Elastic’s Generative AI Momentum

Daniel Newman: Hey everyone. The Six Five is On The Road here at AWS re:Invent 2024 in Las Vegas. It’s been a great week. Lots of announcements. Heard so much about all the things that are happening in the AWS cloud ecosystem, and of course AI has been a central focus. One of the things that we’ve been spending a lot of time during this week doing has been talking to the partners. As we know, AWS has a rich and vibrant partner ecosystem, and I have the chance, the opportunity to have one of my friends back on the show, a partner, actually an award-winning partner this year at AWS re:Invent. I’ve got Ash Kulkarni, CEO, Elastic. Ash, welcome back. It’s been a minute, but it’s great to see you.

Ashutosh Kulkarni: It’s been great. Thanks for having me.

Daniel Newman: So we’re going to hit AWS re:Invent, but maybe before that you and I, you’ve been on my market show, we’ve talked a number of different times. It’s been great to watch the company’s growth. You had earnings what, just about two weeks ago?

Ashutosh Kulkarni: That’s right.

Daniel Newman: Q2 really good results. The market seemed to really like it. Give me the quick rundown. I have to imagine this enthusiasm has something to do with the AI strategy that you’ve been able to execute upon.

Ashutosh Kulkarni: Yeah, AI was absolutely a big part of it. So in terms of the numbers, we grew top line revenue by 18%. Our cloud business grew by 25% and we delivered a strong operating margin. So everything was greater than what the street was expecting. We were really happy with that performance. The team’s been executing really well on multiple fronts. One part of it has been on leading the charge on becoming the platform for what’s called retrieval augmented generation, RAG, as it’s often referred to. This is about building general AI applications. And then our go-to market engine has been really leaning into that. So our customer base is responding wonderfully. So absolutely, AI has been just this nice tailwind.

Daniel Newman: Yeah, absolutely. And it’s interesting, Ash, is I think a lot of investors have been searching for companies, pun intended, that they can attribute AI in a very meaningful way. And what do I mean by that is I think there was a lot for a period from about the time of ChatGPT’s onset, there’s been a lot of this AI washing, oh, we have an AI thing, and then you’d get earnings call to earnings call or you’d listen to the transcript or read the transcripts, listen to the conversations, and the question was how much is AI actually moving the needle for your business?

Now, you and I talked over the course of a couple of years and one of the things I said is you guys were sort of a company made for this moment. So when you think about what you’re doing with search and with RAG, these are two of the biggest problems is how do I actually extract meaningful data that helps me drive my enterprise? So it seems like you’ve met the moment. I am guessing you’re probably hearing that from some of your biggest investors and partners.

Ashutosh Kulkarni: We clearly knew that there was an opportunity because at the end of the day, if you think about what Elastic has always been about, we’ve been about helping our customers find insights in the messy, unstructured data, Word documents, machine generated log files that are everywhere within the organization. And the hardest part of all of this was how do you get insights in a meaningful way from all of that data without drowning in it? And large language models became that one missing piece that was the unlock. And when you take these large language models and combine them with a search AI technology like Elastic, you’re able to very quickly find just the right nuggets within your enterprise data and pass it to the large language model. And now that large language model can respond to you in a conversational style just like ChatGPT, but for meaningful business applications, business workflows. And our customers are loving it.

One of the stats that I gave in our Q2 earnings was we now have over 1,550 customers just on Elastic Cloud that are using us for various generative AI applications. Our customer commitments almost doubled quarter over quarter, Q1 to Q2. So all of those are really great signs. And of those commitments, three of those deals were million dollar plus deals. So we are seeing customers now really lean in and use these applications in ways that are ROI generating. They are value generating for them and these are enterprises. So it’s been great.

Daniel Newman: Yeah, you’re able to basically attribute growth to a specific set of solutions that are AI.

Ashutosh Kulkarni: That’s right.

