The Neo Show: Founder Fridays #16 - Phil Pergola with Hosts Howie Xu & Axel Pilette

This is the Neo Show, where the minds shaping tomorrow speak today. Brought to you by Neo, the browser redefining what AI can do for you.

Yeah. Well, first, hi everyone. I'm Phil Pradola. I'm the CEO of Cloud Zero. Cloudzero is a B2B software company. What we do is we help companies get visibility into what drives their public cloud data and AI spend. And with that visibility take an engineering approach to not only optimizing that spend but improving their unit economics. And so it's great to be here today.

Yeah. So, Howie, it's as you mentioned. I've been in the public cloud since 2017 and the earliest days of the public cloud, even going back to like 2008, I've been helping companies in this area. And so, what you've seen, for me focusing the last eight years, what I've seen is sort of a shift that's happening. A couple of shifts that are happening. One is companies look to optimize their cloud infrastructure primarily focusing on things like compute, network, storage. That's really evolved into more the platform area. First and foremost, you've seen the explosion of these database clouds like think the MongoDBs, the Snowflakes, the Data Bricks of the world. Now you're seeing the model companies out in the cloud, and those are taking different shapes. One, what you see like the OpenAIs, the Anthropics of the world, or you see companies like the CoreWeaves of the world that are like the neo clouds hosting GPUs. You're starting to see a lot of focus there. And what also happens is that through each generation, the problems of the old generation actually don't really go away. So all the same challenges people had with making sure they're using the same compute, network, storage, how you allocate shared costs or manage multi-tenant costs, that doesn't go away. Then you see something like Kubernetes come out, it just makes it harder. Now you're seeing kind of the new clouds. So the world is getting more complex, but the newest things have been the AI clouds. And then before that, the explosion of data, and that data explosion will continue because many of us know much of what we're doing with AI isn't possible without great data.

Yes. I joined Cloud Zero in November of 2021. And at that point in time, I had been an executive at multiple B2B SaaS companies. I had been in the infrastructure space for much of my career. And I was really looking at what's next. And I knew that I wanted to continue to focus on infrastructure as an area. Things like some people like to say that the picks and the shovels, focusing on that area. And so that was one thing, and I continued to see this challenge having been in the same space for four years, I continued to see the challenges getting larger and larger. Just to put some context around that, companies this year will spend about 7 to 800 billion dollars in the public cloud across platform infrastructure SaaS. When I was taking the job back then, it was less, but about 30% of that is waste, and if you apply 30% to 7 to 800 billion dollars, it's greater than the GDP for most nations in the world, so it's not insignificant. And the forecast is that companies are going to spend about $2 trillion dollars over the coming years by about 2030, and so this problem is only getting harder. And as someone who has had a career in working for startups that eventually went on to have successful exits, there's a huge market opportunity that exists, and I was really excited to be the CEO of a company attacking this big market opportunity.

Yeah. So, it's a good question. We spend a lot of time looking at a very simple formula. And it's very simple on its premise, which is if we can get 10% of the service addressable market or about 5% of the total addressable market, and we can convince companies to pay us 1% of that spend to help them manage it, that's a billion dollar revenue company. And if you look at the savings that we deliver our customers, oftentimes it's 20, 30, 40%, so the fee is certainly justified by the outcome. So that's sort of one perspective. I think when you look at it through the challenges, one of the things that we are really focused on is making sure companies understand that being efficient with how they build software in the cloud is not competitive with innovation. And so for many companies, it's where do they want to focus their time and energy? What does good enough look like? In a vacuum, each of the large hyperscalers will attempt to provide value to their customers through free tools, through consulting services. The reality is that most companies who are architecting or building an application in the cloud today are doing it across cloud providers. Maybe they have a primary hyperscaler that they use for their infrastructure platform, but very quickly as you start to layer in security, databases, you start to think of the management tools like DataDog, things like that, it quickly becomes a multi-cloud problem. And then the other piece that I think is really missing, which right now I think we're still in the early adopter stage, is starting to tie the unit economics of the business back to that spend. So the problem right now is that people look at the bill and they think, 'Wow, I'm spending a lot of money. I'm spending millions of dollars in the cloud. That's a bad thing. How do we reduce our bill?' Well, the reality is if you're a company like DoorDash doing deliveries, if your cloud bill goes up because you have more deliveries, that's not necessarily a bad thing. If you're a public company and your revenue doubles and your cloud spend triples, that might be a bad thing. But on the flip side, if you see a 25% increase in your revenue year-over-year and a 5% increase in your cloud, that might be a very healthy thing. And so what companies are starting to understand is that it is possible with solutions like Cloud Zero to actually tie the spend, which is primarily being driven by their business customers consuming products, features, transactions, and their engineering decisions. It's possible to tie all of that back to the unit economics of their business. Now with solutions like Cloud Zero, they can understand the difference between the healthy decisions and the unhealthy decisions. And so engineers, as they're building code, as they're building applications, have the context that they need to make good decisions. And oh, by the way, just like they're building secure software and performance software, building efficient software through the lens of spend is actually a good thing. And that's what good innovative companies do. And it's not at the expense of pace of innovation. And you can often use these unit metrics as a simple way to govern that. So you can empower engineering teams to move really fast but at the same time make sure that the decisions that they're making are in line with your business goals.

I did. I actually have a really big problem, Howie. I was trying to figure out how I get my family of five to Orlando, Florida for Thanksgiving to go visit my family down there. And I actually was using Neo and exploring different ways that we might be able to get down there in a cost-effective way. So I did have some time to explore Neo.

Continue.

