This Week in Startups - Navigating the AI Boom | Startup Finance Basics w/ Kruze's Scott Orn

Episode Date: October 10, 2024

Todays show: Kruze’s Scott Orn joins Jason on the latest edition of Startup Finance Basics! In this episode, they break down the unique aspects and dynamics of the current AI boom, contrasting AI startup revenue, costs, and growth rates with traditional SaaS models (4:11). They also dive into AI tools' impact on productivity and labor efficiency (8:12), the accounting challenges and operational issues AI startups face (14:27), and more! * Timestamps: (00:00) Kruze COO, Scott Orn, joins Jason (1:06) Unique aspects and dynamics of the current AI boom (4:11) AI startup revenue, costs, and growth rates vs traditional SaaS (8:12) AI tools and their impact on productivity and labor efficiency (14:27) Accounting challenges and operational issues in AI startups * Check out Kruze: https://kruzeconsulting.com * Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.com Check out the TWIST500: https://www.twist500.com * Subscribe to This Week in Startups on Apple: https://rb.gy/v19fcp * Follow Scott: LinkedIn: https://www.linkedin.com/in/scottorn X: https://twitter.com/scottorn * Follow Jason: X: https://twitter.com/Jason LinkedIn: https://www.linkedin.com/in/jasoncalacanis * Great TWIST interviews: Will Guidara, Eoghan McCabe, Steve Huffman, Brian Chesky, Bob Moesta, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarland * Check out Jason’s suite of newsletters: https://substack.com/@calacanis * Follow TWiST: Twitter: https://twitter.com/TWiStartups YouTube: https://www.youtube.com/thisweekin Instagram: https://www.instagram.com/thisweekinstartups TikTok: https://www.tiktok.com/@thisweekinstartups Substack: https://twistartups.substack.com * Subscribe to the Founder University Podcast: https://www.youtube.com/@founderuniversity1916

All right, everybody, welcome back to the program. You know, we like to do these startup basic segments. Why do we like to do them? Well, we get asked a lot of questions over and over and over again. We have 20,000 startups apply for funding from our firm every year. It's called Launch. You can go to Launch.com slash apply to apply for funding from our little seed fund. And then we'll meet with probably 5,000 of those founders, just in a brief Zoom call.

and we wind up investing in a hundred of them a year. What that means is we get the same questions about tap tables, about accounting, about human resources all the time. So if you go to this week and startups.com slash basics, you'll see all the basic videos we have. This is the blocking and tackling. And the partner I use for our finances is Cruz, K-R-U-Z-E. and Scott from Cruz comes on the program, and he's a CFA. He's a CLO over a Cruz,

and we talk about the basics. And things have been booming in Silicon Valley in the tech industry recently because of AI, huh, Scott? Absolutely. I actually saw some of the stats today, and it's just, it's mind-blowing what's happening. I thought venture capital investing would be down,

but it's actually up for the year. Can you believe that? Well, you know, we do have to parse out some of the mega rounds, right, because AI, and we're going to get into it today, we're going to talk about, you know, sort of financing, revenue accounting, everything for AI specifically. It's really a lot of work to put up, you know, clusters of hardware. Takes a lot of highly paid sought after developers to build anything good, which means a lot of capital is. being spent and invested. So I bet you there's a fat part of that long tail there where, you know, Elon orders 100,000 H-1s and him, and I heard the story of him and Larry, or Larry Ellison's a story of him with Jensen begging for H-100s, begging to throw piles of cashes into cash over the moat into the invidia castle, as it were. But let's get into it. You know, you've got 175, of your clients that you do, the books for, are AI.

So it's like one in four, basically, for you. What's unique about this boom in your mind? Yeah, well, the boom is very real. We have a really great stat where our clients raised $2.5 billion this year, and one and a half billion of that is the AI companies. So even though they account for, what, 20, 25% of the clientele, they're raising more than half the money, which is pretty mind-blowing. but I think to validate your point,

they just have a lot of infrastructure costs. It's kind of like the rest of the business model is kind of like SaaS. Like, you know, you spend money on marketing, you spend money on headcount, things like that. But what you've found is their infrastructure costs are about 2x higher than a typical SaaS company. And that really adds up across all these companies. That's a really interesting point.

