Capital Allocation with Blair Silverberg and Chris Olivares


Blair and Chris Welcome to the show. Thank, you good to be here. We're talking about capital allocation today and I'd like you to start off by describing the problems that you see with modern capital allocation for technology companies. I'm happy happy to start there. So I think it might be helpful to give. The listeners, a little bit of our backgrounds so I was a venture capitalist at draper. Fisher Jurvetson for five years I worked very closely with Steve. Jurvetson and we were financing are very MD intensive. Technology projects that became businesses things like satellite companies companies that were making chips to challenge the GP you new applications of machine learning algorithm so on and so forth and I think the most important thing to recognize is that the vast majority of technology funding does not actually go to those kinds of companies. The venture space is a two hundred fifty billion dollars per year investment space. The vast majority of the capital goes to parts of businesses that are pretty predictable like raising money in in investing that in sales, marketing and inventory or building technologies that have a fairly low technical risk profile, so the vast majority of tech companies find themselves raising money. From a industry that was designed to finance crazy high technology risk projects at a time where that industry because technology so pervasive you know really do the great work of of many entrepreneurs over the past twenty to thirty years, technology is now mainstream, but the financing structure to finance businesses not has not really changed much in that period of time. Yeah, and then I guess I'll talk a little bit. My my background is I came from consumer education sort of background, so direct to consumer, thinking about how you use tools and make tools that ingrained into the lives of teachers, parents students I was down in the junior class dojo before starting capital with Blair. We were working on the Earth thesis He. He was telling me a lot about this. The the date out. There exists to make more data driven in data rich decisions. How do we go software to make that easy to access in self service and sort of servicing the signal from the noise, and we kicked around the idea and I thought that they were just a tremendous opportunity to bring. What Silicon Valley really pioneered which is I think making software that is easy to use in agreeing to your live into kind of old industry fund raising capital Haitian. The kinds of capital allocation that exist there's. And debt, financing and different flavors of these. Of these things say more about the different classes of fundraising in how they are typically appropriated two different kinds of businesses. So. You have the main the main groups you know. Absolutely correct, so there's. Equity means you sell part of your business forever to a group of people and as Business Rosen succeeds. They'll get a share in that. Success and ultimately income forever. Debt means you temporarily borrow money from somebody you pay them money, and then at some point in time that money's paid back and you all future income for your business, so equities permanent, not permanent. If you think about how companies are finance like. Let's take the P five hundred. About thirty percents of the capital that S&P five hundred companies use to run. Businesses comes from debt. In the venture world that's remarkably just two percent. And the thing that's crazy is this is two percent with early stage seed companies, also two percent with public venture, backed companies in places like the best cloud index, which is like a one trillion dollar index of publicly traded technology companies started their life, and in with injure backing many of them SAS companies, these companies, also just two percent finance with debt, but nonetheless within these these classes, the reason it's obviously economically much better for a business and pretty much every case to finance itself with debt because it's not. Not It's not permanent, and it can be paid back. It's much much cheaper to use debt. That's why you buy a house with a mortgage show. You know you don't sell twenty percent of your future income forever to your bank help you buy a house, but the reason that people use equity comes back to the risk profile so just like. If you lose your job and you can't pay off your mortgage. The bank owns your home. Same exact thing happens with debt in so restorick Louis, if there's very low. Certainty around the outcome in typically early stage investment you're you're doing a lot of brand new are indeed you have no idea if it's GonNa work you cope. You know over time that you'll be successful, but there's really quite a bit of uncertainty equities a great tool because you're. You'RE NOT GONNA lose a business, you know everybody can basically react to a failed. Are Indeed project. Decide what to do next had saints. Equity is kind of the continent tool for high technical risk, high uncertainty investments, and then debt is basically the tool for everything else, and it can be used as most companies do for. Ninety percent of The places that businesses are investing so if you're spending money on sales and marketing, and you know what you're doing and you've been running campaigns before. That were successful, very. Little reason you should use equity for that if you're buying inventory if you are a big business that's. Reach a level of success that on. Means you have a bunch of diversified cashless. Coming in businesses might take out dead on business kind of overall, so it's less important what specifically you're using the money for, but it's important to recognize that most companies are financed roughly fifty fifty equity versus dead, just just intra back companies that. That are kind of uniquely Equity Finance. Scaling a sequel cluster has historically been a difficult task cockroach. DB Makes Scaling your relational database much easier. COCKROACH! DVD's a distributed sequel database that makes it simple to build resilient scalable applications quickly. COCKROACH DB is post grass compatible giving the same familiar sequel interface that database developers have used for years. But unlike databases scaling