Daniel Newman: And that’s that 1500 plus customers. But then on top of that, your RPO is growing as a by-product. And of course everybody that’s invested in that particular space cares a lot about is that backlog building, is that demand building. And so you’re putting this together. I think part of it too, we talked about you being a partner, award winner, ecosystem matters because not everybody is outgoing looking solution to solution to solution. The marketplace and things like that have been really successful because people are kind of building their enterprises in the control plane of the cloud and then they’re looking which solutions can I integrate. Talk a little bit about what you’re doing with the Elastic AI ecosystem because I think it sounds to me like you’re sort of accelerating the partner ecosystem to get more people adopting your solutions.

Ashutosh Kulkarni: The ecosystem is going to be incredibly important for multiple reasons. So first is this is a very, very new space and anytime you’re dealing with a new space like this, the innovation is unparalleled. It’s happening everywhere. Large organizations like the cloud hyperscalers that are building their own large language models, you have companies like Anthropic and Cohere and so on also adding to that mix open source companies like Mistral, but then there’s an entire ecosystem of things like how do you guardrail and protect the output of a RAG system. Companies that do LLM orchestration and all of these taken together is important for this ecosystem to move forward and really for customers to build complete applications that they can deploy in the enterprise. So the way we’ve looked at it is we want to be the most open player in this ecosystem, build really deep integrations with everybody that matters within this environment.

So our customers don’t have to do the integration. The integration just works out of the box. So it doesn’t really matter whether you want to use OpenAI’s LLM or Mistral or whether you want to send some queries here and some queries there. But all of them work really well with the Elasticsearch vector database and our platform to deliver the most optimized RAG solution with minimum hallucinations. And that’s working really well. Our customers are loving it.

Daniel Newman: So the API connectivity’s that can run on a combination of these kind of closed black box as well as in Open. But you’re still very open and you can expose what you’re doing to them. So if someone wants to use OpenAI for business, they can still get the benefit for their custom internal data, the high value data, because that’s really where it’s all at. I mean, to some extent, I don’t know how you feel about this, but I bounce from LLM to LLM. I play with them all. I use Perplexity, I use Gemini, I’ve played with the AWS models. My point though is sometimes you’ll find that one you’re like, oh, I like this one for this thing. I like this for research, I like this for code generation. I like this for getting financial data quickly. But there is a bit of a table stakes element and I think that acceleration to how do we get our data, whether it’s distilled models, whether it’s in RAG, whether it’s fine-tuning so that the outputs has something that’s unique, otherwise it’s all the same.

Ashutosh Kulkarni: It all comes down to your data. If you can help monetize your data, drive automation in your business processes, that’s the name of the game for enterprises. So the data itself is going to be most important. And that’s where we have an amazing advantage in terms of just the incumbency that we have. Out there Elasticsearch today is easily the most widely downloaded vector database. There’s so much data already sitting in Elasticsearch and we’ve made it extremely easy for customers just with a minor change to take everything that they’ve done in Elastic for search and now turn it into a RAG style application. And now with that, we give them the choice of using any LLMs and in my opinion, a few things are going to happen or need to happen for this space to truly become everything that it can be. First, the cost of inference and the cost of running these RAG applications needs to keep coming down because without that, you’re going to see things capping out.

The second thing is privacy and security are going to be incredibly important. Without that, no enterprise is going to really put sensitive data through these applications. And that’s where we are going with our solutions. It’s the most privacy-sensitive and secure solution out there. Everything from document-level permissions to very fine-grain, rule-based access control and then our ability to continually drive the price down for our customers just means that they’re doing more and more with us. And the last quarter, I talked about an automotive company that’s building 32 different chatbots for different use cases on Elastic. And to your point, they’re picking different LLMs for different use cases because I don’t believe there’s going to be one thing to rule it all.

Daniel Newman: If there is, then there’s a lot of companies that are going to lose badly at this point ’cause there’s so much attention. But I think there is enough distribution of use and it’s really interesting and I think it’s worth kind of just a quick double-click, is what you’re effectively saying is what you’re able to do with RAG and Vector in the Elastic database can allow companies to use substantially less compute networking and resources because you can get the right data exposed to the application in a way that’s available that doesn’t require so much horsepower.