And it's because this new hardware layer is very expensive in order to do this. And we've been living in a compute light ecosystem for a long time. Compute got way ahead of what was necessary, storage, bandwidth. I mean, it's not like a SaaS application if you're using Slack or Zoom requires some ton of CPU. It doesn't require a ton of storage. It doesn't require a ton of bandwidth. You know, the bandwidth of a cable modem at home more than enough to do a nice crystal clear Zoom call, your laptop or your G drive, your Dropbox, plenty of room box to store your files. So we've been kind of through almost two decades of not worrying about hardware. And here we are again, racking and stacking and trying to make this data move

quickly. But let's talk about the revenue side, because there is this term air gap, basically all that spend, and then where's the revenue coming from? Turns out people will pay for AI, but it has a big cost. And so people are seeing value here. It's kind of the antithesis of what you and I saw, Scott, with crypto. We were talking about crypto and startup basics and how lucky that was because you had all this vision, going to change the world, but not a lot of application, right? And then we start talking about AI.

I know you've got tax AI company. companies coming at you who want to do co-pilots. We have one tax VPT. I know for production, we have a company podcast AI that's doing the transcripts of this very podcast and hosting our website and helping us make clips. I think they're a cruise client. I think podcast AI is a cruise client, actually. I hope so because that revenue is going up into the right.

I think they're part of your group. So, you know, people are willing to pay for this stuff and they're getting valued from it. But tell me what you're seeing because you actually can take the data across these 800 companies of which, you know, one in four or so are AIA-based, and you can give us an approximation of what's going on here. Yeah, exactly. So we're this really great index for Silicon Valley. And what we're seeing is AI companies are growing 60% faster than a traditional SaaS company.

And so, and we could talk about the expense side of the equation in a little bit. But that's, that's pretty impressive. I think, you know, you talked about why, you know, all the capital costs going into AI companies, and that's very real. But I think the reason why investors are. are happy to write those checks is they're seeing, they're actually seeing growth. Like, this is really happening. And those are on companies that are over a million dollars in ARR. So, you know, this isn't like the little dinkers who are growing 60% a year because they had a

really good three months. These are companies that are substantial. And so that's a real stat. And so to me, I still think we're really early in this whole trend. And you, I mean, you have all these companies that you've invested in. You're probably seeing the same thing. but the winners are starting to kind of emerge. And in the venture capital ecosystem, it's the winners are usually the ones that are growing the fastest, those are the ones that people get so excited about.

So we're seeing it. That stat is pretty mind-blowing to me. Yeah, we're seeing it in our portfolio as well. I mentioned podcast AI. And, you know, when they pitch a podcaster on, hey, we'll make your website for you for 500 bucks a month, transcribe every episode, help you make clips, you start putting that against producer time.

And then if a producer in a podcast, you know, let's see, 2,000 hours, I always do the back of the envelope math. It's when you and I, you do your math perfectly. I do back of the envelope as an early stage investor. 2,000 hours times 30, 40 bucks an hour, 60 to 80,000 dollars a year is what, you know, editors, if you want to do offshore editors would be a fraction of that, half of that maybe. but, you know, U.S. based editors, producers, maybe 60, 70, 80K. So that's 30 bucks an hour. If you can save, you know, I don't know, 10, 20 hours a month of time, which is what

podcast AI saves us per week, you know, then the arbitrage is between those two numbers, isn't it? You know, they're providing us with 100 hours saved, 80 hours saved a month, and they're charging us for 10 hours. That's a really good deal. that's called efficiency, and that is what drives the U.S. economy. Totally. And I remember actually when Sean and Ed were, I think someone of your team sent him over,

and they gave me the similar pitch. It's not just about the time save, but it's about kind of that cognitive relief. I mean, like, for example, you, you're a super busy guy, knowing that that's just kind of taken care of has this other benefit that's even beyond dollars and cents, which I really like. And we're kind of seeing that, you know, we're lucky that we have a bunch of of these AI clients. And so a lot of times we'll like dog food their stuff, you know?