with Cockroach DB's handled within the database itself, so you don't need to manage shards from your client application. And because the data is distributed, you won't lose data if a machine or data center goes down. cockroach D is resilient and adaptable to any environment. You can hosted on Prem. You can run in a hybrid cloud, and you can even deploy across multiple clouds. Some of the world's largest banks and massive online retailers and gaming platforms and developers from companies of all sizes, trust cockroach DB with their most critical data. Sign up for a free thirty day trial and get a free t shirt at cockroach labs dot com slash save daily thanks to coach labs for being a sponsor and nice work with cockroach DB. The capital that is being steered towards a recipient. It's often originating in a large source, a sovereign wealth fund or family office in it's being routed through something like capital allocators cater like a venture capital firm for example or a bank. How does this capital get allocated to these smaller sources? What is the supply chain of capital in the traditional sense? You know it's kind of funny to think about capital and things like the stock market in the form of a supply supply chain, but this is exactly how we think about it so at the end of the day. Capital originate. In somebody savings, basically society savings right you. You have a retirement account or your population like you know in in Singapore and Norway with a lot of capital, it sort of accumulated from. From the population and these sovereign wealth funds, or you're an endowment that's you know managing donations of accumulated over many many years, and ultimately you're trying to invest capital to earn a return and pay for something pay for your retirement pay for the university's operation so on so forth so that's Capitol starts, and it basically flows through the economy in theory. To all of the economic projects that are most profitable, inefficient for society, and so, if you step back, and you think about like how how is it that the American dream or the Chinese Miracle Happen? You know in in both of those cases different points of the last hundred years. Why is it that society basically stagnated? You know the world was a pretty scary. Scary place to live in up until about seventeen fifty, the industrial revolution started. Why is it that you know basically for all of human history? People fought each other for food and died at the age of thirty or forty, and over the last two hundred fifty years that it's totally changed. It's because we have an economic system that converts capital from its original owners. Diverts it to the most productive projects. which if they're successful, replace some old more expensive way of doing something with newer better way and so I think when when I described that like you know I, think most people can step back and say yeah, okay I. kind of see how capital flows through the system, it goes automatically to someone making an investment decision like a venture capital firm ultimately gets into the hands of the company company decides to invest in creating some great product that people love. Let's. Let's say like Amazon and then everybody switches from you know buying goods at some store that may or may not be out of you know may or may not being stock to the world's best selection of anything you'd never wanted. The most efficient price that's society gets wealthier basically through these these kind of steps in these transformations, but it's asking if you step back and think about it like nobody actually thinks it's processes as efficient as it could be like. We asked people all the time. People were interviewing journalists companies. We work with sewn. So how efficient do you think world's capital allocation is? I've never met a person that says it's pretty good. You know we're like ninety percent of the way there. In fact, most people think it's pretty inefficient. They think of companies like you know we work, and some of the more famous cases lately of of Silicon. Valley back businesses that that totally. underwhelmed disappointed. Their initial expectations and I think most people admit that the efficiency of capital allocation is either broken or nowhere close to achieving its potential, and so we basically we'll talk more about our technology and how we do we do. We basically think of this problem our problem to solve. There's an incredible amount of Apache inefficiency in how data that goes from a project or a company, ultimately funneling up to an investor flows, and so you know it's hard to place blame because there's so many people in the supply chain, but. But I think it super clear that if it's difficult to measure whether or not a project or a business is good at converting capital into value in wealth, and you know products that people want, it's nearly impossible for society to become really good and efficient at allocating its capital, so we're we're here basically to make the data gathering data transformation visualization communication of what's actually going on under the out of business as efficient as possible and you know from that, we thank some great things are going to happen to the economy. Goes a little bit deeper on the role that a bank typically plays in capital allocation. If you think about our bank works like let's take. Let's take a consumer bank that most people think about you gotTA checking account. Right, now you've got some money in that checking account. That account actually takes your money or dot and most people know this your dollars sitting in that account. You know just waiting around. You'd withdraw them. Your dollars are actually rolling up into the bank's treasury. There's somebody at the bank working with the regulators to say hey, how much of this money can we actually put into things like mortgages, commercial loans, all of the the uses of capital that society. Has In some some effort to. To, move the world forward and make the economy efficient, and so those deposits basically roll up into a big investment fund, and there's ratios that regulators set globally that say those dollars needed to be kept in reserve, versus how many are actually able to be invested, but with the portion that's able to be invested. It's