Informatica Inc. (INFA) Earnings Call Transcript & Summary

Thank you so much. Small and cozy gathering here, Eric Brown from Informatica. I want to -- I think I told you the story of how I got to know Informatica. Did I?

No.

Okay. So 1995, I want to panel with a cop analyst that talks about what are the things that a private company should do to be a successful stock in the public markets. And I give the analyst side. And there is a gentleman who happened to have founded a technology company just talking about how he went about founding this -- coming up with this idea that led to this company. That was Diaz who founded Informatica, cofounded with Gaurav. So I feel like I've been associated in some way, shape or form from the inception of Informatica in 1995.

Yes. IPO 1 and IPO 2.

That's right. IPO 1, IPO 2. Never got to work on the IPO 1 but 2, certainly, yes.

Before we get into the mechanics of the quarter commentary, et cetera, I just wanted to take this opportunity for you to level us on what -- what does Informatica look like? Just give us an overview how the company has changed post the first buy-out through the stages of private equity ownership? What are we investing into at this stage at a high level?

Sure. before I start with all the things that have changed, I'll just mention the things that have not changed. We've maintained a singular focus on just data management. And we've always been known as the kind of Switzerland of data management, meaning we partnered equally well with people above us and stack, the app vendors, and people below us, the databases, now the cloud databases, the hyperscalers, et cetera. And so we've always done just one thing and focused on that over our 25-plus year history. So that is the one constant. I think what changed at the time of privatization. I think we're actually a classic case study for what you can do under private ownership. In our case, we undertook a complete product and business model transformation. So pre-privatization in August 2015, we were a professional license led company. $350 million here professional license, large, chunky deals that you're trying to get across the finish line at the end of the quarter and maintenance and virtually no subscription business. What we built while we were private is a complete cloud-native platform for data management, we call it IDMC. And while we are building that organically, I might add, so we spent about $1 billion-plus of R&D, really didn't do M&A. So $1 billion of R&D. And again, we're always been very profitable companies. We're able to fund that level of R&D spend. We built IDMC and then transforming the business from perpetual license to initially subscription. And now inside of subscription, we're transforming it to cloud. And I went back actually and looked at the presentation we did for the rating agency circa 2015. For privatization, we described our TAM to $6.5 billion, more of a monolithic product line company centered on PowerCenter. Today, our TAM is north of $40 billion. So we've dramatically expanded the part of the portfolio, and our growth in subscription and cloud has been entirely built on these organically developed net new products.

And tell us, Eric, if you don't mind, the cloud transformation, so you transform from licenses to subscription and now looks like from Informatica 3.0 almost. What does that Informatica 3.0 looks like before we get into the details?

Yes. We're a little under 50% cloud ARR relative to total subscription ARR and Informatica 3.0 would be 80%, 90% subscription ARR comprised of cloud. So we're -- we've got a ways to go against that clearly. But in the not-too-distant future, we'll get past that 50-50 mark with all our product development, go-to-market, and partnerships centered on pushing cloud, cloud, cloud. It's just a matter of time, and it requires a bit of good execution along the way, but that's exactly where we're going with Informatica 3.0.

Got it. Got it. I know that you're the CFO, but you're also an engineer undergrad, right. You're an engineer or...

Organic chemistry major.

Okay. All right. That's close enough. That's close enough. That's really, really tough. I mean there was benzene rings, man, I mean, pretty tough stuff. I guess, so what I was going to ask you was the cloud product road map versus the on-prem road map, is there any difference there that could incentivize the customer to take the cloud churning as the first blush as opposed to on-prem?

Yes, a couple of things. So first of all, the vast majority of our R&D is not cloud. So there's a small amount of, call it, sustaining engineering on PowerCenter. And so like every quarter that goes by all the feature innovation on Cloud DI brings it further and further ahead of PowerCenter. And then, the cloud platform itself, it has -- it's infused with artificial intelligence, as another one of the design tenants. And so if you purchase Cloud DI or any other cloud product that we offer, you get the benefit of the cumulative learning on the metadata that we manage and you get that benefit automatically. So as you look to perform your cloud DI or data quality, you're getting the benefit of learning from all the customers that have come before you. So it's a beautiful AI network effect. And so that's also compelling. Of course, then you get the classic total cost of ownership savings associated with multi-tenant cloud. You jettison your infrastructure your specialized support teams, et cetera, and you only pay for what you actually consume. So those are all the advantages per se.