And it's really fun. It's also just like, I think what it's also doing, I still think we're very early, like first or second ending of this stuff. It's just kind of sparking new ideas or sparking new applications that will continue to see this over the next two or three years. I remember when SaaS first came. I'm old enough to remember like in 1999 in 2000 when the baby SaaS companies were coming. And it was like, you can put the stuff in the cloud.

That's amazing. You don't have to rack and stack servers in your office. I mean, people don't remember this. But, you know, if you had a company of over 100 or 200 people, you were on a couple of floors of the buildings, this is when people went to things called offices. And they worked together on projects called companies. But at that time, you would have a closet and it would be your server closet.

So you found a closet or an old conference room. You converted it. You would have air condition brought into that floor for $100,000. You'd spend a half million dollars on servers. then it moved to the cloud. Similar thing happening here. I like your analogy, though, of, we're not analogy.

I like your insight. It's the first time I've heard it, actually, that lowering stress by just having something automated is virtuous. And that is exactly it. I would have to be on top of a producer, to be on top of updating the website. I got to check it. You know, humans make mistakes.

And now, hey, the website's done. Humans don't make mistakes. So I feel like a third. of the grunt work that we face this year, you know, or of the last two years is being automated. I think it's like somewhere in that number. And proofreading is another example of it. You know, we use Grammally, which is AI-based.

Chat, GPT, Claude also do a great job of this. You dump any text in there and say, check this. Or superhuman, one of our customers. And I think you did their accounting for a bit. Maybe still do. You know, they have AI built into it. So composing an email and not

having somebody have to check, oh, before I send this to a client, can you proofread it? You know, or, you know, a deal memo having to be proofread. It's just kind of already proofread. So I used to spend so much time, but two years ago reading deal memos and making sure commas were in the right place because I'm the best writer on the team. We're top two out of 20. You know, we have two good writers, three really good writers. We have three great writers in the company. now, and we had 17 who were okay or bad or good, now everybody's good. And the three people are awesome because you don't have to worry about where the common goes. Yeah, there's also a nice time savings thing that's happening. I mean, for us, our VP marketing, Healy Jones hooked up

chatGBT through our Slack channel to index all of the content that Vanessa and I have written or all my videos, which is really kind of like the intellectual property of Cruz, right? Like all, just like all this knowledge we have in our brains. And so, Now our accountants can actually go to the questions channel, type in a question, and they get the answer as a Vanessa or I would be answering it. And it's really, really powerful. Now, we always prove free to it. We're always really careful because we're regulated.

We don't want to send out bad information. But just if you're an accountant and you're trying to get back to your client as fast as possible, getting that really, that answer instantly and insanity checking out, of course, it just relieves the stress and it improves our ability to get back to them. And then you multiply that by 800 clients and all the questions we're getting, every single day. It's really powerful. That's not even something we charge for. That's not even something that's revenue generating. It just made us a better accounting firm. I feel like the entire labor force in the United States or the ones globally that are embracing

these tools are leveling up. And the distance between people using these tools and not using them, the gap is really wide. I insisted everybody in our company when you open a new tab in your browser have it and you can get these browser extensions, you know, in the Chrome store or Brave store or Firefox add-ons, whatever it is. When you open a new window, I open by default CHAPT4-O. And then sometimes I'll switch it to clot. Why? Because I want it to be what I open up to and I start conversing. I'll give you another podcasting experience or an experience I had today on a call with a serial entrepreneur who's coming back to us

with a second company and we're investing and having them, you know, do the 125K standard accelerator note. And we were talking about this field and I wanted to know what the terms of service

of a company said about this idea. Boom, I type it in. Then I wanted to know what other companies had tried this. And I said, was there an Uber for blank?

I won't put in what blank is. And I said, boom. I said, okay, put it in a table. Tell me how much money was raised, give me citations, and this new reasoning engine that 40 has really started to give me results that would have been the same as an associate. A venture associate would be on that call, and after the call, they might spend the afternoon doing that work. Now I'm getting

it in real time on the call. For the first time, Scott, I took my Q&A from the Zoom call, and I gave it to the founder and shared it with them. And I said, here's the research. I just did. Now, normally, an associate would give me that. They'd share it with me. I would share it back with the founder. It would take four or five hours. Now, you start talking about associate time. Now, you're looking at 40, 50 bucks an hour.