there to fun. You know building a house to fund a business back -Tory to fund sales and marketing or inventory procurement for some other business, and so a bank was was basically the original investment fund, and a bank has unlike venture funds and other sources of. We typically think private capital. The bank has tricky. Problem were any moment all of the depositors holding the checking accounts could show up and say hey. I want my money back and so that's why banks have to deal with reserving capital predicting the amount of withdraw and classically everybody wants her money at once at the worst possible time, and so banks have to deal with quite a bit of volatility now if you take an investment fund on the other hand. Totally totally different structure, so your typical venture fund will have money available to it for a period of ten years from you know typically these larger pools of capital. We talked we talked about so very rarely. Individuals are investing retirement savings in venture funds, typically sovereign wealth funds down that's. Basically pools of that individuals capable. Win One of these funds makes a commitment to a venture fund. It'll say you've got the capital for ten years. You've gotta pay back. You know as investments exit, but other than that will check in ten years from now. We hope that we have more than we gave you the star with and there there's no liquidity problem because the fun has effectively carte blanche to keep the money invested until some set of businesses grow and succeed and go public and make distributions so one thing that's fascinating. The Tappan in the last twenty five years is private capital capital in the format of these kinds of funds. Have just grown tremendously and so today. There's a little over five trillion dollars. Of private capital being allocated in this way to think like buyout funds venture funds so on and so forth. Funds don't have the liquidity problems of banks. They can make much longer term for looking investments. This is created tremendous potential to make the economy more more efficient by taking out the time spectrum. You know this is why venture investors can do things like finance spacex or Tesla. Really. Build fundamental technologies in the way that a bank never could so this is an amazing thing it. However leads to a very long. You DAK cycle, so the incentive goes down when you take out the time line over which investment needs to pay back. To carefully monitor and understand what's going on in the business day today, so it's pretty interesting thing about the different pools of capital. There's not not to. Make it sound too confusing, but I think everybody will admit that the financial markets are incredibly diverse complicated we track basically about fifteen different kinds of capital, and they're sort of pros and cons with each one, but you know a bank is one. A private fund is wanted insurance companies balancing as another. You've got things like ETF and public vehicles that hold capital so there's quite a bit of complexity and the the structure of the financial markets. All right well. That's maybe the supply side of Capitol on. All kinds of middlemen and all kinds of different arrangements, but ultimately there is also the demand side of Capitol, at least from the point of view of companies getting started which is. Startups or computer in later stage with the maybe they're not exactly considered startup anymore, but they're mature. These companies have models for how they are predicting. They're going to grow, but oftentimes these companies are very. Lumpy in terms of how their their revenues come in how closely their predictions can track reality. So how do technology companies even model their finances? Is there a way to model their finances? That actually has some meaningful trajectory. Sure so first. Companies you know need need a base think of all the places that they're spending our money and. We're pretty. We Do I. Think a pretty good job of organizing this and making it simple so when we look at companies and we can, we can talk more about how the the cabinet machine operates, but when we look at companies, we basically think they're only a handful of places of money. Get spent you spend money on. Short term projects that you hope proficient things, sales and marketing. Houston money on paying for your sources of financing like paying interest on debt, making distributions to your investors, and then you spend money on everything else and everything else can be designing software building products on, and so forth, and so if you break the demand for capital down into just those three buckets. And look at them that way. Some pretty interesting things happen. The first is for the short term investments that you hope productive. You can track pretty granular nearly whether or not they are, and we'll come back to that. For paying back your investors, you sort of know exactly how much you're paying your investors so a pretty easy thing to track, and then for the operating costs you know most people will help us. Apax, that you're paying to keep the lights on things like Renton the your accountants, the CEO salaries on and so forth these are these are table stakes expenditures. You need to stay in business and so. Amongst each of those three things, there's different things that you wanna do to optimize and I'm happy to go into more detail sort of go through each one. If you think that'd be useful. Yeah Bliss a little bit more about about how these companies should be a modeling, their revenues are that is meaningful to model their revenue so that you can potentially think of them as targets for for capital allocation so. If we think about. Understanding what company might be a viable recipient of capital? How can you accurately predict the trajectory of that company, or or do they? Would they present a model? Would they develop a model good through a little more detail? How a company would serve justify? It's need for capital. So typically what what most companies do and this is not terribly useful or accurate, but I'll tell you what most people do I mean by the way like how central the entire economy predicts, predicts demand for capital works like this. Companies take. Their income statement on their. Balance