Got it. Got it. I'm curious as CFO, you probably see data integration within Informatica. How is that working out in Informatica? How heavily are you invested in your own cloud internally to do all the integration of data and quality?

So very heavily. I'd say that we run our business with north of 75 different systems, all except one, are cloud-based. And so the 1 system that we use to run our businesses is not cloud-based is our license key generator. We really couldn't procure that from someone else. But...

Every for SaaS apps, the new company.

More or less, yes. So like, for example, I was there when we did the whole pre-IPO. We took on on-prem, believe it or not, legacy PeopleSoft ERP system upgrade to Oracle Cloud ERP or Workday, you use your Salesforce, you use using Clarity and on and on and on, all cloud-based applications. And so what we do is we take that data, we pump it into a data lake. We use Azure and then we use analytic tools. Combination of Tableau and now more recently, more Power BI. That basically powers the intelligence that we use to run our business. And so we use our cloud DI to build that data like use data quality, data catalog. And so we practice what we preach. And of course, we use our cloud MDM. As we set up our mapping of relationship with customer accounts and customer targeting, it's a very complex exercise and hierarchy, who are you selling to where? It's like 6, 7 level. How do you acquire the lead and how do you pass it through? We're one of the larger users of Cloud MDM internally as we compare ourselves to our other external customers. So yes, we practice what we for [ HTL ].

That's great. Have you gotten this question before?

No.

Perfect. Late evening coffee inspiration. That's -- I really like how you walk through the 75 systems and that's orders of magnitude complicated integration setup you've got going on. It's not just one talking to the other, but one talking to potentially 2 or 3.

It's a classic kind of cloud-to-cloud interactivity and use of state of the art elastic data lake.

Yes. And so those Azure. I'm curious as the CFO, how you chose Azure and sort of -- it's incidental of this discussion. I'm just curious, since you have a big ATL implementation data integration, data quality, MDM, et cetera, et cetera.

Well, I think that in fairness, we use all 3 and soon 4 OCI, the 4th upcoming of the -- for our IDMC platform. So Azure is kind of #1 by -- excuse me, AWS is #1 by consumption. Azure is a second, rising quickly. And so I think using the data lake, Azure Data Lake was most sensible for us. And I think also our internal developers had some marginally higher affinity to that versus the overall [indiscernible] so it's a close call. But again, we do business with all the hyperscalers.

Exactly. Exactly. And I'll get to the good stuff about the things that Amit has been talking about that 85% plus of cloud workloads are from cloud-based data warehouses and not from on-premise data warehouses. We'll get to that in a second. Let's get to the macro. Let's get the macro out of the way. You've been through quite a few cycles, and I've been through a couple of cycles. We saw large deal elongation with respect to close, linearity the quarter were a little bit tough. How is it all shaking out from where you are today? Are you incrementally surprised or things sort of stabilized versus your assumptions?

I'd say roughly 2/3 into the quarter, we see the same thing that we saw last quarter. So there's no surprise, positive or negative, versus like our opening assumptions for the quarter. There's still a lot of net new bookings to be done in month 3, but that's very typical for any quarter, not just Q3. So no net change there.

Yes. Meaning, you experienced back in the loaded linearity in Q2. you assume the same to be the case in Q3, and it's not surprised you in that regard, right?

Correct. And when we look at what we have to do in our coverage ratios and things like that, it looks very similar to the last couple of years of Q3 as well. So we're trying to pattern match as best as we can, like I want else in it. It looks like as expected, and it looks like a Q3 of prior years.

Got it. Got it. But Q2 was a surprise, but things flipped and see the sales cycle has got a longer and...

