like it's got this different material instead of the standard studs and drywall. It's just a different material, and it kind of all comes together.

(Tom Chi at 00:17:09) And notably, it's actually a higher quality material. Right? So most of the time when people talk about low-cost housing, then the very first thing you do is trade off material quality. Hey, we don't need good countertops. It's all going to be plastic. We're not going to have wood floors. It's all going to be laminate. Right? This is very standard kind of trade-offs that people do to get to, quote unquote, low-cost housing.

(Tom Chi at 00:17:32) What we see with Kubi is that about 70% of the cost structure of building is the labor cost. And with Kubi, we've been able to use cobotic labor, so combining robots plus humans. But given that we're combining robots plus human, we're able to actually reduce the labor fraction really substantially. But in exchange for that, we actually have opened up so much in the reduction of the labor cost that we're actually able to spend a little bit more on materials. Because it turns out that higher quality materials are also oftentimes easier for machine tools to work with programmatically.

(Tom Chi at 00:18:08) Right? So you don't get that many swings at trying to put plastic together, but you can reweld metal. Right? And the metal is the higher quality product that you could be working with anyhow. But we typically don't do that because we're oftentimes presuming that labor is a fixed cost, and we just need to keep on trading off material quality in order to get to lower cost housing.

(Tom Chi at 00:18:31) But the ceiling on that's very low. Right? Only 30% of the build cost is the actual material. Right? So even if you brought down all the quality of that by half and the cost down by half, which would be a terrible building to live in, you've only saved 15% out of the cost structure.

(Tom Chi at 00:18:46) But we've saved closer to 60% of the cost structure by focusing on the labor side of things. And it's been both the combination of cobotic efforts so that basically some of it goes to automation. But it is also we have redone a lot of the internal fixtures and production machines so that you actually don't need skilled labor to get into this business. Right? One of the first homes that Kubi assembled in the US was actually assembled just by four people.

(Tom Chi at 00:19:17) Three of them had never worked in construction before, and two of them are just average-sized women. So these aren't big, beefy construction workers that have been contractors for 20 years. These folks were trained in one week because, you know, very similar to the way that we updated printers and copiers, then we basically done that with a lot of the cobotics inside of Kubi. And specifically, if people don't know what effect I'm talking about, a long time ago—though a lot of your viewers might be too young for this—there used to be this job called copy machine repairman. And it's because every business in the world had a copy room where they printed things and copied things because we were using a lot of physical paper.

(Tom Chi at 00:19:59) Now those machines back in the day were pretty complicated and expensive, and it meant that whenever it broke down, then your office is shut down for a little bit until you can get the copy repair person out there to fix the machine. And then at some point, HP and Epson looked at this, and they found out that 95% of the time, it was just jammed paper somewhere in the handling chain. And they basically said, "Well, what if we could just redesign the printer so you didn't need to call a copy machine repairman? You could just open it at a particular spot and pull out the paper that's stuck." Now in order to do that, they needed to redesign the paper handling route through the whole thing so that it touches the perimeter as opposed to go through the middle.

(Tom Chi at 00:20:40) Right? Because if it goes through the middle, you need to take apart the whole machine to get the thing. Right? But because they were able to control that part of it, they made it so that people with literally no training could just look at a blinking display and say, "Open up panel two. There might be a paper jammed here."

(Tom Chi at 00:20:55) Pull it out. So a lot of our manufacturing fixtures in our home production are actually designed like that. So you actually don't end up needing to have years and years of training. We've made the machines intrinsically safe to interact with, and we've made the type of human interaction for them something that is walk-up usable to folks that have closer to a week of training as opposed to many years of training.

(Joel Beasley at 00:21:21) That's brilliant. So in that $230 per square foot down to $80 to $110, that includes the material cost, the labor cost?

(Tom Chi at 00:21:29) In cost. It's the landed cost of the house on the site that you can move into. Yes. It's the full build. Yeah, of—it has been markedly reduced in cost.

(Joel Beasley at 00:21:42) Cobotic—the word you've used a lot—it sounds fairly self-explanatory, but I don't know.