Sheet historically. And they they basically have this excel file got a bunch of you know, rose and have different things like my revenue, my you revenue that sort of linked or my expenses that are linked revenue Mukasey could sold so on and so forth, and they grow each of those rose by some number that they hope to hit so if you want your revenue to double next year, you'll say my revenue one hundred dollars today I wanted to be two hundred. Hundred dollars twelve months from now I'm just GONNA draw a line between those two points and every month. There will be some number that's on that line, and that's why monthly revenue I want my expenses. You know everyone knows. Expenses are going to have to go up if my revenue goes up but I don't want them to go up as much as my revenue, so I'm going to draw a line. That's you know somewhere less than a doubling. and. You pull these lines together on one big excel file and there's your you know they're your corporate projections. In general, this is true for big companies small companies, but that's not actually how. Company revenue works because if you go back to the three categories, we talked about before, and you just focus on the one that talks about the short term investments. The. Way Company Revenue Actually Works is a company this month. Let's say they spend one hundred dollars on sales marketing. Well. They're hoping to get a return on that sales marketing, and so they're hoping that in the next you know six months. That's paid back. Twelve months that's paid back. You can actually track every time they spend money on sales and marketing. how quickly it gets paid back so it's that level of precision that can accurately predict revenue, and so what we do is we basically just get a list of every time? Money was spent on one of these short-term investments, so you sales and marketing for for an example, and then we get a list of all of the revenue that was ever earned. And we attribute between both of those lists causing effect. And we do that using a bunch of techniques that are pretty commonplace in your typical data, company or machine learning company. We use some math things like factor graphs. We use simple kind of correlations. We have You know a whole kind of financial framework to. Guess. What attribution should be because you learn a lot as you see different businesses and you see a bunch of different different patterns, which you can basically cluster on, but it is this linkage between spending on something like sales and marketing emceeing seeing revenue, go up or down, but makes or breaks a business, and you want to look at it and I is. Not a bundled. Entirety which is how financial projections are typically built? Okay, well! Let's talk a little bit more about what you actually do so if you're talking about early stage technology companies. Describe how you are modeling, those companies and how you are making decisions as to whether they should receive capital. When a company comes to capital they they come to our website. They sign up for this system that we built which which we've called the capital machine. And the first thing that they do is they connect their accounting system their payment processor typically, so think like a strike, and then sometimes they'll provide other things like a pitch deck or a data room, or whatever other information they have prepared. The system pulls down. All of the date in the accounting system and the the payment processor, and we look at other systems to these are the two key ones that all all dive into detail, and so, what ends up happening is from the accounting system. We get a list of all the times. Businesses spend money on these things like sales and marketing that we were talking about before. From the payment processor we get a list of all the revenue transactions in crucially we get it at. The level of each. Each customer payment, and so you know we scrub I all we really care about is having a customer ID, but once we have data at that level. We can start to do this linkage and say all right look. You know this business spent. A million dollars on sales and marketing and March of two thousand eighteen in April of twenty eighteen, and we saw revenue grow by twenty percent. That was a pretty substantial chain. You know what actually happened here. You can typically identify the subcategories of sales and marketing and start to do this link between these two, and this is really the you know the magic behind our our data science in our team pairing with our engineering team to figure out this problem and solve away that is, that's robust. Bud once we have these two data feeds, and the system goes through, and does all of these attribution. Populations were able to present that back to accompany a pretty clear picture of what's going on, and so we'll say things like hey. Your Business is pretty seasonal, and in the summer is when you're typically more more efficient at converting your sales and marketing dollars into growth so I, you want to finance growth in the summer. The second thing is only about eighty percent of your businesses financeable. There's twenty percent where you might not know it because you're not looking at this level of detail, you're busy building your business, which is exactly exactly what you should be doing, but Twenty percent of your businesses, not efficient. You're spending money on on your sales and marketing categories, product lines, and CETERA that just shouldn't exist and so if you get rid of those. If you double down on the part of Your Business, it is efficient. Then we predict your revenue will be act fifty percent higher, and we'll tell you exactly how much money you need to invest to raise money to to raise the revenue by fifty percent. We give you a bunch of charts that allow you to see how history and projections merged together and dig down. Inspect how we do that linkage to make sure you agree, but. This is what the capital machine does at its core. It Converts Company data into a fully audited completely transparent picture of. How business works where it sufficient where it's not efficient. And then that's where our technology stops, and where balanced she comes in, and so we then take this information, and we make balancing investments directly