I'd say, yes, we saw elongated sales cycles. We didn't really see slippages because we're able to kind of close out the quarter pretty much as expected. We had to work harder as we went through the quarter. I think that's fair to say. There's just more scrutiny. I think that one of the things that we benefited from, we didn't anticipate this as we underwent this cloud transformation, but our business is now comprised of more lower ASP transactions. So before I mentioned $3 million, $4 million large chunky license deals that we have to close for the quarter, kind of like make or break deals. So we don't have that. We have a large number of 300,000, 400,000, 500,000, 600,000 CB deals. And in this environment, we've seen it before, where something has got a 7-digit price tag or higher as a purchasing commitment. That's kind of where the scrutiny tends to get applied. So it's not as we we're not seeing scrutiny, but we're still not doing these large chunky deals that get enough scrutiny to potentially cause them to slip. So we saw scrutiny, but not slippage in Q2.

Got it. Got it. I've been tracking this half-hearted joke all day long, and I don't think people have been laughing for it, but my recommendation for you've heard it right. Extended sales cycles and more levels of scrutinies to add more sign here fields and your doctor saying, you got to center around to the CFO. If you want to take a look at this as simple. What are you guys doing?

I mean you can sign more often, just as I was.

Yes, yes, more purchasing cycles, just more [ Signcare ], electronic and send the hello to DocuSign and get people to.

Well, it's funny you should mention because you can actually operate more efficiently. Like the old way you have like a sales rep that literally camped out an office waiting for like last signature, you get a PDF back to...

Tremendous value proposition. Right now, this is the time.

Right. Exactly. So yes, we, like everyone else. We'll use like DocuSign, for example, to kind of just reduce latency here to try to get these large deals.

Link up with an Anaplan to do the revised planning and...

We're not that integrated. We're not real time. We're like rapid batch. Okay. Got it.

Got it. Got it. So Eric, you talked about external payment terms with some customers that caused you to take a more last year -- not last year, but more conservative view of cash flow guidance. And you -- we all went through 2008, 2009. How -- what is different in this cycle? Because back then, was a liquidity crisis. People do not know they would have access to money. This time, it's not a liquidity crisis. It feels like, hey, price, they're going to go higher, so I should be making a long-term contract today with a tech company like Informatica. So I get visibility into my price as opposed to holding back and you guys might raise prices. It sounds like there's counterintuitive behavior, not only with you guys, but from other companies, customers too. Where are you today as you reassess your cash flows and you're ready to collect invoiced customers?

A couple of comments there. So first of all, we want to make sure we draw a distinction between kind of delayment in payments and like ultimately getting paid. We're still ultimately getting paid. We have very diversified, high-quality customer base. So there's no new issue with bad debt or anything like that. And for us, it's interesting, if you look back at 2021, you may recall, we actually did a much better-than-expected job in operating cash flow. We actually produce more working capital, positive working capital than originally expected. So in Q2 last year, we had brought our DSOs down to 72 days. And typically, we're in mid-70s. And so in Q2 this year, we thought for the year, we could hold it at roughly that same level, in Q2, people wait paid us and our DSO being 76 days. So 4 days DSO delta for us on the AR balance at the end of Q4 translates to about $25 million cash flow delta. So we saw that in Q2. I'm assuming we have elevated DSOs. It doesn't improve between now and the end of the year. So that's why we reflected that in our revised cash flow guidance update. I have informed a view as to 2023 yet, we've just said, look, it remains elevated through the end of this year. So I think that other companies may have seen it. I think we had a relatively better-than-expected starting point last year was a year-over-year comp on our DSOs. And so we started from a better or lower-than-expected number, then we reverted back to a higher number. That's kind of the story for Informatica.

Could you not insert inflation accelerators, inflation penalties, inflation induced penalties, late payment that is based on some cost of capital...?

We, I guess in certain contracts. We're generally pretty customer-friendly. We always have the tension on the sales team to get deals. I mean they're still paid monthly. So there's always that tension on the other side, which is healthy.

Yes. I want to talk, I want you to talk about the consumption model. It was actually a bit of a surprise that we first talked about that accounts for 50% of net new cloud bookings, and in aggregate terms, 30% of cloud ARR. How do you structure your consumption contracts? Is there not a base minimum and then some variability? Or is it all variable? How do customers structure these contracts? And how do they eat into these segments?

