The Twenty Minute VCClickHouse CEO: AI Margins Need to Improve | Revenue Concentration Should be a Concern
EVERY SPOKEN WORD
65 min read · 12,601 words- 0:00 – 1:52
Intro
- AKAaron Katz
We're just getting started. This seems to be accelerating at a unprecedented pace. I mean, we haven't seen revenue growth like this in our lifetime. We went zero, 12, 50, 200, and we'll finish this year north of 500.
- HSHarry Stebbings
Our guest today just got front-of-shirt sponsorship for Craven Cottage and for Fulham, most importantly. Welcome Aaron Katz, Founder and CEO of ClickHouse, industry-leading online analytical processing database management system. Try saying that after a couple of tequilas. But they've just crossed 350 million in ARR, and Aaron is incredible. He was one of the leading execs at Elastic for years. Before that, he spent 12 years working with the one and only Benioff at Salesforce. Now he's obviously the Co-Founder of ClickHouse, which is worth over $15 billion.
- AKAaron Katz
We need to get to a billion dollars of ARR as quickly as possible. I'd put the over-under at December 2027, and I would take the under. We could take the company public next year if we wanted to.
- HSHarry Stebbings
Ready to go? [upbeat music] Aaron, dude, I've done over 1,000 shows. I've never had a setting quite like this for a show, so thank you so much for, for hosting at Fulham Football Club.
- AKAaron Katz
Yeah, total joy.
- HSHarry Stebbings
Now, [chuckles] I wanna start with just a little explainer on what is ClickHouse and what does ClickHouse become in a five-year period? Just to set the scene there.
- AKAaron Katz
Yeah. Well, as one of our more recent investors, uh, I think you understand the thesis behind it, which is it's the world's most popular open source database, and it satisfies a very broad array of use cases, and it's used by nearly every AI native company, An- Anthropic, OpenAI, Weights & Biases. Um, it's known for its lightning-fast query execution and, uh, extreme resource efficiency in terms of storing
- 1:52 – 3:49
Are We Still Early in the AI Cycle?
- AKAaron Katz
vast volumes of data.
- HSHarry Stebbings
When we look at, bluntly, the fastest-growing AI companies, I think the single biggest question that I have right now is, where are we in the cycle? You know, I'm really good friends with very smart people, and they go, "Harry, I've seen this before. The levels of debt that we're seeing is insane. The, the prices and the valuations are so exuberant." And then I also look at adoption and revenue scaling, and I have these two paradoxical data points.
- AKAaron Katz
We're just getting started. You know, one of the few benefits of age is experience.
- HSHarry Stebbings
[laughs]
- AKAaron Katz
And so I've been through a few of these before in terms of the internet and mobile and social.
- HSHarry Stebbings
How does it actually compare as a builder? 'Cause we hear a lot of pontifications from venture investors. How does it compare for you?
- AKAaron Katz
And those cycles, in my experience, were much more gradual. This seems to be accelerating at a unprecedented pace in terms of how quickly these agentic experiences are maturing and how quickly these companies are growing. I mean, we haven't seen revenue growth like this in our lifetime. And so... And the demands for the, on the systems of these agentic applications is un- unlike anything we've ever seen.
- HSHarry Stebbings
Totally agree with you there in terms of the revenue scaling. A lot of people always look to pick holes. One of the holes they often pick in the revenue scaling is the gross margin profile. Do we just see a new world of lower gross margins? When we look at, you know, companies like FireEye, Slim was on the show, and she was like, "Yeah, we're at 30, 35%. We hope to be more over time." Do we just have a lower gross margin world, or do we actually scale into traditional SaaS margins over time?
- AKAaron Katz
You know, a lot of these companies just seem to have unlimited access to capital right now. So they... And they're growing so quickly that I think they can operate with gross margins that most public company investors would not be satisfied with. I think as, as long as they can demonstrate a path to margin expansion over the course of the next few years, while still growing at these unprecedented levels with very healthy balance sheets, I worry less about gross margins like we did five years ago in traditional
- 3:49 – 4:46
The Biggest Risk in AI Investing Today
- AKAaron Katz
enterprise software.
- HSHarry Stebbings
What should I worry about then as an investor? As I think about navigating this new world, again, you have the best customer base that you could almost ask for. Now, I was talking about, you know, some of the fast-growing companies on our wall the other day, and you're like, "Well, they're both customers." And it was just universal, the best companies were using ClickHouse. What should I be concerned about, worry about when I'm investing, say, looking at these names?
- AKAaron Katz
If you were to come down to one, if you were to say the s- what's the single biggest risk?
- HSHarry Stebbings
Yeah.
- AKAaron Katz
Would be durability of revenue, because the switching costs that you and I talked about are very high for infrastructure software. The switching costs can be very low for agentic applications. Um, and we're seeing that with these model providers that are leapfrogging one another, what seems like every other week. And I would call into question the durability of some of the revenue for some of these AI applications.
- 4:46 – 6:16
Is Anthropic’s Revenue Actually Durable?
- HSHarry Stebbings
When I look at, like an Anthropic IPO at 2 trillion, and you look at Claw Code being the kind of dominant Trojan horse behind that, would that not fall into the low switching cost?
- AKAaron Katz
I would put it in that category. Now, it's, that's not showing up in our use of Claw Code-
- HSHarry Stebbings
[laughs]
- AKAaron Katz
... because our Anthropic spend is up 100 times from what it was at the beginning of the year.
- HSHarry Stebbings
Can I ask, how much do you spend on Anthropic?
- AKAaron Katz
A significant amount. You know, we, I, we weren't doing enough at the turn of the calendar year, so I sent an email to the company, and the subject was the AI Awakening, and I said, "I think we're moving too slowly, and I'm not seeing the adoption of these, uh, coding applications, for example, that I would expect to see for a leading database provider like ClickHouse."
- HSHarry Stebbings
Mm.
- AKAaron Katz
And the company rallied, uh, to the call, and we've seen just this explosive growth in terms of our use of these, of these applications. Um, and we're now looking at adopting more open weights models whereas just the traditional Frontier Labs. But I think Anthropic specifically and OpenAI have the benefit of being both a model provider and the harness provider that the open weights models right now are behind in.
- HSHarry Stebbings
Do you think that is the right approach though? Do you not think it's better to be independent because then you can actually do optimal model routing for different tasks versus being tied into one model provider because they are your harness also?
- AKAaron Katz
Yeah, I mean, for example, some of these open weights models we'll use for code review-
- HSHarry Stebbings
Mm
- AKAaron Katz
... but we won't necessarily use to ship production code-
- HSHarry Stebbings
Mm
- AKAaron Katz
... because we have concerns about the output inference.