in companies, and so primarily at this point we lend money to technology companies that we see from their data are eligible for non dilutive funding. We make capital available to them directly. We basically allow them to access it through the capital machine. We use one system to communicate changes to the business. No keep both sides and form so on and so forth, but this is the kind of analytics layer that's essential to making these capital allocation decisions more efficient, and so I think you could imagine a day at least for us in the not too distant future when it's not just US using our balance sheet in this tool to make investments, but in fact, just like excel, every investor can benefit from a similar level of analytics and transparency, as can companies by getting more accurately priced faster access to capital less friction so on and so forth. Get Lab commit, is! Get labs inaugural community event. Get Lab is changing how people think about tools and engineering best practices and get lab commit in Brooklyn is a place for people to learn about the newest practices in devops, and how tools and processes come together to improve the software development life cycle. Get Lab commit is the official conference. Forget lab. It's coming to Brooklyn new. York September Seventeenth Twenty nineteen. If you can make it to Brooklyn, on September Seventeenth Mark Your calendar, forget lab, commit and go to software engineering daily dot, com slash commit. You can sign up with code commit s E. D.. That's COM MIT S. E. D.. And Save thirty percent on. Conference passes. If you're working in devops, and you can make it to New York. It's a great opportunity to take a day away from the office. Your company will probably pay for it, and you get thirty percent off if you sign up with code, commit S, e. There a great speakers from Delta. Airlines Goldman. Sachs northwestern, mutual, T, mobile and more. Check it out at software engineering daily Dot Com slash, commit and use code. Commit S. E. D.. Thank you to get lab for being sponsor. The inputs specifically if you think about a model for determining whether or not, a company should should be eligible to receive capital. I'd like to know how the the models are built. The the data science models that you're building are constructed from the point of view of the inputs. So how are you determining or how do you like company comes to you? How do you turn that company into some structured form of data that you could put into your models and determine whether it's worthy of capital. Yeah I mean it comes down to what what the data is your down so when we talk to a system like striper transaction records system, you know that that's the revenue of the company now where things get interesting when we connect to balance sheets in penalizing, it's of accompanying really onto understanding. Weighing. What exactly these numbers mean, and that sort of where we made our pipelines were built from the ground up to give us that granular. Of A company's cash family revolutions. Where's the money going where they allocating? And it's savable greenway or you once. What do you understand that data through that Lens? That let's build pretty sophisticated financial models Linda. And you know as soon as you have the picture of Company You can really do a lot of flexible analysis on the back leg distributed computation. Come stuff that you would never be able to excel and quite frankly a lot of these companies don't have the stacking internally or really the tools to understand for themselves, so you'd be surprised it you know when we surface this analysis back to the company by virtue of just being transparent on how we're making decision how it is perceived their business, the signals that were uncovering. These operators the CEO's the CFO's that are really focused on building company. Really surprising. They're really making these insights really transforming. How they think they should have capital. Should invest growing business. Are there any? Sources of Third Party data that you can gather to improve decision making. There are at a macro economic sense, and so it's actually quite useful to look at public company performance and say hey. SAS businesses in general. Most people notice, but facilities in general are seasonal in the fourth quarter. Budgets basically expire and people come in, and they buy a bunch of SAS. Software and so to take concepts like that basically shapes of curves, signals and apply them to private company. Financials is useful. Crucially though there is no private company. Data repository of any kind like it just doesn't exist, and you know notoriously even even with small businesses. It's actually quite quite difficult to get access to any sort of meaningful credit data, and so, what ends up happening is these aw. These businesses. Give you a picture of their business directly as an investor and you have to interpret it directly, and that's basically how this works totally unlike consumer credit, there's no credit bureau that people paying so most investors are analyzing the state and excel. Excel notoriously breaks when there's about a million cells worth of data, and so we've got this great visualization showing our data pipeline, and it's basically a bunch of boxes, and there's a little tiny. Tiny box in the bottom of corner that's excel, and there's a bunch of other boxes across the entire rest of the page that are nodes in our in our distributed computations, but accelerate very very limited, and so it makes it impossible to actually understand what's going on in business from the source data, and it's at the source that you see this variability in this linkage between profitable capital allocation decisions in unprofitable capital allocation decisions. Describing more detail, the workflow so a company comes to you and they're going to put their inputs into the. Would you call the capital machine? What does that workflow look like in a little bit more depth? Yes when they come to the website, they creighton count much like you would on. Twitter facebook account. When your details your email, you terrify your email, and then you on what's recalling like the capital portable on there? You have et CETERA. Tools to connect your sins record and these are