Yes. It's, first of all, as a backdrop. I mean we've historically been very fair fully viewed by the Gartners of the world in all dimensions, but then pricing complexity. If you may recall, 2 to 3 years ago, Gartner is like, "oh, their pricing is too complex, too many licensing scalers." So we introduced consumption-based pricing about 1.5 years ago to address that. So our IPU, Informatica Processing Unit, I think of it as a token-based system and a customer will pay for a certain amount of IPUs per month, call it, 100 IPUs per month. They will commit to a 2-year contract. So they'll -- and our 2-year contract will invoice year 1 up-front, and then 12 months later invoice year, year 2. So the cash flow is much like our normal self-managed. In fact, it's identical in terms of cash flow characteristics. The rev rec is perfectly linear. The way it works as an IPU is interchange for any combination of services, cloud DI, can be data quality, enterprise data catalog, et cetera, use an IDMC and they can change and modulate demand in real time, quarter end, month end, whatever they want, spin it up, spin it down. It's all interchange and drawn against that IPU balance. And it's a take-or-pay monthly balance. This is the key thing. So if they only use 80 IPUs in a month. The 20 unused disappear. They don't roll forward. So that's how we get perfectly linear. So it's not pure after effect consumption-based price, which you see with the hyperscalers, it's a variation on that. But we know when we signed the contract, the ARR, ACV, TCV is known with 100% certainty, and the month-to-month linearity is assured because of the take-or-pay characteristic of the month.

So it's 80% divided by a number of days, months and the last day, they don't use of the 20, boom.

Yes, that additional 20 IPUs in that given month kind of disappears. And then calendar day 1, you restart the clock, it's 0 usage. You have an entitlement of about new 100 IPUs for the month. And then you go up. Now...

Rev rec is based on the 100 IPUs?

Correct. 100 IPUs.

Okay. Got it.

Now -- so that's what happens if people use less than. And by the way, we spend a lot of time up-front sizing because we don't want to oversell IPUs. And so we've got rigid kind of control and calibration over that. We kind of take that out of the sales team's hands and bring in a different team to do that.

They overextend their contracts and to 120 IPUs?

Will allow some excess. But if we see, for example, like 3 months in a row greater than 15% to 20% usage, we'll send them a bill and say, look, it looks like you have change the usage profile, we need to upsell you go from 100 to 120 IPUs, this could happen midway through the contract 18, for example. And so at that point in time, we get a small step-up in ARR for that extra 20 IPUs in this example.

It's like dealing with my kids. You get 45 minutes of video games a day goes to 50, 55. You have to sleep at 10:30, 10:35, 10:40. Overages. I should slap a consumption model overage price and take away some of their...

Higher bandwidth, I think is clearly.

Got it, something like that. So we talked about the consumption model. We're now what, 9, 10 months into this macro funk. How are renewals happening for some of these cloud contracts? At this stage, these are probably renewals of deals that were down in Q3 last year. How do you see the pipeline shaping up for renewals? How our customers, especially on the consumption business model, you're likely to see changes? Maybe I'll size down a little bit or maybe keep the same level of commitment, but you're seeing the Mongos and the Snowflakes even the Datadogs walked in wane a little bit. What is your expectation of how these consumption contracts go for renewal?

Again, we signed a 2-year contracts. So the first kind of 2-year cohort will come up from new in Q1 next year. And so that will be our first real set of data points. What we're seeing is pretty good gradual adoption over time. Again, the day that you sign the contract, you really don't see much consumption in even month 1. They may have stood up some of their ultimate application in a preproduction mode so they can flip that in, partially start consuming IPUs, but no one clicks into 100% of their IPUs in a month 1, 2 or 3. So it scales. And so they seem to be scaling overall the rate that we're seeing, and we feel that we don't have to get to exactly 100% usage on month 24 when the renewal is due. We don't necessarily tell people they can do a partial renewal. That's the other thing too. That's not part of the contract. And so if somebody is short, they may have the option of renewing at par or not at all. And again, we operate mission-critical production applications. We have a, I think, a good amount of leverage in that scenario should there be underutilization. The other thing too is like you just don't allow it to get to that point because now we see like we're with these installations should ramp months 3, 6, 9, 12, et cetera. And we've added adoption services to our team. It's part of the business model transformation. There's a whole new function we didn't have before. And so like as soon as the sale is completed, we've already kind of calibrated the IPU usage based on the use cases. So like the customer adoption team has a runny head start in terms of working with the customer. And that's very much akin to what you see with the hyperscalers really to the adoption. We want to encourage utilization and burn down from day 1 here. And so we have to do a good job on that to get to par value renewals and of course, beyond.