- 6:16 – 10:04
How Should Companies Think About AI Token Spend?
- HSHarry Stebbings
Totally, yeah. You said there about, "I love this, the AI awakening." The challenge then becomes, and we had the president of Uber on the show who was much more skeptical of the ROI that it generated internally, um, how do, do you think about token budgeting cost when suddenly you're incentivizing this, "Hey, run free," and then the bill might come at the end of the quarter?
- AKAaron Katz
Well, the primary measure that we care about, as you know, is revenue growth. And if we see the sustained revenue growth that we've experienced over the last three years, and we're a very efficient company, as you know, um, some would argue we're too efficient, then I worry less about the expense that we're incurring on coding agents, for example, because we're covering a very broad surface area, and our roadmap is accelerating at a pace that we've never seen before. And as long as we can continue to ship features that satisfy this very broad set of use cases, then I can always rein in token consumption. And the cost of tokens, as we know, is going down, not up, and so they're becoming more efficient, not less efficient, but our revenue's growing faster than it ever has. So I'll take that trade any day of the week.
- HSHarry Stebbings
You said that you're too efficient in some people's eyes. Where should you have spent where you didn't spend, and how do you reflect on that?
- AKAaron Katz
So we've got about 100 quota-carrying salespeople, for example, and I think you know our revenue scale is significantly more than 100 million.
- HSHarry Stebbings
Yeah.
- AKAaron Katz
And so our average, uh, rep productivity is quite high relative to the industry average. Our competitors that we're going up against, some of these very large data warehousing companies, some of these very large observability companies, have thousands of salespeople. They wake up every morning thinking about one specific use case. Our salespeople wake up every morning thinking about 10 different use cases, and there's only 100 of them going up against an army of 2,000, 3,000 sellers for some of these very large data warehousing companies. So I think the one thing, and if I look back over the last two years, that I f- wish I had done differently was increasing sales capacity.
- HSHarry Stebbings
That's so interesting. So you should've invested more in sales team, sales leaders earlier?
- AKAaron Katz
Well, I really wanted the pressure for the first few years to be on product and engineering.
- HSHarry Stebbings
Mm.
- AKAaron Katz
Because I looked at the two most popular infrastructure software companies over the last 10 years, and I distilled that down to Datadog and Snowflake. They were both successful, but through very different avenues. Datadog had this PLG self-service, developer-led motion, so you could get started, deploy an agent, instrument your application, never talk to anybody in sales. Snowflake went heavy after the enterprise through, you know, very expensive sales and marketing. I just thought it was gonna be a lot easier to follow the Datadog playbook than the Snowflake playbook, and it pr- it proved to be the case, but at some point you need to layer in an enterprise sales motion on top of some sort of PLG distribution.
- HSHarry Stebbings
What do you know now about layering on that enterprise sales motion that you wish you'd known before you started?
- AKAaron Katz
You know, I spent 12 years at Salesforce, and so I was kind of a, a student of, of Marc Benioff's playbook, and so many lessons learned through that experience, and this was a long time ago. I joined that company 24 years ago.
- HSHarry Stebbings
What was your biggest lesson from working with him, if there was one takeaway?
- AKAaron Katz
That you can overestimate what you can achieve in one year and underestimate what you can achieve in five. When I started, it was a three-year-old startup, and it was basically a glorified contact manager, you know, Salesforce automation. Um, we essentially said, "What you're traditionally using ACT, or GoldMine, or using a spreadsheet, you can use Salesforce for." But Marc had this bigger vision, and he said, "We're going after Siebel, SAP, Oracle, Microsoft." We didn't have the product set to go after those competitors, but he was such an incredible marketer that he created this perception in the industry that some of the largest companies in the world could adopt this technology and that we would deliver on a roadmap that would satisfy the requirements over time, and
- 10:04 – 12:01
AI Is Pulling Product Roadmaps Forward by Years
- AKAaron Katz
he did.
- HSHarry Stebbings
You said the word roadmap multiple times in different contexts there, but, um, I do a show every week with Jason Lemkin and Mario Driscoll. It's very successful and popular, which is fun, but Jason said last week that if you are not well into your 2027 roadmap already, you are behind. Are you seeing dev acceleration because of AI tooling, and how do you measure actual ROI internally when attribution is really difficult?
- AKAaron Katz
Yeah, we're shipping products faster than we ever have. We're entering new product categories, um, two years ahead of where we thought we would.
- HSHarry Stebbings
Really?
- AKAaron Katz
Yeah, both organically and inorganically. Um, we've made six acquisitions over the last four years that have propelled us into new use case areas, and our organic roadmap, shipping features like stateless workers, which is essentially infinite compute, we've shipped those faster than we ever thought possible.
- HSHarry Stebbings
How do you think... Nikesh Arora is a very good friend of mine. He's a very brilliant M&A machine. How do you think about the buy versus build versus distraction? What's that internal decision-maker for you when you think about those six acquisitions?
- AKAaron Katz
If I think that our product and engineering teams can, uh, innovate in a specific area that we're not in today, then I'll let that play out organically. If I see, uh, a founder or a group of founders that are building on top of ClickHouse, that are getting into a category that I think is gonna be, you know, a, a future component of what we build as an ultimate data platform, then I think about doing something inorganically. And we partnered again with six different founders, uh, actually more than that. Some of these companies have multiple founders. Our most recent one earlier this year was Langfuse-
- HSHarry Stebbings
Yeah
- AKAaron Katz
... out of Berlin.
- HSHarry Stebbings
Yeah.
- AKAaron Katz
Um, which is three incredible founders entering agent observability. Every enterprise in the world is gonna need this technology.
- HSHarry Stebbings
In terms of, like, the agentic future that we face, how does software and product decisions change when you lo- no longer cater to humans but you cater
- 12:01 – 16:12
What Happens When AI Agents Become the Customer?
- HSHarry Stebbings
to agents?
- AKAaron Katz
Well, even four or five years ago, software applications were designed for a specific persona that had a role within an organization, and their query patterns were very predictable in whether or not you ran a report or you looked at a dashboard. Agents don't have personas And so while you would traditionally use a specific application for observability or data warehousing or CRM, agents expect that they're gonna traverse across all these applications, and they're not gonna be constrained by access, they're not gonna be constrained by latency, and the experience is gonna be defined by the slowest point in that chain. And so what's the number one requirement for agentic query patterns? Low latency, because they're executing dozens of SQL queries simultaneously across all these different systems. And the most important requirement is the unpredictability of those query patterns, the responsiveness, and the fact that they're much more exploratory than a traditional human, you know, report or query.