typical offload. So you know people are very familiar with you. You know you say hey, let's connect by quickbooks you in your credentials and sort of be as secure way, and you click okay and the system checkmark by your quickbooks in the system start pulling that data out of regular cadence and. Depending on what system you're connecting you of the characteristics of that's not go systems of record, and how much data you have you know. The data's available anywhere from ten minutes to a couple of hours later and you know once we have Dr. System, we run that through our partake analysis pipeline in the users as a company. You get you get charged. In Tableau kind of call it, the insight Saban's these refused that we think would be helpful for you as an operator company understanding about Your Business in separately. We also get views of that data that are useful to our our internal investment team. Whoever is looking to capitalization systems? Are there certain business categories that are a better fit for modeling in better fit for the kind of. Predictable capital returns that you can, you can expect with the investments that you're making so like you ride sharing or Gig economy businesses or some businesses. What are the categories that are the best fit? Say Very few categories don't shit from the from the perspective of of linkages, but they're certainly models at their easier to think through and easier to understand, but our our system can underwrite today A. Lease on a commercial aircraft, a fleet of ships and Insurance Agency ask company the most important. Thing about our system is that the financial theory that underlies it is very general, just like p. e. rate is very general, and so that's kind of sounds crazy like. A lot of. A. Lot of people say what what businesses the best fit for your your system and you know it's kind of like asking what businesses the best for Warren Buffett like Warren. Buffett is a generalist. In any business, and he has a framework in his own head to figure out how to make ship comparable to American Express our assistant has a very similar framework. It just operates at the level of transactions instead of at the level of financial statements, but certainly within. That framework there's some examples that are just easier describes I think like you know thinking through the fishing of sales and marketing something. That's a lot more obvious than thinking through like the stability in refurbishment of commercial aircraft parts, which is a key question you know. Pricing pricing refurbished parts, which is a key question if your financing commercial aircraft and Our team, the ambassadors that use the capital machine internally which we primarily do internally do a little bit of partnering with without the groups to to use this as well. These people are all specialists in some particular area, but it's crucial to understand. They're looking at the exact same chance as all the other specialists and all the other areas, so it's like literally the the Fast Company and a commercial aircraft will have the same series of charts at investors. Are there two two draw their conclusion? Is the question for Chris. Can you describe the stack of technologies that you built in more detail? Yeah Yeah. Of course on the front, we are react type script, xjs, you know everything is on aws, and in the back, and we're. We're all python, and in really the reason for that is if you're doing any serious machine, learning or data science today can't really get away in python stack, so we're all python them back in. We have flasks. As a as our API late here and That's the that's a high level. And get a little bit more detail about how the data science layer works. Yeah, yeah, yeah, of course, so we put on the dea into basically a data lake the that goes down into Ardito pipeline in that's all air orchestrated on top of each called airflow, and we use a technology called desk for are distributed computation, and I think that this is a good choice. Choice for us at this moment you know I see us doing a lot of work on. You know using a spark in other distributed technologies in the future and his team and it turns out that when we pull this data down organizing the data was really important to us as we build a lot of attractions to make accessing that data, really easy for quantitative analysts. Important central to our whole technology is that we're able to do a lot of different financials experiment very quickly on top of this so the the implications of that really cascade down all the way into. You know what technologies where choosing how we structure our delayed. Even even how strokes are teams, so it really is brought up locations across all product. How is it when you're analyzing company that you have enough data that it warrants a spark cluster because I can imagine? The financial data around the company. How can there really be that much data to analyze how you do surprised in a lot of these transactions systems taking up the companies have been around a couple of years and their direct to consumer. These data sets can be can be pretty large. You know we're talking about in the millions and millions and millions of transactions that were pulling down and storing. Storing and that just on a per company basis. You know that's not even talking about if we wanted to. Benchmarks Cross companies, and also if we want to do scenario analysis, so you know one of the things we was part of a pipeline is take this data, and through like nine ninety nine hundred thousand simulations to understand the sensitivity of different variables on the performance of Your Business and If, you're starting out with starting that already large. Sort of a multiplying effect. On how much data the system is the old process? is you go through those different stages? And, can you tell me a little more detail? What would a typical spark job? Look like for a company that you're assessing. Yes, so first episode is ribbon. Our our financial didn't ingestion parts, so we download something on the order of you know forty fifty bytes of Tim's action data for for a company. We have to do all the work to interpret and understand what that means in reorganized that data in a way that are downstream analysis and primitives can. Make