All of this is abstracted away. And so if you're not using these tools, you're falling behind. Yep. And you talked about something that was really interesting, where it's now your default opener on your web browser, which is, that's kind of the next stage, I think. I mean, we're very fortunate. We work with a company called Perplexity.

And I know a bunch of people, I mean, it's a classic cruise. It's like when Superhuman, you know, when Ruel came over, Vivek, they come, the two people and an idea, you know, two or three, it's probably been two or three years now. And all of a sudden, now they're exploding. And I know in our company, a lot of people just use perplexity instead of Google now, you know, and that would have been mind-blowing to think about two years ago or three years ago. But these defaults are starting to change.

And it just makes so much more, there's so much more potential. It's really exciting. So, yeah, these companies, it's for real. I know. Tell me about the accounting. I mean, that's what, let's get down to breast tax here. They're going to be charging consumers and enterprises for this product. And there's a lot of expense.

So how do you, how does anything change or is it just standard SaaS accounting? And then how do people charge for this? Because I think sometimes people don't take into account what these, you know, chat GPT bills could be, you know, using some of these models. And so, man, you could get. a bill as a CEO and all of a sudden see $7,000 $8,000 from Cloud or chat GPT or whoever and be like, whoa, that's a little expensive.

I mean, try million dollar invoices from the cloud. I mean, we have a lot of companies, not, I'd say probably five to 10 that are spending something like 500K to a million dollars a month just on their cloud infrastructure. Yeah. And so there's kind of the revenue accounting, which we'll talk about in a second, but the expense accounting is really, really important,

especially because, I mean, a lot of these, you know, this is Amazon, Microsoft, Google Cloud, they're fantastic companies, but these groups are growing so fast. They, their invoicing is not really how you want it to be, let's put it nicely. And so we see companies getting like double invoiced on accident or they forget to bill the company for June and all of a sudden the next month they get a $2 million bill instead of a $1 million dollar bill. So this is like, this materially affects your burn. So we're, we're actually doing a ton of work right now just doing these like spend analysis for AI companies because you miss a million dollars here, a million dollar there. That could actually dramatically impact your cash runway.

The companies are spending a ton of money with this. And what's also interesting, Jason, which is, it shows you how a nascent thing is, this is. A lot of our big AI companies are spending a lot of money with like Anthropic or Open AI. They actually have like executive to executive. It's not like the sales guys talking to this, the CTO talking. It's like the CEO to CEO kind of conversations because these are really big contracts. They're spending a lot of money with each other because sometimes they're aggregating each other. So it's, this is like real, real dollars. And look at this. I mean, the gap here is unbelievable. AI, we're looking at spending on hosting computer and software as a percentage of revenue. These AI startups are spending over 50%

and it's been growing since 23, and then you look at standard SaaS, it's fixed 19%, 18%, not even above 20. So this proves what we've been talking about. It's supposed to be getting cheaper. They're supposed to be lowering the cost

by 90% a year, but I think people are finding new things to do. Every time you solve something, we just talked about the podcasting applications, you then find another reason to use AI. And that's why this is such a profound shift in technology.

Once you start using it, you cannot stop. It becomes the paradigm shift for you as a executive, as a founder. You're just going to look at every problem and say, how does AI solve this? And then that comes with an expense.

Totally. And I mean, Corweave was one of our clients for two or three years. And that's one of the major kind of compute power companies out there. And like they, all these companies are just constrained by their capacity. They can't add enough capacity. So they have pricing power. Like, you told that story about Elon asking Nvidia for the, for the processors, right? Like, if that's the case, everyone's got pricing power. So what's kind of interesting to put

this in like a venture landscape is we're seeing, I think we're seeing a deeper J curve, meaning these companies have to invest a lot more just to get going. And it's, I don't, we, we know they're growing revenue faster at 60% faster, but is that going to be enough? Do they need to be growing 200% faster than a normal SaaS company. That's what's going to play out here. But yeah, they're, they have dis-economies of scale, which is, which is a little scary as an investor or even someone working in the industry. But I think, I'm kind of like, I think this is what you were saying. I think entrepreneurialism, capitalism is going to come by and start working those costs down Moore's Law, all these things that we've seen historically in technology starts driving

economy to scale. It's just going to take a little bit longer. It's one of these things where you're now using in so many different places, you're going to want to do line items. If you're doing some, like in my example in our firm, and we're just using off-the-shelf stuff, but if we were going deeper, you might be using some part for your podcasting media business. You might be using some for your legal accounting.