- HSHarry Stebbings
Sure. Can I ask you, when you see the explosion of agent queries in this way, you'll also see the increasing awareness from agents to be more cost efficient. To what extent do you think you see a race to the bottom on pricing with the awareness from agents that they can't have an explosion of queries and cost being the same?
- AKAaron Katz
Well, I don't see agents necessarily being cost efficient.
- HSHarry Stebbings
[laughs]
- AKAaron Katz
Uh, I don't see them thinking about consumption and budget like humans do. Um, but-
- HSHarry Stebbings
Me and my mother have that in common. [laughs]
- AKAaron Katz
But the, but the volume of queries is exploding at a rate that we've never seen, and so it's requiring these systems to completely rethink, you know, their pricing models, their consumption patterns, access patterns. And not only are agents hammering these services in an unprecedented way, but they're actually now selecting the underlying infrastructure. And so, you know, while you could go to Claude or you could go to ChatGPT and say, "You know what technology should I use for this specific use case?" I'm thinking about a future where the agents actually select the infrastructure stack behind the application, and positioning ClickHouse to be the default database for the next generation of applications that agents are building, not humans.
- HSHarry Stebbings
And just so I understand, when we fast-forward to that three-year preference stack for agents, it's number one latency, it's number two...
- AKAaron Katz
Efficiency
- HSHarry Stebbings
... efficiency.
- AKAaron Katz
Yeah.
- HSHarry Stebbings
And-
- AKAaron Katz
I mean, Tesla, for example, is ingesting a billion events per second into ClickHouse. Like, that throughput is unprecedented, and so you need the ability to both ingest that efficiently, store it efficiently, and then be able to query that efficiently at a, a fraction of the cost of traditional database technologies. There just isn't another technology in the world other than ClickHouse that can satisfy those requirements.
- HSHarry Stebbings
When you think about that agent buying process, trust and security are so important. Everyone is saying that we're at this golden age in terms of cybersecurity, and we see more and more security hacks. How do you feel about the security vulnerabilities that come with the agentic future that you are planning for in three years out? More on the enterprise side. We obviously see it on the personal side, but more on the enterprise side.
- AKAaron Katz
Well, I think it's gonna require multiple deployment models. So you need to be able to consume services via a, a, a cloud offering through any one of the three major hyperscalers. You're gonna need the ability to manage that data on-prem behind your VPC in your firewall. So you're gonna need to have the flexibility to deploy these applications as you best see fit, especially in the indus- in the enterprise where you've got highly regulated industries, a lot of da- data privacy concerns, a lot of compliance requirements. And so as a supplier to these customers, we think about how do we support them dependent on their deployment preference. And, you know, that's a tricky roadmap to maintain, because most companies pick one of those avenues. You know, I mentioned Snowflake and Datadog. They're primarily cloud services. You know, you look at more traditional technologies that run on-prem, and so if you force your customer into a specific lane, you know, you're limiting the addressable market that you can go after.
- 16:12 – 17:59
Every AI Agent Will Need an Identity and a Budget
- HSHarry Stebbings
When you look at the agentic future of the three to five years that you're planning, what seems insane today that you think will be quite commonplace in three to five years?
- AKAaron Katz
We can talk about the fact that agents will need to have an identity-
- HSHarry Stebbings
Love that
- AKAaron Katz
... that they don't have today. They'll need to have a budget, and how do you authorize an agent to consume a service?
- HSHarry Stebbings
What do you mean agents need to have an identity and a budget?
- AKAaron Katz
You know, if you ask Anthropic how they chose to use ClickHouse, they'll tell you they asked Claude, "What technology should we use for this specific observability use case?" And Claude suggested ClickHouse. I'm thinking about a future where they say, "Hey, we need to build an application. Provision the underlying stack." So you've got a database, you've got networking, you've got compute, you've got storage, and the agent's actually making that selection process. But they need to have authorization, they need an identity, they need to have a budget, right, to be able to consume those services. We're not there yet today. So if I were in your shoes, I'd be thinking about what companies are best positioned to give that agent everything they need to build a software application.
- HSHarry Stebbings
I'm sorry. Help me understand. Where should I be looking then, and why is that not included in the harness? Like, why is that not in the settings and preferences of the harness and the model provider?
- AKAaron Katz
Because most companies aren't just gonna let their agents run wild and build whatever they want and consume as many resources as the agent deems fit. There's gonna need to be some sort of governance and oversight with that consumption, and we're not there yet today. There's some human that is observing that consumption. They're monitoring the agentic spend. They're putting controls in place, uh, to make sure that things don't get out of hand, that they don't access enterprise data that they shouldn't, that they don't spend a certain amount of money that they're not authorized to. If you look out three to five years, those agents are gonna be
- 17:59 – 21:20
Will Every Company Have Its Own AI Model?
- AKAaron Katz
fully autonomous.
- HSHarry Stebbings
When we think about... I, I, I said one that I think is very clear is also, like, specialized models. You know, we both know Leonard Firewas. I think every company will have their own model, you know, trained on their own data. They'll supplement their own data with, you know, additional data. But I very much see that being common. Do you see a world of millions of specialized models? Do you think you actually have your Anthropics and your OpenAIs take the large majority of enterprise, and only very specific cases have specialized models? How do you foresee that given the access point you have?
- AKAaron Katz
I think we're gonna have both. I think you're gonna have specialized models for a specific use case, like legal tech. If you look at Harvey, for example, and I know they're, I think, leading the category in terms of specialization, but I think the large frontier labs are still gonna be the dominant providers in the space.
- HSHarry Stebbings
Can you help me out here? I'm an investor in Lagora. Why is that a better approach than Lagora's, who obviously have not decided to dedicate their resources to building out specialized models?
- AKAaron Katz
They, they consume less ClickHouse is the short answer to the-
- HSHarry Stebbings
[laughs]
- AKAaron Katz
... to the question. We're an infrastructure provider, so like we're picks and shovels at ClickHouse. So we're not, like, picking winners in these categories. Lagora could be a bigger company than Harvey a year from now. I, I can't predict the outcome of these specialized providers. I don't know their revenue scale.
- HSHarry Stebbings
What's the average customer spend on ClickHouse?
- AKAaron Katz
Uh, it's a good question. So, you know, w- we define a customer once they hit a, a certain, uh, revenue scale.
- HSHarry Stebbings
Yeah.