sense of and use for useful analysis so really the first step at this point job is is transformed the datum some it's useful, and then there's all the work on what are the clusters in order to machines and analysis in the computational. Resources needed to run simulations. You know not not just say local computer locally owned of fall over the only about thirty to sixty four gigabytes of Ram what league, so that's where workflow comes in creating easier faces into data, clusters and being. Should you know when you run a job? You know when it fails. You know it's done. You know when the team can't okay. This part of analysis done I had intermediate date asset to do more analysis on now get back to work is a lot of the time we spend developing internal tools to make. One other thing that'll mentioned that I think's important is. A lot of the underlying technology in our data pipeline it's no different than like what a tableau or you need. Traditional BI business would have access to, but what's fascinating when you have a vertically specific domain so financial data in our case you can make a lot of interpretations about the date of the let you do much more intelligent things, and so for example we. Don't have to make your own charts as a user of the capital machine. We make all the charts for you can of course. As a business we work with. Give us ideas for charts. You can mock up your own. We we basically have an interface for for business. The I team's to to write some code if they if they want to bought when you have clients who are thinking about financial risk, financial attribution across all of the companies that we see distilling that down into a series of indicators that are detailed, but generalize -able, and then publishing that back to all of the companies that use the capital machine to run their own capital, allocation, decisions and access, external fundraising and capital. Some pretty amazing things happen in so it's only with a vertical view. You actually having these we, we call our data scientists Kwan's, but but actually having these people who you know typically are graduate level economists, thinking for the first time about using transaction level data in their analysis, which is notoriously not not available to to normal economists that you get the kinds of insights and analysis the actionable for businesses, and then in terms of the data pipeline that then means we actually store a bunch of intermediate data that's opinionated in that way, and that makes it much faster to access much easier to benchmark much more useful across a network of companies, versus just that isolated excel model that. Explains only one business. One thing I'd like to ask you about. Capital intensity so there are kinds of businesses that are capital intensive for example where you have to pay upfront for a lot of ridesharing rides, and you know as Uber or lift. His has known in much detail. You allocate all this capital two things to subsidize rise because you try to win a market, there's all kinds of other capital intensive businesses. How does capital intensity change? What makes sense with regard to the equity financing the debt financing that you are shepherding for these companies? That is a great question and be because of where you focus in your audience. You totally get the most financiers don't so. The first point exactly like you said. Capital intensity means a business consumes a lot of capital. It doesn't mean a business has a physical factory or plant or railcars, so it is absolutely true exactly like you said that there are a lot of tech businesses that are incredibly capital intensive. If you are capital intensive business that means UNI especially if you're growing, you need to raise a lot of external capital, and so it is even more important that your capital or a big portion of your capital base is not dilutive. That's that's just essential. Table stakes because what you see with these businesses, the ride sharing companies are great. Example is by the time one of these things actually goes public the early owners in the business on a very very very miniscule. KEESA that business, still if you contrast that to company like Viva Systems which I think is one of the most capital capitol efficient businesses in venture history, I think that this race something like twelve or fifteen million dollars total before it went public in a at a multi billion dollar market cap. So capital intensity. Is a synonym for dilution your own way less. Than you think when you exit entities even more important that you figure out a way to raise capital non ludicrously upfront. Some broader questions zooming out in in getting your perspective. Do a thesis for what is going on in the economy right now where you look at. The fact that We have. Obvious pressures to. Reducing the size of the economy through the lack of tourism, the lack of social gatherings while the stock market climbs higher and higher, and it appears that the technology side of things is almost unaffected by Corona virus is there. Is there a thesis that you've arrived at or or their set of theses that through conversations with other people, you've found most compelling. Sure the most important thing to realize about the stock market is that it discounts all cash flows from all businesses in the stock market to infinity, and so the value, the stock market about eighty percent of the value. The stock market is. Pretty far into the future like more than three years from now, and so if you believe that the current economic crisis and this is why there's always a. At least in the Western, world, last two hundred fifty years after an economic crisis. If you believe the crisis will eventually revert, and there will be a recovery, then it only makes sense discount stock market assets by anywhere between ten and twenty five percent. If you believe businesses fundamentally going to go out of business because of this crisis, that's a different story, but that explains why something as terrible as Kobe nineteen and a pandemic. Only discount the stock market by by roughly thirty thirty five percent in a in March, but that's not what's actually going on today as you mentioned and so stock market prices now have completely recovered. That is something