(Tom Chi at 00:21:46) Means in cooperation with robots. So there are things that are fully autonomous where literally you could turn off the lights. Right? So as part of my education—so I'm formally trained in physics and electrical engineering with a specialization in robotics and signal processing. And I did my graduate work with Raffaello D'Andrea who won the MacArthur Genius Prize and created this company called Kiva Systems, which then got acquired by Amazon and became Amazon Robotics. Long story short, way back in the day, I was working with the team and with that thesis advisor to go build a lot of the sort of pure autonomy routines. And if you actually go to a bunch of these Amazon fulfillment centers today, there's half of the building where it's totally dark because no humans need to go in there. And actually, robots don't need light to navigate. At least these robots don't need light to navigate.

(Tom Chi at 00:22:40) Right? And given that, then there are situations where there's parts of a factory that are just fully automated. That's not cobotic. That's—but the materials are just coming out, and the robots are literally handling the thing tip to tail in this part of the process. Then later on, then, sure, if a human being is interacting with a robot and together they're solving a thing, then that's called a cobotic use case.

(Joel Beasley at 00:23:04) Okay. And this company, Kubi, they've made their own cobotic entities?

(Tom Chi at 00:23:10) Yes. From scratch. Yeah. So they have proprietary—they have a proprietary technology stack that—because they vertically integrated some of these things, they were also able to get additional cost savings that other folks could not touch.

(Joel Beasley at 00:23:28) That's interesting. This also is somehow better for the environment too?

(Tom Chi at 00:23:33) Yes. It is better for the environment, and this is, to me, a little bit more mild, but the numbers still are quite material when you realize that construction is the largest material user of all the industries that we have in the world. But it's very normal for construction projects to have a 20 to 40% cost overrun and also material wastage. And where does material wastage come from? Well, it's quite normal for a bunch of materials to be delivered on a different day than the labor comes to go install the stuff.

(Tom Chi at 00:24:06) Or, you know, the material gets delivered and it gets rained out for a month, and because of it's too soggy for you to go work on-site. Right? And during that time period, you'll have some materials spoilage. You'll also have a bunch of materials that don't quite fit, so you'll have these offcuts. Long story short, it's normal for you to have a big pile of waste that comes with even new building construction.

(Tom Chi at 00:24:27) And that's not them trying to be wasteful. That's just the way that it works when you're building everything on-site, and you have to deliver standard-sized materials that will then be customized into the specific build you're doing. And you're doing the logistics and management of it. The labor is there at the different time than a material, which is on the backdrop of unpredictable weather, and all these things basically create time delays, production delays, material wastage, all sort of thing. And that drives about a 20 to 40% overrun is the average in construction.

(Tom Chi at 00:25:00) Right? That's not an unusual case. We're almost always beyond budget when we do construction. So in Kubi, we don't have any of that. So just imagine not wasting 30% more.

(Joel Beasley at 00:25:12) Just the act of not wasting.

(Tom Chi at 00:25:14) Yes. Just the act of not wasting helps a lot, and we build tighter builds on higher quality buildings. And given that, then it means the insulation, the energy, and money you spend on HVAC is better used. The lighting, all these sorts of things are just higher efficiency, higher quality.

(Joel Beasley at 00:25:37) And so your firm, you research all of this. You understand all of it, and then you invest in companies that are already doing it, or do you seed fund companies, or where are you at?

(Tom Chi at 00:25:47) Well, our first placement tends to be in the seed round or the A round. So, sure, when we enter at seed, then sometimes the companies are still quite small. It might be a four-person team, might be a six-person team. So in most people's definition, still quite early. That said, was there a point before that where there was literally two people at a lab bench being like, "Yeah, maybe we should start a company"? Yeah. So there is even earlier places to get in, but we're typically getting in there at the point where we can see that the core idea is going to have advantaged physics over the rest of the industry. And this is actually a real benefit of working on physical businesses because there is a lot of conventional wisdom in the industry that says, "Well, you know, only software really makes sense for venture because that's the place where you can get these big multiples." But some of that is just its own belief eating itself.

(Tom Chi at 00:26:40) Right? We definitely have overvalued a bunch of software companies beyond what they're actually worth, and we've undervalued a bunch of hardware. But the stuff that I really like about physical businesses is beyond the fact that you don't end up with discretionaries, you end up with full winners or not, then because it's physical, everything in the business needs to follow the laws of physics. So you can actually get a huge amount of insight just by tracing down the four most important things to track in physics, which is matter, energy, time, and space. So if I can tell that a new approach is able to get a better outcome using half of the matter, it means that they're going to be unit economic advantaged.