- AKAaron Katz
We've got a long tail of customers that, uh, consume some of our services, but once they actually are in production at scale, we count them as a production customer. And, uh, there's a chart that you can see with our, our revenue growth and our, and our customer growth, and you can see they're growing at a, at a s- at a similar rate, but revenue's actually growing a little bit faster because our customers are growing faster than new customer acquisition, which is the most important function for a company of our size, is how many new customers can we onboard in any certain period, because we've seen the expansion characteristics of these services are unlike anything I've seen in my career. You know, we had over 130% net dollar retention, uh, at previous companies. We're over 200%. Because these use cases expand, you could be using us for data warehousing, then you use us for real-time analytics. Again, these historically were like siloed applications inside of an enterprise. People are now saying, "We want to put all of this in one data repository. We want to build applications against it. We want to expose it to our customers. We want to expose it to our partners." And so we're just seeing this rapid growth. Average customer spend, I mean, we've got customers that spend tens of millions of dollars with us every year. We've got customers that spend thousands of dollars with us every month, and everything in between. So averages can be a little bit misleading. If I were to look at the midpoint, probably around $100,000.
- HSHarry Stebbings
When they are making that buying decision, which competitor do you fear the most?
- AKAaron Katz
The one that isn't in the market yet.
- HSHarry Stebbings
Huh.
- AKAaron Katz
Like, our competitors are right in front of me. I can see them. I know their strengths. I know their weaknesses. And what I worry about is the technology coming from the rear view mirror. I worry about the next ClickHouse. Like, people really didn't see this technology coming. It was open source 10 years ago. That's when I first discovered it. It burst on the scene. Thousands of companies adopted it, but there was no company behind it, and so people dismissed it as just another popular open source database. There's been a lot before it. There'll be a lot more after it. So I worry about what's the company that's gonna disrupt us in the same way that we're disrupting the competitors
- 21:20 – 22:35
Open Models vs Frontier Models: Who Wins?
- AKAaron Katz
in front of us.
- HSHarry Stebbings
You mentioned, obviously, ClickHouse being an open project and, uh, you know, the amazing early traction that you had. When we look at the percent of tokens that are now going through open models, it is increasing exponentially almost it seems, and it is removing from frontier models. What percent of tokens will go through open in three years versus frontier?
- AKAaron Katz
Well, the easy answer is 50/50. In the same way that what percentage of enterprise software today is open source versus proprietary.
- HSHarry Stebbings
Mm.
- AKAaron Katz
I think it's a pretty even distribution. Look at the big two data warehousing-
- HSHarry Stebbings
That would be a controversial prediction, though.
- AKAaron Katz
No, no, no. Look at the big two data warehousing providers, Snowflake and Databricks.
- HSHarry Stebbings
Oh.
- AKAaron Katz
Right? Snowflake is primarily a closed ecosystem. Databricks is built around open source. Like, would you argue which one's gonna be bigger?
- HSHarry Stebbings
Databricks.
- AKAaron Katz
Perhaps, right? But then you add Datadog. Datadog's primarily proprietary, right? Other open source observability tools are open. Which one's gonna be bigger?
- HSHarry Stebbings
Databricks.
- AKAaron Katz
[laughs] In observability, I would argue Datadog over Databricks.
- HSHarry Stebbings
I just think out of that-
- AKAaron Katz
Data ware- Well, you never know. Three to five years is a long time.
- HSHarry Stebbings
It, it's a very long time.
- AKAaron Katz
I'm not gonna underestimate Databricks.
- HSHarry Stebbings
It's, it's-
- AKAaron Katz
It's a great company.
- HSHarry Stebbings
Do you know what's shocking when you say that? Is like five years ago, ChatGPT wasn't out.
- AKAaron Katz
Maybe.
- HSHarry Stebbings
That, that's the-
- 22:35 – 24:46
Could OpenAI and Anthropic Become Infrastructure Companies?
- AKAaron Katz
Yeah
- HSHarry Stebbings
... stark wow.
- AKAaron Katz
I mean, are they gonna get into infrastructure? Will they be offering databases as a service? I wouldn't dismiss Anthropic and OpenAI as core infrastructure providers.
- HSHarry Stebbings
Really?
- AKAaron Katz
Correct. Yeah, not at all. I don't see them as competition today, but you see how they're entering new categories so quickly. Just imagine the surface area they're gonna cover in three years.
- HSHarry Stebbings
They're building their own chips. This would not be-
- AKAaron Katz
Right
- HSHarry Stebbings
... extraneous.
- AKAaron Katz
Yeah.
- HSHarry Stebbings
Uh, totally get that. But the conventional wisdom would be that you see 90% through open, 10% through frontier. You know, i- in the conversations that I have today, everyone basically says frontier models will be used for cancer, climate change, extremely valuable but very few tasks, and then the rest of everything will go through open. You don't agree with that?
- AKAaron Katz
I, I don't because I think it, especially in the enterprise, they want provisions and protections that potentially open weight models, especially those that come out of China, cannot provide around indemnification, for example, and output inference, and it's gonna limit the use cases that those open weight models are adopted for. Now again, it's very important that we distinguish between open weight models and open source software. Okay? We're in the latter category.
- HSHarry Stebbings
Can we just-
- AKAaron Katz
My predictions on-
- HSHarry Stebbings
Can, can we just-
- AKAaron Katz
... enterprise adoption around open weights models is very different.
- HSHarry Stebbings
Totally get that. For those that do not know, and I like to be not just for kind of Silicon Valley inner engineers, open weight versus open source in a minute.
- AKAaron Katz
Let's focus on open source
- HSHarry Stebbings
Yeah
- AKAaron Katz
So open source is essentially where anybody can inspect the source code, anybody can modify it, depending on the license that it's governed by. You can deploy it with no attribution to the authors of the software. You can modify it. You can monetize it without any relationship of the people that are actually developing it. You can contribute to it. You can fork it. Now, open source licensing has evolved significantly over the last five years. That affects some of those
- 24:46 – 28:01
Are Enterprises More Scared of Frontier Labs Than Chinese Models?
- AKAaron Katz
implementations
- HSHarry Stebbings
I'm pleased that you said that, that you don't agree with that, and that actually there will be more concern around using open weight models and potentially Chinese models. Um, when you, when I speak to people on the show, they actually say that the biggest enterprises say they're more scared to work with frontier providers than they are open source Chinese models
- AKAaron Katz
Because people don't trust the statement zero data retention. It's basically you saying, "Just trust me." Like, "I got you." You know? Some people will take you at your word, many will not. And so I think a lot of companies worry about sending their source code, for example, to a frontier lab
- HSHarry Stebbings
Do you see that?
- AKAaron Katz
I do. I see it every day. I mean, we have that own concern internally, right? Because you worry about the output from that code generation. You worry about, you know, third-party indemnification if you were to consume code that's being derived from another repository
- HSHarry Stebbings
So what do you do in that case? You then go to an open model?
- AKAaron Katz
You limit the use cases. You know, you can use it for code review, for example, but maybe not to push production code into an environment that your customers are using
- HSHarry Stebbings
But then what do you trust? You trust an open Chinese model with the more sensitive data?