Ashutosh Kulkarni: That’s right.

Daniel Newman: Because the bigger the model, the more compute required. So I always say Nvidia makes great stuff, but it’s an F1 vehicle. And so it’s great on the strip here, we’re in Vegas, but if the strip’s got traffic, it’s no fun driving an F1 car. It’s got to be open.

Ashutosh Kulkarni: You got it. You got to bring the price down and you got to make sure that the answers are accurate and you’ve got to reduce the risk.

Daniel Newman: That’s why there’s options for different solutions. It doesn’t mean that one won’t continue to be great, it just means that the market’s going to get bigger and there’s going to need to be different options. That’s why we’re hearing so much about smaller models, stilling models is because it’s not efficient and obviously we’ve got an issue. We only have so much power, all these liquid-cooled racks going in data centers, but sometimes you just need a very efficient architecture and it sounds like you’re getting there. So let’s talk a little more specific about AWS since we are here. How is the partnership, which seems to be very strong, influencing your strategy? What are you learning about this as AWS is going through their own transformation with AI?

Ashutosh Kulkarni: Yeah, so first of all, the news of the week for us was we won the AWS Global Partner of the Year for generative AI and data, which was great. So the recognition that-

Daniel Newman: Congratulations.

Ashutosh Kulkarni: Thank you. In their ecosystem, they see us as one of the absolute leaders in innovating in the area of vector databases, in RAG and so on. That’s been fantastic. And the way we have always approached AWS is first and foremost, they’re a wonderful marketplace and channel for us to work through because our customers, many of them want to run Elastic, our platform on AWS. So we make it possible for our customers to purchase Elastic through their commitments that they’ve already made through the marketplace to AWS. And I see it in three phases. I always think about it as build with, market with and then sell with. And you have to start with the customer experience. So that’s all about building these deep integrations and that’s what has really helped us absolutely improve our experience and our relationship with AWS like never before. This year, we had a prime spot when it came to the convention hall where we were exhibiting. We are a diamond sponsored this year, and it was just fantastic to see the foot traffic. I’m very excited about what this means for the future.

Daniel Newman: Other than that long walks around this event, which anyone that’s been to-

Ashutosh Kulkarni: I got my steps in.

Daniel Newman: Yeah, definitely one of the most compelling kind of partner exhibits out of all the events I attend every year is this one. The ecosystem AWS is built is second to none.

Ashutosh Kulkarni: It’s amazing.

Daniel Newman: It is really, really well done. So let’s kind of move a little bit more to the broad Elastic viewpoint here. You and I have talked, like I said a handful of times. I think you rose from the engineering and product side. In your last answer I could hear it, a salesperson might come from a different lens and you can definitely sense that, but how do you sort of now looking at it through the lens as CEO, you’re a couple years now, right?

Ashutosh Kulkarni: Three years.

Daniel Newman: Yeah. Gosh, it’s gone fast.

Ashutosh Kulkarni: Time flies.

Daniel Newman: Congratulations. But sort of this progress in transformation, how do you kind of evaluate it? How do you grade it? How do you characterize it?

Ashutosh Kulkarni: So I take a much broader perspective on the opportunity ahead of us. When I think about what we represent as a search AI platform, Elastic is all about helping our customers with unstructured messy data, like I said, and there’s multiple orders of magnitude of that unstructured data in most organizations than there is structured data that sits in databases that’s used for BI and so on. But the challenge ultimately has always been how do you find the right insights out of that data in real time? And that’s been Elastic. That’s been our core strength. When you add the power of large language models to that mix, now all of a sudden you can present a very conversational way for people to explore that data. And then it unlocks all kinds of fascinating use cases.

Observability, which now represents 40% of our business. This is all about very quickly finding what’s in your environment that might be slowing down your application. How do you monitor your systems, keep them running at the optimum speed and so on. Security, which now represents over 25% of our business. This is all about cyber security. How do we prevent threats? So for me, the opportunity landscape is just massive. My aspiration is that there will be a day when every single organization, big and small will need a search AI platform and we will be it. And that’s what we are going to work towards.