(Tom Chi at 00:27:22) They literally use 50% fewer atoms to go make the thing, but the quality is as good or better. Well, we know that is going to drive a unit economic advantage. It's built into the core physics of the approach. Right? If we know they can get a better result using one-tenth the energy, we know that from the core physics that they have an advantaged approach.

(Tom Chi at 00:27:41) So we're able to go assess this from the very early days of a company. Right? A lot of times people say, "Oh, you can't really tell what the economics are going to be until you get to that first of a kind facility, and that's going to be $150 million worth of investing before they get there." No. No. Those people are incorrect. You can actually tell from the core physics approach in the first place. And you can actually tell the exact efficiency of the existing industries by analyzing their core physics approach. And if you look at their core physics approach, they will also tell you exactly where to innovate to win. So just to make this really concrete to people, 90% of the cost structure of all physical businesses fall in just three categories: feedstock costs, processing costs, transport logistic costs.

(Tom Chi at 00:28:25) And let me explain why 90% is going to fall in these three categories. The feedstock cost is the cost of all the atoms you're going to work with in order to make your physical thing. Because remember, you're making a physical thing, you will touch atoms. The processing cost is the cost of the energy and the labor that you use to go transform those atoms into the end products and outputs that you want. And then the transport logistic cost is the cost of moving the atoms that you're going to work with to you and moving the end products away from you.

(Tom Chi at 00:28:54) And you can see that how between those three things that's going to be most of the cost structure of any physical business. It's obvious. Right? But given that, when you actually see the physics advantage, the physics advantage of any new company translates directly into the unit economic advantages based on those three areas in the cost structure. Because if you're able to save on matter, what you're really saying is you're saving on feedstock.

(Tom Chi at 00:29:19) If you are able to save on energy, what you're really saying would be is saving on processing cost. If you're able to reduce, you know, time and space, then what you're really saying is you're also reducing transport logistics. And every industry already has a profile for how that 90% is played out. For some of the industries, like the petrochemical plastics industry, most of the cost is the processing cost because they need to do molecular cracking between 1,100 to 2,500 degrees Celsius in order to go break their core materials down to form their monomers, and then they polymerize their monomers into polymer chains. And then sometimes they add other ingredients like cross-linkers and all sort of thing.

(Tom Chi at 00:29:58) But at the end of the day, they all have this, you know, processing energy bottleneck of needing to heat something up to 2,000 degrees Celsius in order to make anything. And that actually tells us that that's exactly the weak point in the cost structure. If we're able to go find alternative polymers that are able to be produced either at low thermal processing—because 2,000 degrees Celsius is called high thermal processing or high-temperature process heat—so if we are able to find alternatives that are more compatible with the Earth that either use low thermal processing, so let's say 300 degrees Celsius instead of 2,000 degrees Celsius, or use mechanical processing in order to go produce the good, then this is going to categorically save on processing costs, which we know to be the fat part of the cost structure of that industry. So we have about 30 industries to go after, but we know exactly what the fat part in the cost structure is for all of them.

(Tom Chi at 00:30:52) And we know from physics first principles which types of technologies are going to be able to hit the fat part of the cost structure versus not. So the entire thing is honestly way more straightforward than software investing. The stuff tells you—the industries tell you exactly where you need to innovate to go save the cost out of the thing. The industries are not that discretionary. They all gravitate to the lowest unit economic cost full stop.

(Tom Chi at 00:31:16) So if you achieve that and you can produce that, then you will win the entire board. You don't end up in this discretionary race. Like, what's the market share of ChatGPT compared to Claude today? No. No. You win the whole thing. Go to any physical production around the world. You're not going to see a hundred approaches. You're going to see one or two.

(Tom Chi at 00:31:36) And those are the cheapest unit economics that we know how to make that physical thing.

(Joel Beasley at 00:31:40) That's kind of brilliant. I love that the market will actually tell you where to innovate through physics. That's awesome.

(Tom Chi at 00:31:51) There's a little bit because we did so many years of discretionary investment with software. And the users do tell you they like a thing more than another, but they can't decide when there's enough options that are about good enough. But at the end of the day, why use Gemini over ChatGPT or vice versa? I don't know. They all seem good enough to me.

(Tom Chi at 00:32:12) Right? And I work in this space. Now look, if we really get in the details and if you want to talk about the differential improvements in the models, who is the furthest along in sub-polynomial training and inference, sure, we can have a deep technical conversation about it.