- AKAaron Katz
It depends on the use case
- HSHarry Stebbings
'Cause it, again, this is what someone said it on the show the other day and I was like, "No, you've got it wrong." Like, "You must have got that confused." 'Cause they said, "For all, like, kind of less sensitive things we use, you know, Anthropic, and then for everything more sensitive we use an open source Chinese model." And I was like, "You mean the other way around?" And they were like, "No, no, no. That's the right way around"
- AKAaron Katz
I think when you, y- y- when you need the, the legal protection that most enterprises do, you're gonna wanna work with one of the frontier lab providers. I think there's too much security concern around some of these open weight models
- HSHarry Stebbings
Is it justified?
- AKAaron Katz
Today I think it is. If I look out a year or two from now, I think a lot of these concerns will be addressed
- HSHarry Stebbings
You don't buy the backdoor to the CCP or con-
- AKAaron Katz
You know, I personally don't like to speculate that, you know, by using some of this software you're somehow gonna be engaging in some nefarious-
- HSHarry Stebbings
Chinese espionage
- AKAaron Katz
Yeah, exactly. I mean-
- HSHarry Stebbings
Yeah, yeah
- AKAaron Katz
... I, I, it's quite fantastic to go there
- HSHarry Stebbings
Listen, you're CEO of one of the kinda most prominent open companies in the world in terms of ClickHouse. When we look at the American open models, we significantly lag behind. If I were to say, "Aaron, I want you to spearhead open American models," what would you do to encourage, incentivize us to dramatically leapfrog China now in our open ecosystem?
- AKAaron Katz
Well, obviously I'm a huge fan of open source and open weights models, so I don't think that any sort of government intervention, uh, is, uh, is wise, um, in terms of limiting the adoption of these technologies, the distribution of these technologies, because I do think, you know, open source and open weights models are the future, and the future is defined, I don't know, three to five years. It's really hard to look out further than that. Um, I personally wouldn't take your capital and say, "I'm gonna deploy it to develop an open weight model"
- 28:01 – 36:25
Why Enterprises Are Moving Back On-Prem
- HSHarry Stebbings
Do you see more and more enterprises want to go back on-prem in this day and age?
- AKAaron Katz
We do.
- HSHarry Stebbings
Yeah.
- AKAaron Katz
Yeah. And e- even companies that I thought would never be going back on-prem are talking about going back on-prem. Like some of the most innovative digital native companies in Silicon Valley are now thinking about moving their stack from, you know, one of the hyperscalers to an on-prem environment
- HSHarry Stebbings
When we think about agentic workflows, a lot of what we've said, kind of agent identity, um, how sophisticated are traditional enterprises when you speak to the CEOs, when you sell to them, what you see, and what's the chasm between what you see and what they know?
- AKAaron Katz
Narrow
- HSHarry Stebbings
It is?
- AKAaron Katz
Yeah. I mean, I was at Canary Wharf yesterday meeting with some of the largest financial services companies in the world. They are leading the way in terms of their adopting new technologies in a way that I've never seen in the past. A, a lot of the times, historically, the sales cycles into these big firms would be measured in years-
- HSHarry Stebbings
Mm
- AKAaron Katz
... not quarters, right? They're now adopting technologies much faster than they ever have before
- HSHarry Stebbings
So you're seeing sales cycle compression in this day?
- AKAaron Katz
Yeah
- HSHarry Stebbings
Yeah. Gotcha
- AKAaron Katz
Um, I mean, open source aids in that, because you can get started without any sort of vendor relationship. Uh, PLG products like Datadog and ClickHouse aid in the fact that you can just spin up an environment without ever talking to anybody in the sales organization. You can have this frictionless experience to where you can evaluate, deploy, scale the product without a traditional enterprise sales process
- HSHarry Stebbings
I think people just fundamentally misunderstand the GTM and the business behind it. What do investors get most wrong when analyzing your business?
- AKAaron Katz
Investors often say like, "Where's, where's the moat?" Like, "How difficult would it be for me to simply redistribute ClickHouse? Uh, what's the risk about one of the hyperscalers offering ClickHouse as a managed service?" You know, it's been done before where AWS or Google or Microsoft takes your open source and they redistribute it as a managed service, so now you're almost, like, competing against your core database. Uh, that has been an issue in the past. I think if you can maintain a competitive advantage with your cloud offering or your proprietary features that are very, very difficult to replicate, then you can maintain that moat And I think very few open source companies get that right
- HSHarry Stebbings
I do want a weird transition back, but when we said about kind of agent preferences and agent, you know, Anthropic, using Anthropic to choose ClickHouse, does that mean, like, developer relations, developer community becomes less important? And how does the future of brand change when agents become decision-makers?
- AKAaron Katz
So I talked about you're building an application. Like, let's you're building a dating app, right? So you can build it over the weekend. You th- you can vibe code it, right? And in theory, that agent can select the stack, right? It can select a managed Postgres service 'cause you wanna support transactions. It can a managed ClickHouse service 'cause you've got analytics, uh, and everything in between. You still have an enterprise buyer. You still have a huge data warehousing project at a top bank or a telco that's gonna be driven by an engineer or a developer, and there's still gonna be a human that's making that architectural decision on are they gonna use ClickHouse? Are they gonna use Snowflake? Are they gonna use Databricks? You've got an observability workload. Are they gonna use something off the shelf like Splunk or Datadog, or are they gonna embrace open source and use something like ClickHouse? Those, for the next decade, there will still be a human involved in that decision loop.
- HSHarry Stebbings
Totally get that. So it doesn't change the investment and commitment that you have towards brand.
- AKAaron Katz
Not at all. Yeah. Because budgets are still controlled by people, right? And budgets correlate to technology decisions. And so a few years ago, I started doing things around awareness. Like, for example, for AWS re:Invent, Amazon's big cloud conference in Las Vegas, or Google Next, I partnered with The Chainsmokers and I had them perform 'cause I said, "There's not one party that everybody at re:Invent wants to go to. It's a bunch of, like, shitty restaurant buyouts and happy hours. I want one party that 60,000 software engineers are, you know, jumping over themselves to get access to." So I partnered with Alex and Drew, and we started performing these concerts, and they started performing these concerts in, in support of ClickHouse. Um, so then I thought about how can we even drive broader awareness? And so a sports sponsorship came to light. I'm not the first person to do this, as you know. You know a lot of other companies that are sponsoring Premier League teams.
- HSHarry Stebbings
Sure.