You might be using some where your developers are using a co-pilot. Those are different cost items and cost centers. And so getting one, giant bill from your web provider, your web services provider, you need to break those up. I don't suppose they do that very elegantly in the bill. No. A lot of times we're going, but you're making an excellent point, especially research and

development and cost of goods sold. So cost of goods sold is the cost to deliver the service, just kind of in a normal way to your customers. Research development is what, you know, your advanced team is working on building new capabilities. And what's really important for startups to break that out is, first of all, venture capitalists really care about your margin. And what we've been kind of talking around is things like gross margin, right? If you have a really terrible gross margin, no matter how exciting the business is,

VCs are going to be hesitant to invest. And then the second part of that is for your R&D tax credits, money you actually get back from the government as a tech startup. You only are able to claim your research and development costs. And so breaking that out allows you to claim the R&D compute expense on that credit. And so these companies are getting $250 to $500,000 back as a payroll tax offset, meaning they don't have to pay payroll taxes. So just that simple act of breaking that out, the accounting pays for itself by 10x, right?

You're just getting a huge benefit there. You want to do that. And here, as we're talking, for those of you who don't know the J-C curve, I just said, explain and show the J-C curve to chat GPT-4. Oh, I'm not even using the new preview of 01. But, you know, I didn't realize there was a political j-curve. I didn't either. Yeah. So in political systems, reforms or shift might initially lead to instability or dissatisfaction. However, once the

changes are institutionalized, the benefits or realize stability and satisfaction prove again, showing a j-curve. So it might be like, oh, my God, we give women the right to vote and everybody gets upset, but then it comes back and everybody's happy, whatever it is. I mean, I'm picking a very dramatic example there, but here's, you know, more representations. And I said, just explain the J-curve, which basically means you'll lose money, you go into a deficit, and then you recover, and here's Tesla's, and here's Uber's. Those are two famous businesses, and it says here, Tesla faced significant financial losses early years as then invested in R&D production facilities and building a brand, the cost of

electricials society, and same thing for the J-curve for Uber, subsidizing rides, expanding globally and legal and regulatory challenges cost a lot of money. So you're going to go into a deficit and then hopefully it returns. If you're going to do this, you need to have a partner. Cruz is a great partner. They really, really help a lot for startups because one of the things that drives me personally crazy on my partner, Mike Savino, on the business, who's a bit of a bean counter himself, I mean, that in the most positive way possible, because we don't want to waste beans.

He asks a lot of tough questions. Mike sends us emails asking tough questions about accounting and finance. Well, because you know what, he's so good at accounting and finance that we'll have these discussions and you only see the ones he can't answer. So if it gets to your desk, it's because like he understands most of this stuff, but something new comes up. You need to have a great partner. Scott's that partner at Cruise. I just love having him here to help us with this. AI is going to be changing everything.

And really, you know, what I'll tell you is I've seen as many startups fell because of legal. accounting and HR issues, which I'll put into operations, as fail because of not getting to product market fit. It's an equal amount. So, you know, eight out of ten startups fail hard, return nothing to investors, you know, and it's just an experiment that fails. That means four of the 10, in my estimation, they'll because of operational issues, which in a way are self-inflicted and you can remove them by having a great partner. Cruise is our partner for accounting and all this finance stuff. And then, you know, Becky DeGraw over at Wilson, Cincinnati helps us with the legal stuff.

Here's your call to action. Go to cruise consulting.com slash twist. Talk to Scott. He's my guy. And if you want to learn a ton this week in startups.com slash basics. And he's Scott at cruis consulting.com. K-R-U-Z-E. Thank you, Scott.