Daniel Newman: Yeah, it’s very aspirational and it’s also a massive opportunity. In the end, most people will consume AI through the lens of some sort of managed service, whether it’s the SaaS application itself or a managed service. I mean the DIY companies, we’re seeing them. We know who they are. They’re building these massive clusters, these racks, these tools. But most companies, the way SaaS had its own sort of revolution and even enterprise software did, it’s about making the business need available and making it simpler to consume. And so I think that’s where you’re heading. And I mean, I think that’s the opportunity.

Ashutosh Kulkarni: It’s a multi-billion dollar opportunity. So we are very excited about it.

Daniel Newman: So is that what you’re most excited about as we head into 2025? Give me across the tech and AI and Elastic landscape, give me the most exciting thing for the year ahead.

Ashutosh Kulkarni: From our perspective, the two things that are really driving our momentum, first and foremost, it’s genitive AI like we talked about, but it’s not just about customers going from search to semantic search to these RAG style conversational applications and then agentic workflows for entirely automating business processes. That’s one big pillar for us, and it’s been just fantastic to see the momentum. The second pillar for us is we are infusing AI into our observability and security solutions. And that’s helping us disproportionately win and take share both in security and observability. And we are seeing customers consolidate onto our platform, displacing incumbent vendors in those spaces. And so we are leaning in hard into this idea of a search AI platform. So that’s really exciting for me.

But long-term, what I also see is the industry is now getting a lot of things right. You are seeing this week at AWS at re:Invent, there were a lot of announcements around custom silicon. And when you see all of that, what’s going to happen is the price of compute is going to come down, the price of inferencing is going to come down, and that will benefit every layer of software that sits on top of this hardware. It’ll benefit companies like Elastic and effectively it’ll raise all tides. So I’m super excited about what this means for the whole AI ecosystem.

Daniel Newman: And I love that you called that out. We’ve been evaluating and watching that part of the market very, very closely. First of all, that whole segment’s going to grow a lot. And the idea that you can basically have workload-specific silicon, which you always could, but there was kind of idea that everything needs to be a giant GPU. And it’s like, no. I mean a lot of use cases, I mean, heck, you can still do a lot of inference on a CPU.

Ashutosh Kulkarni: Absolutely.

Daniel Newman: But I mean there are these very efficient inferencing in the cloud, chips, managed services, and it’ll certainly benefit Elastic and the industry as a whole.

Ashutosh Kulkarni: Effectively what you’re seeing here is things that, you’re not seeing displacement of activities that we were doing before. You’re seeing net new things happening. You’re automation of workflows that just were so tedious and so painful. Yeah, I talk about we have customers who use us to build chatbots for customer support and call center automation and so on. And yeah, I always ask this question, you talk to a call center representative that’s worked a shift and see their energy level at hour one and their energy level at hour seven, there’s a marked difference because nobody calls a call center to thank them. They usually call a call center because they’re upset about something. So if you can automate the grunt work there and allow that person to actually deal with the harder problems, that’s a huge unlock of human potential. And now you’re driving meaningful return on capital, it’s going to raise all tides. This is what really excites me about AI.

Daniel Newman: Yeah, nobody after an eight-hour shift of being yelled at that their dog food didn’t show up as feeling good about themselves. But yeah, absolutely. Ash Kulkarni, thank you so much for joining me here on The Six Five.

Ashutosh Kulkarni: Thank you.

Daniel Newman: It was a lot of fun to have you here. Congratulations on everything.

Ashutosh Kulkarni: Thank you.

Daniel Newman: And thank you so much for joining us for this episode of The Six Five. We’re On The Road here at AWS re.Invent 2024. Hit subscribe. Join us for all of our coverage here, and of course, be part of our community and watch all of our shows when you have time. But for this one, this episode, time to say goodbye. I’ll see y’all later.