- AKAaron Katz
And so we, we did a financing earlier this year, and then we extended it and brought in some strategic investors like yourself and David Sacks and Michael Dell and JP Morgan. We didn't need the capital. We had a billion dollars on the balance sheet. Uh-
- HSHarry Stebbings
I, I quite like the affiliation to those names, so thank you very much for including me-
- AKAaron Katz
Yeah, yeah
- HSHarry Stebbings
... and David Sacks and, uh, JP Morgan. Very helpful.
- AKAaron Katz
Happy to.
- HSHarry Stebbings
[laughs]
- AKAaron Katz
So I thought what better way to spend this new investor money than to sponsor an English Premier football club in London.
- HSHarry Stebbings
Why football? Why English Premier League? You can sponsor F1. You can sponsor LaGore do golf as well. Why football?
- AKAaron Katz
Well, I love the sport, but let's put that to the side for a minute. I think the value of these sponsorships obviously come in two forms. The first is awareness, and as we saw last night against Chelsea, this is a game that's being televised globally. So you've got millions of viewers that are looking at your brand, and you're getting impressions obviously. The second, which is obviously easier to quantify, is hospitality, and we're sitting here at Craven Cottage along the Thames. I think this is arguably the best sports experience in the world, and I've been to many. We had 20 executives last night attend an intimate Michelin-grade dinner. We had, you know, the C-level executive come from Paris, one of the largest banks in Europe, just to experience that, and those types of relationships are extremely important, especially as we move upmarket.
- 36:25 – 38:36
What Founders Get Wrong About Fundraising
- AKAaron Katz
ways, if not both.
- HSHarry Stebbings
Dude, I was talking to your investors and, you know, many of them said every round I've wanted to invest more in, in Aaron and ClickHouse, and he always cuts me back What do you know now about fundraising that you wish you'd known when you started?
- AKAaron Katz
The credit really goes to Yuri and Alexei. Like, they're- my two co-founders are spectacular. The best engineers I've worked with in my career by a very wide margin. And so I'm happy to be the interface to the investor community and maintain these VC relationships, but really it comes down to how differentiated our engineering culture is. In terms of cutting back investors, I mean, as you know, when I evaluate an investor relationship, it really boils down to the value that they're gonna bring. The customer introductions that they're gonna make, uh, the advocacy that they're gonna help with.
- HSHarry Stebbings
How do you determine that? Everyone sells a good game. VCs, we sell cash. We get good at selling.
- AKAaron Katz
I reference them like you would in any other relationship.
- HSHarry Stebbings
Hmm.
- AKAaron Katz
So I talk to the companies that they've invested in in the past. I say, "What's it like to work with Harry? What customer relationships has he made that have been valuable? How has he helped with awareness from his social presence? Has he helped with recruiting? Has he engaged with your employees? How does he work with other investors? Do people perceive it as a positive to have him on the, on the cap table?" And those all need to be a unanimous yes before we start working together.
- HSHarry Stebbings
Do you think you raised aggressively enough? We're seeing a new world by capital. It's more and more a moat in a lot of cases.
- AKAaron Katz
You know, I'm trying to build a generational company that outlives me, and I think right now a lot of people look at fundraising as a very short-term exercise, and they wanna have this consistent and steady step-up in valuation. It's good for your employees. You give them liquidity through tender offers. It's good for recruiting. It minimizes dilution. It bolsters your balance sheet. Uh, it lets you forward invest. All of those things are true, but I'm thinking about a company 20 years from now, not two years from now. So whether or not we raise that 15 billion or 25 billion in 10 years is irrelevant, right? We need to have the right investors involved. We need to build a very durable, long-lasting, sustainable
- 38:36 – 39:33
ClickHouse’s Path to $1BN ARR
- AKAaron Katz
company. We need to get to a billion dollars of ARR as quickly as possible. The most important metric for the company is new customer acquisition. We add hundreds every month. Our gross retention is north of 99%. Our net dollar retention is north of 200%. The addressable market we're going after is absolutely enormous. So I think less about valuations perhaps than I should.
- HSHarry Stebbings
When will we hit a billion in ARR?
- AKAaron Katz
Within the next two years, if not sooner.
- HSHarry Stebbings
Give me a date. We can do a bet. We can both do a bet.
- AKAaron Katz
All right. Well, I'd put the over-under-
- HSHarry Stebbings
Yeah
- AKAaron Katz
... at December 2027, and I would take the under.
- HSHarry Stebbings
You think you'll get there before?
- AKAaron Katz
I do.
- HSHarry Stebbings
I'm gonna go for Feb '28. I think now that's like 18 mon- that's more than 18 months. Yeah, that's 18 months.
- AKAaron Katz
I would definitely take the under on that timeframe.
- HSHarry Stebbings
Would you really?
- AKAaron Katz
Oh, yeah. Yeah, yeah.
- HSHarry Stebbings
Really?
- AKAaron Katz
Yeah.
- HSHarry Stebbings
We're at 250 now.
- AKAaron Katz
We're well
- 39:33 – 46:51
Can Startups Compete With OpenAI and Anthropic for Talent?
- AKAaron Katz
north of that.
- HSHarry Stebbings
Oh, well, that's unfair. You said there about tenders. We see them more and more for employees, and I think talent acquisition is one of the hardest things today. I actually got in a lot of trouble the other day for this. I said, "If you are trying to hire A-star talent today, you can't. OpenAI and Anthropic simply pay, and they go to the front end model providers." Is that true, or was I being glib?
- AKAaron Katz
I think it's true, depending on the category that you're in.
- HSHarry Stebbings
Hmm.
- AKAaron Katz
I think if you're a digital native AI startup in San Francisco, it's a very difficult employment environment 'cause you're competing against OpenAI and Anthropic and others, um, that are extremely well-capitalized and are putting offers that are extraordinarily aggressive-
- HSHarry Stebbings
Yeah
- AKAaron Katz
... into the market. I think if you're a infrastructure provider like ClickHouse, uh, we look for a slightly different, um, profile. You know, we're looking for database engineers, people that have experience with distributed systems, um, slightly different than what the Frontier Labs are hiring for. We employ people in 27 different countries, um, which gives us a competitive advantage. So I can hire engineers in Portugal and Germany and Singapore. Um, you know, we've got single-digit, uh, attrition, so we've got extraordinarily high retention. Um-
- HSHarry Stebbings
So-
- AKAaron Katz
We have done some structured secondaries, um, and we'll continue to do so over time, but not with the frequency that I think some of the younger companies are doing.
- HSHarry Stebbings
So I am, yeah, controversial in many ways. Uh, one of them is because of my vocal, uh, expressions about remote work. Why am I wrong?