Got it. Any reparations, especially after Snowflake talked about lower consumption trends with customer-facing, consumer-facing e-commerce companies. Obviously, you guys are at the front end of that whole cycle. So if they're seeing less consumption, you should have seen it even earlier or maybe it's very, very quick? Any correction with Snowflake and what their customers are going through with respect to slow down? And does that pertain to you, your business? Or are you more insulated from what's happening with Snowflake consumer?

Yes, I think the usage is it's a little bit different because I think a lot of the use cases would be data going from on-prem system to Snowflake for advanced analytics. Typically, we're serving production applications. And so there's not as much of a kind of a discretionary ad hoc analytics characteristics built to it. It's a little more...

You say production is it OLTP?

You're taking and transforming data necessary to drive business processes as opposed to performing some kind of what-if analysis. And so that's more kind of ad hoc. I would say, as opposed to look, I have this process. I need to run it, the data has to transform day in, day out. And so like we look at our transaction count. Let's take another proxy for IDMC and that's scaling at or ahead of our cloud revenue growth rates, which indicates it's a positive, only half that rate that might indicate a potential problem. That's not what we're seeing there. So it's a slightly, we're in a slightly different part in the, I guess, the data chain relative to, I think, what some of their customers have.

Got it. Got it. Next I want to talk about, I want you to talk about CLAIRE, the AI engine. At this stage of the economic cycle when labor costs are going higher. IT talented short supply, is there wood to chop with the AI and CLAIRE?

Absolutely. I think we as part of the overall business transformation, circa 2015, 2016, we actually decided to do something that hadn't been done before infuse as a core service, not as an optional module, but it's a kind of a core service in IDMC AI. And again, you think about what we do, it's -- you just think about managing federation of data from like a dozen on-prem systems to like 3 different cloud systems and all the permutations in between, and then you enrich the data as well. Like identifying the data, understanding like what's the customer record, what's the correct address, finding a match and tracking that lineage. So you end up with client, a very clean, pristine well-defined data set. If you do that repeatedly time and time again, this is like what we do on CLAIRE, like cloud DI running, get it again, the algorithm sees more and more things, these more and more matches, which get more human validation is like correct, not correct. That's like how you train a learning algorithm you throw a ton of data at it, it pattern matches, and then to give them absolute examples, binder, yes, that's correct. That's wrong. And I think it gets better and better. And so in our business, I think it's like really the only way to scale this problem because we're dealing that with a growth of data, like the technology set can handle that and the price performance and everything, all those curves are bending down nicely to support just more data. What we're dealing with is like data sprawl, data complexity, right? People are using data from all types of sources, internally generated and externally procured and enhanced. And so like trying to curate and understand it, you can throw a lot of bodies at it, throw machine algorithms at in. So we designed the platform to employ machine algorithms.

Something you can charge more. Is there a premium price? Or does it come with the platform?

It comes with the platform. It's not a separately priced SKU. So it's very different. Yes, it's not like an option. It's built into. And we -- our teams, whether it's a data quality dev team or DI team or any other team uses and leverages AI in the cloud and the cloud platform for learning algorithms and better performance of their apps. So the first time a customer subscribes items, like I said, they get the benefit of all the learnings have gone before. The anonymized data, it's metadata, to be clear, and the algorithms benefit from that.

Any thought to pricing leverage? Do you have pricing leverage with your product family?