(Tom Chi at 00:32:28) But in the grand scheme of things, when you get down to the consumer choice of why do I use Alexa versus Google Assistant, why do I use Claude versus ChatGPT, it's fully discretionary. And I honestly hate that about software. It sucks. It's tough for investors.

(Joel Beasley at 00:32:44) It could be very niche. It could be very like, there might be a specific thing that Gemini does that's very useful to a very small number of people that, like, that's why they use Gemini. But I love that this takes an unemotional, calculated approach to saying, hey, there's a physics bottleneck here. And if we solve the bottleneck by either finding polymers that can be heated at a lower temperature or improve the ability, like, the cost of getting to that temperature, whatever it is, that's affecting everyone. That's a thorn in everyone's side. And if we go in and solve it, they're not gonna think twice. Everyone's just gonna switch to that because it's the smart move. It's what you would do.

(Tom Chi at 00:33:25) It is also the way that all of the industries have advanced up to this point. So it's literally been the rules of the road for hundreds of years. If you look at every major innovation, why did the cotton gin make a big difference? Because, basically, the cost structure on the labor went down pretty dramatically with it. Why did we start using water wheels in order to mill grain and stuff? Because it was also better unit economics than the thing before it. Like, literally from the beginning, this has been the modus operandi of the whole thing. And I think it's because we had this, honestly, I think kind of stupid conventional wisdom in venture capital that only software makes sense for venture capital. The reason that that's so stupid is that it just shows how short our memory is. Like, the place is called Silicon Valley because all of the initial investments were in hardware. We literally invested in things that were silicon based, like physical atom things. That's how we built the entire venture capital ecosystem. And the fact that we had twenty years of software primary investing, and because of it, everybody forgot what founded this whole thing, that's insane. Right? For people to now pretend like, oh, hardware investing is so impossible. You could never make venture returns on it. It's like, that's how we made this whole thing, guys. That's the reason we're called Silicon Valley, not Lines of Code Valley. You know, we're not called React JS Valley. Right? We are called Silicon Valley because we did physical things. That is how venture capital was created, full stop. And when you win at those physical things, like, why is Taiwan such a big issue right now? Because they make 90% of the advanced chips. Like, when you actually get good at the unit economic, it's not easily replaceable. You win the whole pie.

(Joel Beasley at 00:35:14) Jensen didn't forget about that, though. He pushed forward.

(Tom Chi at 00:35:18) And made by a Taiwanese entrepreneur that did not forget and now one of the most successful companies in the world. Right? Did you see that interview?

(Joel Beasley at 00:35:26) With him on Joe Rogan?

(Tom Chi at 00:35:28) I'm not following the podcast too much, but I would expect him to do smart things because he's done many smart things.

(Joel Beasley at 00:35:37) How big is, like, your fund? What's the name of it? How can people learn about it?

(Tom Chi at 00:35:42) Yeah. Our firm is called At One Ventures. It's because it's about being at one with ourselves, nature, and the universe. The goal is to help humanity become a net positive in nature, and you can check us out at At One Ventures, atoneventures.com. And, yeah, things are going quite well. We have raised two early stage flagship funds. We also have kind of a leaders fund and a co-invest fund, and we are well on the way. We're about halfway done raising our third fund. So altogether, you know, we've raised on the order of $800 million across those vehicles. And I think by the time we get to the end of the fund three fundraise, we'll have raised, you know, about a billion dollars.

(Joel Beasley at 00:36:29) Very cool. Now was there any play on words with At One Ventures? Like...

(Tom Chi at 00:36:35) Yeah. It is on purpose. So, the entrepreneur, Stacy, from Evernew was actually the person who first suggested the name, but I really liked the name because the purpose of our firm is to help humanity become a net positive nature, which means, yes, we will make, you know, good returns for our LPs, and we have been. But we also are doing it for a little bit of a larger purpose. And for me, like, those five letters, A-T-O-N-E, if you parse it as "at one," then it has kind of the center of a lot of eastern spirituality. Right? Like, oh, you're one with the Tao. You're one with, you know, the chi, the universal energy of consciousness, da da da. And so that idea of being at one with things, at one with ourselves, nature, and the universe, it's a very kind of eastern sense of spirituality. But when you jam all the letters together and you type it in a URL, you can also read it as "atone." And if you think about the kind of Western spirituality, specifically the Abrahamic religions, right, then atonement and forgiveness and all that sort of thing is very central to those spiritualities. So I basically, it's like, oh, in five letters, we're able to go hit the two major spiritual traditions of the planet and kind of indicate that we're doing a little bit more than just making money for folks.