- AKAaron Katz
I don't think you're missing anything. I'm of two minds, and I contradict myself constantly about this topic because I spent 12 years at Salesforce. I was in the office every single day, five, six days a week, 10, 12 hours a day, and it was during this extremely formidable time in my career. I learned so much from those experiences being in the office amongst my colleagues and peers, learning from people with more experience than I do, than I had at the time. When I started this company, it was during COVID, and so we kinda had to be distributed by design, and I started it with some Europeans. And so people were in Europe. I was in the Bay Area. My co-founder was in Utah. And so, you know, w- w- we started the company. We grew very quickly. We found engineers that had a very unique skill set, and they weren't all in the Bay Area. They weren't all in London. They weren't all in New York, and so we built a distributed company. Fast-forward to where we are today, we're almost 800 employees. We'll be 1,000 by the end of the year. We are introducing, uh, in-person options for our employees. We don't have this draconian return to work mandate or return to the office mandate, but we do have offices, and we have a huge office in Amsterdam. We have an office here in London, New York, the Bay Area.
- HSHarry Stebbings
Is that specialized around different functions?
- AKAaron Katz
Not at all.
- HSHarry Stebbings
Really?
- AKAaron Katz
Yeah, both engineering and go-to-market, and we're seeing, you know, increased participation across the employee base and increased demand for office space. And if I look out a year from now, we're not gonna have six or seven hubs. We have an office in Singapore. We have an office in Sydney, Australia. We have an office in Tokyo. We're not gonna have six or seven. We're gonna have 16 to 20 offices around the world within 12 to 18 months.
- HSHarry Stebbings
If you could have your way, would you not have everyone be in office in some way?
- AKAaron Katz
I, I wouldn't and I'll explain why. Uh, we're a very international company, uh, by almost every measure. Over half of our revenue comes from outside of the US. 40% here in EMEA, 10% in Asia. Over half of our customers are outside of North America, and so we need to support our customers in a variety of different languages, in a variety of different time zones. I mentioned we're live in 36 different regions around the world across all three hyperscalers. There's no way that you can centrally manage that from one location. You need to have people in every single time zone. You need to have relationships with the hyperscalers in region. We go to market with AWS, we go to market with Google Cloud, we go to market with Azure. I flew to China to launch a partnership with Alibaba. Like, you're gonna have... You're gonna need local language speakers to maintain those partnerships, and you can't do it from one or two or three centralized hubs.
- HSHarry Stebbings
What phase of company growth was most uncomfortable? When were you the teenager at the wedding?
- AKAaron Katz
The most uncomfortable phase was immediately following our product launch. It was at the exact same time ChatGPT launched. And these database services are different than consumer services. They take time for companies to evaluate and adopt. It's not like you're just gonna use ChatGPT overnight, you're gonna get to 100 million users in two months. So there was this six-month period, and it, it shows up on that bar chart of revenue growth, where revenue was slow out of the gate. Now we've hit this inflection point and revenue's surging now that we have over 4,000 customers. But those first six months, you know, I raised $300 million at a valuation that was hard to justify 'cause we had no revenue. We had no product. We had no customers.
- HSHarry Stebbings
Your price is quite chunky. I mean, at, like, 50 million in revenue, you were priced at, like, $6 billion.
- AKAaron Katz
Again, just a point in time.
- HSHarry Stebbings
[laughs]
- AKAaron Katz
Like, you don't get credit for where you are.
- HSHarry Stebbings
I get your job-
- AKAaron Katz
Venture investing is-
- HSHarry Stebbings
... on my LPs
- AKAaron Katz
... you're paying forward.
- HSHarry Stebbings
My LPs are like, "Dude, but your DPI is not what, you know, uh, Gilly Renan's is." I mean, like, it's a point in time.
- AKAaron Katz
Right.
- HSHarry Stebbings
Yeah.
- 46:51 – 48:26
Is 0 to $100M ARR in One Year the New Standard?
- HSHarry Stebbings
that. Can I ask you something I struggle on and another thing that I get chastised for, is you said about the infrastructure slope on revenue or the slope on revenue being slightly slower because it's an infrastructure play. It's very normal. I often say that triple, triple, double, double's dead. You know, the one to three to nine to, you know, it continues is dead, and we need to be zero to 100 million in a year now.
- AKAaron Katz
Yeah.
- HSHarry Stebbings
Am I glib and wrong, and that won't take into account infrastructure plays? Or is that actually just the new world that we're in?
- AKAaron Katz
Well, as I mentioned previously, I think the durability of revenue is the most, uh, underestimated attribute of these companies. Meaning, how high are the switching costs when somebody's using your product? And for any category that goes from zero to 100 million in a year, I worry what's the competitive moat that they have to preserve that 100 million from that customer going to something else. So ours was a bit more gradual. We went zero, 12, 50, 200, and we'll finish this year north of 500. Which, in the database world, is faster growth than we've ever seen, including all of the competitive companies that I mentioned earlier today in terms of the first three years of revenue growth, and it's over a very broad customer base. So we've got very little concentration risk. The basket of AI companies that's using us, and nearly every AI company is built on ClickHouse, from Harvey, Sierra, Decagon, Anthropic, OpenAI, et cetera, represents less than 12% of revenue. And so even if half of that goes away, the winners are gonna offset
- 48:26 – 53:44
Does Revenue Concentration Still Matter in AI?
- AKAaron Katz
the loss from the losers.
- HSHarry Stebbings
Is concentration risk a valid investor concern? You know, you see a lot of people say like, "Oh, yeah, I'm an investor in McCaw. Oh, McCaw's 90% of their revenue comes from frontier model products." And I'm like, "So does Nvidia's. Seems to be working for Jansen." Is concentration in terms of revenue a valid investor concern anymore?
- AKAaron Katz
I think about it as an operator as a very valid concern. If I've got one customer that, or one category or one industry that accounts for more than 10% of revenue, I spend a lot of time thinking about it and that seems to be kind of the industry standard, that threshold that if some dimension of your revenue base accounts for more than 10% of your revenue, you've got exposure and I wanna limit exposure, right? The goal is predictability, sustainability, durable growth, and if I've got one category or sector or customer that can have such a negative effect if they were to leave the platform, that's a concern for me.
- HSHarry Stebbings
We said about switching costs there, and I think one big mistake often investors think is, "Ah, you know, once I get to a certain scale, I'm gonna move off Elastic and build my own. Once I get to a certain scale, I'm gonna build my own payments processor." Shopify still use Stripe at the scale that Shopify's at and actually most often you just don't because it's not your core business. I'm intrigued how you think about that, especially given your time with Elastic and whether we do overestimate the, "I'll just move at scale."
- AKAaron Katz
Yeah. It comes down to customer value, right? You need your customers to continually see value from your service which means you always need to be ahead of your competition in terms of the ROI and the TCO calculation that your customer's gonna think about.
- HSHarry Stebbings
On TCO?