that we think is a little bit of out of sync with reality but I. I mention you know we're not. We don't spend too much time about the stock market beyond that we just look at you. Know Private Company fundamentals. We try to understand what's actually going on in individual businesses across all businesses that are network to see what you know what we can understand, and you know what kind of conclusions we can draw, and so if you take that Lens and you actually look at what's happening to businesses due to Cova nineteen, it's fascinating. Some businesses like think the food delivery space have gotten a lot more efficient, so those businesses lot like ridesharing businesses back twelve months ago, there was sort of a bloodbath between bunch of companies competing in local markets to acquire customers all all fighting Google and facebook console, and so forth you subsidies drivers, etc.. That's essentially stopped. These businesses incredibly profitable, the cost acquire customers has fallen by more than half a lot of cases. The channels were slot less competitive, and so if you're running one of those businesses. Now is a great time to be aggressively expanding. Weird things like commercial construction businesses. They're actually a handful businesses that we've seen do things like install windows and doors and commercial buildings whose businesses have accelerated because all of these buildings are closed down. Construction project timelines have gotten pulled up. All of these orders are coming. Do in they're you know sort of rapidly doing it solutions? There's obviously a bunch of other businesses have been that have been hurt by by the pandemic, but our general thesis are we've studied. Pretty detailed way the Spanish flu in nineteen eighteen, you know. These things eventually go away. There will be a vaccine. Economy will get back to normal, and as long as we can stay focused on working through this as as a society and of maintain our our fabric of of kind of economic progress then. DESAGUADERO values today will eventually make sense just sort of a question of of win for the stock market, and then if you're if you're actually running business in thinking about your own performance in isolation, really being clear about is now the time to invest and grow my business now the time to be very careful with my expenses interest, get through this for the next year or however long it takes for there to be a vaccine. So the way to think about your company, if I understand correctly if I was to to put in a nutshell, is that. I think of you as a data science middleman between large capital allocators, and and start ups deserving of capital, so the the sovereign wealth funds the banks the I guess. Funds of funds. These kinds of sources are essentially looking to you for guidance on where to direct the capital, and you're on the on the other side, absorbing data and creating opportunities from these startups to source the good directions of that capital. Just wrap up. Would you put any more color around that description or or refining anyway. Yeah I mean I. think that at the core of what capital is is where the. Core Technology Ambler of sort of. The private market if you think about public markets today, you've clearing-houses like the New York Stock Exchange, and you have companies that provide analysis on top of that like Bloomberg, you know we see a tremendous opportunity to shift the paradigm where you know the place where all the financial transactions happen. is also the place that collects the data improvise information for those making these decisions and yeah, so I think capitals really at the center of making a transparent technologically enabled financial marketplace. Guys. Thank you so much for coming on the show and discussing capital, and I guess one last question is. Do you have any predictions for how capital allocation for startups will look differently in five ten years? Sure so! The first prediction. And this is happening now. I mean the the infrastructure is. In place both within. And others. Most startups fairly early in their life. Think is equity only way to do this and. So. That's a cultural shift. That's that's already happened. People are starting to ask that question. The second prediction is. Seed and series a funding will be entirely unchanged. After series. There'll be a bifurcation between businesses that. Are Really. Capital intensive gigantic rnd projects think like SPACEX. The series, B. C. d. e. enough are really about building and launching a rocket. Those businesses will by and large not. Turn outside of equity to finance themselves, but there's very few of those businesses. Pretty much every other business businesses that you see raising a series B. Serie C. Will like any normal business in the entire rest of the economy raise maybe half of that capital nine allegedly either in the form of debt. Royalty financing factoring all of the other instruments that normal companies use to finance themselves in the void delusion that will happen roughly three years her. Now that'll that'll kind of we'll see obvious obvious signs of that from very early very early base, and then the final the final thing is. Steve Case talks a lot about this. With the rise of the rest, he's got this great venture fund that invests explicitly outside the coast, so kind of the rest of America and we've seen that there's there's a pretty dramatic distinction between being a coastal business non-coastal business from capital access perspective, but there's no distinction from an actual performance perspective, and so we'll start to see some of the regional. Differences in bias sees around where capital flows, go away. And so I would maybe put that on a five year timeline like raising capital is actually much more predictable, much less biased, and that's great back to the beginning of our conversation. That's great for the economy I mean every project or business that can convert capital, two products and services that people love should get finance. No questions asked doesn't mean it doesn't matter what the color of your skin is. What background you have whether you went to college didn't go to. College doesn't matter. You have a business with data that can prove whether people love it

Coming up next