Yeah. When I think about what AI means to me, I can look at it through the lens of the typical B2B SaaS company and then there's a nuance that's a bit unique to our business. So first and foremost, what AI means to us is we're looking at how does AI impact the productivity of our employees. And we're a fast growing organization. We're hiring lots of people. Our focus is really around how do we increase the productivity of the people that we're bringing on by giving them AI. So what does that look like? We're a software company. So that could be our engineers using different co-pilots. It could be our salespeople. For example, if I am meeting with a prospect, one of my sales executives asked me to meet with a prospect, the good sales reps will use generative AI to prepare a briefing for me that used to require a lot of manual effort on them. Doing research, pulling together information from our internal data sources. Now they do that in an instant and then I get a really nice PDF which is, here's everything I need to know before I meet with this prospect. So our pre-sales teams, our post-sales teams, our customer success teams, all of them are using generative AI. A second lens is how do we use AI inside of our product? And when we think about AI, we think about it both in the context of more traditional AI like machine learning as well as some of the more forward generative AI use cases. So we think about how are we using AI in our products. So on one side, for example, we have time series data, so we're doing things like anomaly detection would be a simple use case. Using AI to help customers forecast what their future cloud spend is going to be. But also things like instead of interacting with a traditional data explorer or BI tool, having a natural language conversation about what drove our cloud spend to increase. And so we're embedding AI into our product in lots of different ways both from a user experience perspective but also behind the scenes in terms of things like MCP servers. So that's embedding AI in the product. So the first use case is productivity of the employees. Second use case is embedded in the product for the benefit of our customers. And then the third use case for us is that our customers, because of the nature of what we do, they are consuming AI associated with that. They are driving up the number of tokens that they are consuming. They are driving up the associated spend. And through that, we're helping them understand the difference between healthy AI spend and unhealthy AI spend. The reality is that most companies right now, most businesses, they're not looking to reduce their AI spend. If anything, they're getting pressure from their boards to invest more in AI, in which case the bills are going to increase. But what companies are starting to understand is it's important to understand which of those AI investments have some type of return and which ones do not, and how do they reallocate their investments to either try new experiments or invest in the things that are working. So we have this third category.

Yeah. Examples of helping our customers from a value perspective in terms of reducing their spend, and I'll use some AI examples as well. First and foremost, one of our typical customers might have tens, hundreds, thousands of software engineers, all of whom are interacting with the cloud, with data, with AI, and they will often times spin up these resources that are like a utility. So for the non-technical folks, think of it as you enter a room and you turn the lights on, and then when you leave the room, the lights are still running. For some of our customers, that activity of turning the lights on might cost them tens of thousands of dollars a day, which is significantly more than what running a light in a typical room might cost. And so one engineer might make a decision that costs tens of thousands of dollars a day. And those resources might be the equivalent of leaving the lights on overnight. And we just need to shut those resources down. Or in other cases, using AI as an example, some models cost 10 times more than other models. So understanding the cost profile and then tying it back to the business might help better inform which models they might want to use for a particular use case. Again, our customers are primarily businesses. Profitability matters. Not all of them are profitable, but making sure they have healthy unit economics. So for example, if you're like a DoorDash, the cost per delivery, if you're someone like an Uber, a cost per ride, if those economics don't make sense, then their businesses don't make sense. And so it's really important that companies are building a solid business that can be sustainable, and we're helping them get the data to understand that they're doing that. So through both visibility as well as uncovering things like you've left the lights on, we're helping our customers not only save money but be more efficient.

Yeah. So what's here and now right now, and I've seen this play out over the last decade, and it's not unique to the cloud. It's that a really big bill shows up. So what's happening now and what has happened in my experience over the past decade is at some point in time a really big bill shows up. I'll just use my own business as an example. We're doing experimentation with lots of AI providers. I was experimenting with Neo recently. We work with Anthropic, we work with OpenAI. We're experimenting with some others, we're experimenting with different data providers. One day finance gets a bill and says, 'Whoa, where did that $10,000 bill come from? I wasn't expecting that.' And then what happens is they end up going to the engineering teams and they say, 'Hey, what happened there? I was caught by surprise. Is that going to continue? Is that something we need to plan for?' And those are the questions that start happening with a lot of the consumption of AI. Right now we're still in the early days. I was actually talking with someone yesterday and they had commented that they had used a 100 billion tokens and I asked a very simple question. I said, 'How much did that cost? How much did a 100 billion tokens cost you? Is it thousands a month? Is it tens of thousands a month? Is it hundreds of thousands a month?' And this person was like, 'No, it's costing our company like tens of thousands a month.' And I said, 'Oh, is anyone starting to pay attention because you're spending tens of thousands of dollars a month on something?' And she's like, 'Oh, yeah, totally. When it went from thousands to tens of thousands, we started getting phone calls.' Which is like, 'Hey, what's going on with this bill?' And so oftentimes what happens is you see the bill surprise being the thing. And what happens often times is that the finance teams will all of a sudden say, 'Whoa, hold on a sec, we got to slow this thing down.' And then what happens is the innovation slows down. And that's not good for the company because nobody wants their innovation to slow down. It's not good for the vendor whether it's an OpenAI vendor or a cloud vendor or an AI vendor, a cloud vendor, a data vendor. It's not good for them because they're not getting revenue. It's not good for the end customer who is the beneficiary of that innovation. Nobody wins when that slowdown happens. So part of our mission is to really eliminate the bill surprise and eliminate what I call the trough of innovation, which is that period of trying to get control of the bill and then getting back to a point where companies feel like they have good governance around it and they can continue to invest with confidence. Because inevitably they always get there and they always continue to invest. But we need to try and eliminate this trough of innovation. We need to try and eliminate the bill surprise.