- AKAaron Katz
Total cost of ownership.
- HSHarry Stebbings
Ah.
- AKAaron Katz
So if you think about how much does it cost to run ClickHouse, how much does it cost to run a, a, a c- comparable service? And you wanna have that advantage, right? Those are the two primary attributes that you're gonna think about in terms of value creation for your customers.
- HSHarry Stebbings
Do you think it's important to have an internal enemy?
- AKAaron Katz
I don't. I, I worry about creating, uh, adversity inside of a company.
- HSHarry Stebbings
Really?
- AKAaron Katz
Well, you have so much adversity outside of your company, right? You gotta deal with competitors disparaging yourself and the company. You've gotta think about how you stay ahead of the competitive landscape. You gotta think about geopolitical concerns and that's before you have any sort of personal strife in your life. That's just work. So why would you want to introduce that into your company? I wanna make my employees' lives better.
- HSHarry Stebbings
To create fire-
- AKAaron Katz
Not more difficult
- HSHarry Stebbings
... in them that they wanna beat someone.
- AKAaron Katz
Oh, that just comes down to hiring the right people. Like, we only hire people that are insanely competitive.
- HSHarry Stebbings
Has your hiring changed in an AI world how you determine talent, how you discover it?
- AKAaron Katz
You know, I think we're a little bit unique. The average age of our engineering organization is in the mid to late 30s which is a little bit different I would imagine than in a, a typical venture-backed company that's only had a product for three years. So we typically look for people that have a bit more experience. Um, it's not to say we don't bring people into the company straight out of college, but we wanna make sure that they're paired up with, you know, somebody that's been building distributed systems for a period of time.
- HSHarry Stebbings
A lot of people question the value of college today. Do you think that's justified?
- AKAaron Katz
It's something I think a lot about. We've got two teenage daughters and so, uh, as they think about college-
- HSHarry Stebbings
Would you tell them to go?
- AKAaron Katz
I would but I think it's simply around the life experience. You know, the, and again I don't wanna take anything away from like the academic, uh, output of a four-year degree. Uh-
- HSHarry Stebbings
But with the greatest of respect, you should and I mean that, and again this is why you're popular and I'm probably controversial. Like you should, like the, the curriculum process to update them is so long that by the time it's been updated it's already out of date.
- AKAaron Katz
I think it depends on what you wanna study.
- HSHarry Stebbings
Well, if you're studying obviously neuroscience then it's relatively important.
- AKAaron Katz
Yeah, if you wanna go into medicine I think that we're gonna need doctors. You want that human interaction, right? If you're studying software engineering I think you're gonna be entering a market that's very uncertain in four years' time.
- HSHarry Stebbings
Do you think this could all get a bit creepy? You said that we're still gonna need doctors. Well, I'm not really so sure if I'm totally honest. Like I use ChatGPT for most of my medical queries now and it prevents me from seeing a lot of doctors. Uh, doctors are pretty unaffordable to most people. Wait times in the UK for a GP are months.
- AKAaron Katz
Yeah.
- HSHarry Stebbings
I don't know, dude.
- 53:44 – 1:05:46
Quick-Fire Round
- AKAaron Katz
I think that will be the future.
- HSHarry Stebbings
So listen, we're gonna do a quick fire round so I say a short statement, you give me your immediate thoughts. Does that sound okay?
- AKAaron Katz
Yeah, why not?
- HSHarry Stebbings
What job does not exist today that you think will be extremely common in five years?
- AKAaron Katz
An AI finance function.
- HSHarry Stebbings
Hm.
- AKAaron Katz
Solely dedicated on AI consumption inside of an organization. That's all they wake up thinking about.
- HSHarry Stebbings
In terms like token resource management?
- AKAaron Katz
Correct and then that job will be, uh, made irrelevant five years from then because AI agents will govern themselves.
- HSHarry Stebbings
Which job will never be made irrelevant? VCs like to think it's ours.
- AKAaron Katz
Professional f- football players. [laughs]
- HSHarry Stebbings
[laughs] Dude, I'm 30. It's too late for me now.
- AKAaron Katz
Professional athletes aren't going anywhere anytime soon.
- HSHarry Stebbings
Um, in the UK we have a game, okay? It's called, and forgive the crassness, Shag Marry Kill. Um, you know, shag is short-term buy, uh, marry is long-term buy and kill is, you know, a la pubelle. No, not for me.
- AKAaron Katz
Yeah.
- HSHarry Stebbings
Um, you have Meta, you have Microsoft and you have Nvidia. What is your shag, marry, kill on them?
- AKAaron Katz
Which one would I shag, which one would I marry, and which one would I kill? I met as a customer, so I don't wanna put them in the latter category.
- HSHarry Stebbings
[laughs] Fine.
- AKAaron Katz
Microsoft's one of our power users. We power the largest analytical workloads at Microsoft, so I'll probably marry Microsoft. I had the opportunity to meet Satya, uh, a couple months ago, it was quite impressive. Um, [tsks] and, uh, we do business with both Nvidia and Cerebras and a lot of the chip manufacturers, so I'd... I mean, I'd like to shag all three of them, but unfortunately-
- HSHarry Stebbings
[laughs]
- AKAaron Katz
... un- I, I don't, unfortunately I don't think that's possible.
- HSHarry Stebbings
Do you think we'll see a much more distributed chip ecosystem in the next few years? You see Etched and you see a lot of other providers, Cerebras being one of them.
- AKAaron Katz
Yeah. I think a lot of the hyperscalers and frontier labs are gonna be very relevant providers in the chip ecosystem.
- HSHarry Stebbings
What's one widely held belief about AI that's pretty agreed upon that you think is actually pretty wrong?
- AKAaron Katz
A widely held belief around AI that's generally agreed upon but is wrong is that it's overblown, and that we're in a hype cycle, that we're in a bubble. And I'll kinda finish it with where I started. We're just getting started. Like, I, you know, I can't pick the winners and losers, um, but I think the winners are gonna far offset the losers.
- HSHarry Stebbings
Do you worry about the levels of debt being taken out exceeding a- any historical norms?
- AKAaron Katz
I, I worry about the public exposure to these companies, um, when they're publicly accessible. Uh, right now it's private capital. Um, and so, you know, a lot of investors stand to lose money, a lot of investors stand to make a lot of money. But I worry about-
- HSHarry Stebbings
Totally, but Nvidia represents a huge amount of 4- 498s for a lot of Americans, no?
- AKAaron Katz
Nvidia?
- HSHarry Stebbings
Yeah.
Episode duration: 1:05:56
Install uListen for AI-powered chat & search across the full episode — Get Full Transcript
Transcript of episode Wna2Ds4dt50