(Joel Beasley at 00:38:01) I love it. You know, in six months when someone asked me about Tom, um, the what I'm gonna take away, the thing I will remember is that he was one of the first people I've met who can is working on improving the world and making it better without being frou frou. It's like it's not lame. It's not, it's like it's very real. It's like we look at the physics, we understand the industry, and then we're just putting money where there's a huge advantage, and, you know, you explained your setup before. And I'm like, that to me is cool because it's being cool with like, there's always so much association with, like, saving the planet and being lame. Right? But you're, like, saving the planet and being cool. And I would tell you before this interview that that's not possible, but you have changed my outlook.

(Tom Chi at 00:38:52) Look. I understand why people would think it's not possible, and I also understand why people might think it's lame. And the issue here is when people need to make a lot of noise about saving the planet, then a lot of that is kind of performative. For us, I call us the most exciting boring firm on the planet because if you go look at how we actually help folks through manufacturing and how we get things to scale up for production, it is the most, like, close to the metal stuff that you could possibly imagine. It is the most literal practical stuff you can ever imagine. And on these interviews, I oftentimes need to hold myself back to be like, oh, should I really talk about the fixtures and bring up the lines and the tooling and all that sort of thing? Because that's actually what it takes to make this stuff. Now that said, if I'm talking with a very technical audience that also does physical production, I'm happy to let her rip because I want as many people that actually can make a difference on those things to go make that difference. And I do think our firm is discovering a lot of the right inroads to be able to get to those unit economics and a scalable production in ways that are also work better with the capital, the capital markets and the funding environment. Right? Because that's another piece of the equation you need to solve. But we're boringly ridiculously practical on all of the physical production and business model stuff because all of those businesses that make the physical world around us, there is no BS to have in them. Right? They basically need to physically produce the thing. It's not like, oh, you can imagine this thing being a thousand times bigger. That's why we've moved the multiple up so much. It's like, no. No. On the stuff we make, you don't need to imagine it. It's literally just cheaper and better, and it's based off of cheaper and better physics. It's faster, cheaper, better, greener, but it's a disciplined way of doing that. And it turns out that, honestly, I think if you have this type of scaffolding and you can approach it in this disciplined way, I think it's easier than software investing. And it's not because I don't know software. I was a professional software engineer for a number of years. I worked as an executive, you know, toward the top of two multibillion dollar software businesses. I know what it looks like to write it from, you know, the actual cube, or I guess people get to write from home now. That's nice. Um, you know, I know what it's like to author that stuff from scratch, and I know what it's like to run multibillion dollar businesses that realize, you know, the kind of potentials of software at scale. But I will tell you that even with all of that, I find the hardware stuff to be a lot more direct and straightforward because there's very little discretionary on the other side of the winning unit economic. If you have the winning unit economic, you will probably win the industry. But in software, you can have things, oh, my code runs 500 times faster. Well, I just like the design of this other one better. Right? That there's, the it's you can literally do the best engineering and it doesn't totally register in B2C use cases. And...

(Joel Beasley at 00:41:48) Well, you're picking stuff lower in the stack. Right? Like if I got a phone call and there was a credible energy company that was half the price of my current energy company, I'm so abstracted from the brand of energy. As long as it's reliable and known in the mark, I could switch it out and you would switch over very quickly.

(Joel Beasley at 00:42:04) Yeah.

(Tom Chi at 00:42:05) Very fast. And I love that because, like, actually, most money in the world, and this is probably a useful note for the audience as well. Like, you hope that most money in the world moves because people have these, like, deeply considered high quality diligence, that sort of thing. But, actually, most money in the world moves on the principle of the less bad place. So if I have, like, $5 billion invested in a particular direction that is yielding me 3.4% per year, if I can find a different spot where I get 3.8% per year, I'll move $5 billion like that. Now was that the best, smartest, highest use of $5 billion in the world? Did you do the most that you could with that money in the world? No. Of course not. And that's not how most money in the world moves. Most money in the world moves because it's a less bad place. What we're basically doing by creating the best unit economic in the world is we're making the cleanest technologies the best place for you to move the money to. And people make those decisions just like you're saying. Like, if I can get energy that is cleaner, and, honestly, they don't even care if it's cleaner. I'm gonna say 95% of the customers of our portfolio companies could care less about the environmental impact. They just see the headline on how much their margin improves. They're like, wait, when can we take a look at this? We'd be doubling our margins? We have nothing else that comes anything close to doubling our margins. Let's get into it.

(Joel Beasley at 00:43:26) And it pulls their attention.

(Tom Chi at 00:43:28) Yeah. Oh, yeah. The, well, the other thing is that most startups, and this is another reason that people complain about hardware companies, they're in this, like, B2B hell, you know, sales hell, where they're like they don't have something that's convincing enough. So they need to keep on knocking on the door over and over to be like, hey. I know we talked six months ago, but things have developed a bit more. Can you have another meeting? Right? And you're kind of, like, almost, you know, begging for the presence of these core executives, right, that can make the big decisions in the industry. When you are coming with a new physical approach which rewrites the unit economics of that industry, they're knocking on your door every six months.

(Joel Beasley at 00:44:06) Because they wanna look smart and they wanna save money. Yeah.

(Tom Chi at 00:44:09) And they literally have no other pathway to double their margins, triple their margins because these industries have asymptoted to their current physical production. Most of them have not changed their physical production in any meaningful way in thirty to fifty years. And think about how much technology and capability we've developed in those last thirty to fifty years. That has never been really looked at. This is a completely kind of underpicked area where there's going to be a huge amount, a huge number of wins exactly because it's underpicked.

(Joel Beasley at 00:44:44) That's brilliant. So you did your first two funds are done, and you're raising your third currently?

(Tom Chi at 00:44:50) Yes. Our first two funds, so the fund one portfolio is completely built out. We're doing a couple, you know, bets out of reserves as things unfold. But fund one is doing really well and not just compared to other climate funds, compared to all funds in its vintage. And fund two is, we just signed our first unicorn term sheet in fund two for a fund two company, and that's gonna pop us up into the top quartile. So we are doing quite well compared to all of venture capital, not just other climate firms.

(Joel Beasley at 00:45:23) So if you compare yours, that's a very good distinction. You compare your firm as far as getting returns, your funds compared to all other venture funds. You guys are on par or better.

(Tom Chi at 00:45:34) That's correct.

(Joel Beasley at 00:45:36) That's great, man. Congratulations. Making money and making the world a better place, you must sleep well at night.

(Tom Chi at 00:45:44) Well, definitely in terms of working on things that I consider to be some of the most meaningful things, then have no issues with that. I will tell you that our firm is a bunch of really technical thinkers, so we're in the details of things. So sometimes I do, like, wake up at night and it's like, oh, you know how we could get more yield on the line. Right? So I'll have that sort of thought, but that's less of a tossing and turning because, like, the reason I'm excited about the more yield on the line is it means we're really realizing the type of vision of, like, being able to get the best unit economics out there and clean up the industry by their own internal force. Right? If you try to go and do it from the outside, like punish them with regulatory or have all the consumers shame them, all that sort of thing, then, I'm not gonna say that never works, but it's really, it's really incredible. Yeah. It's not that scalable.

(Joel Beasley at 00:46:38) As humanity, that's not what we want. Let's just be smart.

(Tom Chi at 00:46:41) It might only work, like, 15% of the time. That's a very low hit rate. But when you actually have the best unit economic, everybody gravitates to the best unit economic. If you don't, you're out of business. Like, everybody else will surpass you if you don't gravitate to the equipment of production that has the best unit economic. So it's great business if you're good at producing the equipment with the best unit economics. And to me, it's like such a turn the crank approach to investing. Now that said, I wanna say it's this easy. It is this easy, but you do need to hire slightly different people than we have for most venture firms. Because most venture firms are people with pure finance backgrounds or if they have an entrepreneurial background and they built anything, it tends to be software. And you're not getting the rest of the view on things. Right? And it's not that software building isn't important. Software is one of the five major types of building. But when I was staffing on my team, then there's thousands of fields of science, but if you wanna be able to go touch everything that industry makes, it actually turned out there's only really five types of engineering that make everything. And that is electrical engineering, mechanical engineering, chemical engineering, bioengineering, software engineering. That's it. Everything in the world is, you know, one of those five types of people or a combination that actually making the thing. And your viewers might say, well, how about civil engineering? To me, civil engineering is just like really large scale mechanical engineering, but sure.

(Tom Chi at 00:48:03) If you want to fight me on it, it's fine. Then you could say there's six instead of five. But in the grand scheme of things, there's basically five types of building that make everything. And we've staffed the team so that we have representatives of all five types of building, and we've staffed the team so that we have more than one for each of the five types. That means we don't have a single source of failure on any of the evaluations for any of the types of building.

(Tom Chi at 00:48:29) And because we cover all five types of building, we can go after literally anything that can be made. Anything that can be made is going to be a combination of those five things.

(Joel Beasley at 00:48:38) That's brilliant. I love what you're doing, Tom. I'm watching the time. I want to make sure we get you out of here on time.

(Tom Chi at 00:48:43) We've got to finish up here.

(Joel Beasley at 00:48:44) Yeah. But let's do a call to action. What do you want the call to action to be? Invest in your fund, buy the book. What do you think?

(Tom Chi at 00:48:51) I guess that if you do LP work and you want to reach out about that, then do that. If you want to buy the book, then that's amazing. It actually hit the best seller list last week, so they're going to reprint the cover to say that. So if you want the limited edition that doesn't have a banner that says that, then buy it now. And look, if you want to spend no money, then I have a lot of free lectures online because I am very committed to trying to teach the best things that are working.

(Tom Chi at 00:49:19) And oftentimes, I do public speaking, but I often call myself the anti-entrepreneur—not because I dislike entrepreneurs. I support entrepreneurs as my day job. But most entrepreneurs get up there and they talk about what they're about to do. But given that, that means they haven't done it yet. And given that they haven't done it yet, the signal-to-noise of communication is kind of muddled because some of the things that they say they're going to do is not going to turn out that way. Other things, they won't be able to do, and other things they will do. It's great. But when I do public speaking, I'm almost always talking about things that we already did do. So I can just tell you exactly how to do it.

(Tom Chi at 00:49:57) So the signal-to-noise is like, yo, no. If you do this, then you will get this result. And you'll see that in the book, and you'll see that with a bunch of my talks online. If you just want to search for my name on YouTube, you'll see a lot of talks on product development, rapid prototyping, how to work in particular industries, and that sort of thing. And myself and the whole team are the types of people that have practically made these things in the past. I'm a named inventor at 77 patents. I've patented maybe 4% of what I've built. We do know how to actually make the stuff. And given that we know how to make the stuff, then the types of communications that we do that are meant to teach people to make stuff, then it really gets close to the metal for people that care about the work.

(Tom Chi at 00:50:38) And I think—maybe this is the right note to end on—when people say, "Oh, we're going to invent AGI or artificial superintelligence, and that's going to solve climate," at the end of the day, they're still going to need to do the stuff that we're doing right now, and this stuff is very hard to wrap an AI around. I know this because I've worked on AIs in the past. And as my brain tries to bridge that distance, it's like, no, you're still going to end up needing to learn the stuff we're learning right now. And that will probably be the thing that trains the machines if AI is going to have an important role in the climate fight. We're building the training data right now at minimum, but we are also practically solving a lot of the problems on the way there.

(Joel Beasley at 00:51:21) Brilliant. Well, I'm glad when I meet people like you. I'm like, I'm glad this world has some smart people that are making the world a better place. So thank you so much, man. I really appreciate you doing this.

(Tom Chi at 00:51:33) No, thank you so much. And I'm happy to hear from the audience. Also, if you're a builder and you have other better ways to build, that's the other thing I like about this space. Builders are very accretive with each other, right? And a lot of folks that I call talkers—they kind of fight with each other a lot, especially in the political space. It's like, "Oh, you're on the left. You're on the right. I'm not going to listen to you." But in the builder space, all of that falls away because we do agriculture, industry, all that sort of thing. It means I do a lot of my work in red states. But when you're in red states with other builders, we just love each other because builders need to have build integrity. The things that we make need to work. And whenever you're around a person that can make things work, it doesn't matter where they come from politically. You just love working with them because you're like, "Oh, you got this extruder to work. Oh, you got this conveyance to work. You got this robotic welding step to work." Freaking, let's work together. And this is the type of energy that I think is going to be the source of repair for us as well, right? Finding the commonality is where we have deep integrity with each other that we're able to build on it creatively as opposed to everything being some divisive, scattering fight no matter how minor the issue.

(Joel Beasley at 00:52:48) Respect from competency is going to make a comeback.

(Tom Chi at 00:52:52) Absolutely. And the future will be built by the builders, so you might as well get in with them now because that's where the action's going to be.

(Joel Beasley at 00:53:03) Tom, we made a podcast. How do you feel? I feel good. Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.