I think we do. I think number one, you see evidence in our actual just primary renewal rate, 95% of maintenance, 94% on sub overall. Our cloud renewal rate incidentally is higher than that average. Our cloud NRR rate, there is higher than our composite average kind of as you would expect. The business has proven to be quite resilient. Maintenance renewal rates are higher than expected. So even as we've kind of moved away from perpetual license generation, that installed base, again, because we're running, again, mission-critical applications. In some cases, they run flawlessly for like 8, 9, 10 years. The last thing someone is going to do is not pay the maintenance bill in that, again, mission-critical workload. So that's a form of kind of pricing leverage. We don't have to negotiate renewals. If we get par value, we have 5% attrition. But otherwise, we get par value. So that's the form of pricing leverage. The other form of pricing leverage is we've had this built into our sales compensation plan for the last 3 to 4 years. Our sales team will get paid multipliers. So not every dollar of ACV is treated equivalent. And a breakpoint we've had every year relates to discount from list. And so if you're a salesperson and you're going to sell a product and the breakpoint, I'll just make up a nice round number, let's say, the multipliers is 50%. Let's say that at a discount greater than 50%, you'll get a 0.75x multiplier on the deal. And if it's lower than 50%, you get a 1.5x multiplier the deal. You'd be surprised how many deals you'll see with discount.

Like 49...

49.9%. People figured this...

[indiscernible] there.

Yes. The other thing about complaint, you can't have to put to any. You have to resist the temptation to like overengineer. They have to be relatively simple. So in this case, it's really simple, over or under 50%, like anyone can understand that. So we've had built in that pricing leverage intentioning for the last 3 to 4 years, and it's actually served us well. And you see that reflected in our non-GAAP gross profit margins, which at 80% plus or minus. And we have a hybrid model, like no perpetual, some on-prem, some maintenance, some cloud, blend it all in 80%. That's pretty healthy. So it indicates pricing leverage and also pricing discipline.

Yes. I want to touch on the maintenance installed base. it's a blessing and a curse, right, it occurs because it's a big anchor on your growth rate. It's a blessing because it could be something. Where do we stand now? I had a chance to come out Informatica world and good buzz, good vibes with your customers and partners. So you have some goodwill. I mean Informatica is a company that takes really good care of your customers. Good job chief customer officer. It happens to be somebody that I've known from my grads school. But how do you take that goodwill that you've built with your customers to help them, especially ones that are sitting on maintenance to see the light at the end of the tunnel? Because that 1% moving to the cloud going to 2%, that's going to take a long time. It feels like you need not linear progression, but not only near progression like geometric, there's got to be some buzz. Look at these customers that just switched out maintenance and vendor cloud and boom, everybody else wants to do it, could be in for some nonlinear progression with maintenance conversions.

Yes. It is possible over time. I mean, and you're right, it's a blessing and a curse because when people look at our GAAP revenue growth rates, it's muted by $540 million maintenance ARR, growing at minus 5% per year. So if you have 2 net new perpetual license going in, you've got $0.5 billion decline. Mathematically at 5%, and that's kind of best-in-class, and you really can't do better than that. And that's, by the way, we have an abundance of ARR metrics. And so you can decompose sub-to-cloud, self-managed, et cetera, separately for maintenance. And so it is kind of a tale kind of 2 sides of the org. So here's the positive way to look at it, put a positive spin out first, and I'll directly respond to your question, Kash. So we've driven a 40% growth in cloud business. That's 90% based on net new workloads and only 2% on migrated workloads. So that's a pretty positive thing. It's like we're driving this business at scale, profitable and there's net new work that's out there just as there's healthy demand. We've only migrated 2% to date. We are not end of supporting the maintenance space. We've made a decision.

So you say 2% -- is it 2% of the maintenance base revenue or 2% of your cloud revenues that came from maintenance.

It's 2%, the maintenance migration has translated to about 2% of the cloud ARR that sits today.

Okay. Okay. So is there a way to look at the maintenance dollars that have converted over?

Well, I'll give you the rough [indiscernible].

Should [indiscernible] because it's a 1.4 multiple, right?

It's going to be about start with 2% of maintenance, roughly 10% migrated to roughly 2x that in terms of cloud.

Got it.

So 10 going translating ultimately to 20. So it's more than 2% of cloud ARR. So it's like 20 over 3 75.

Okay.

We've decided not to end of support...

20 was sitting in maintenance is something smaller than that divided by 1.4.

The maintenance, it would have been, yes, 10 out of 500. So that's the 2%.

Okay. Got it. Right. That makes sense. Okay. We had it drawn math, you got it right. Yes.

So the question is like, should it be a carrot or stick approach. Like the stick approach would be, look, we're going to end of support. We're going to force people to migrate. And that's akin to telling someone when we're going to turn off your ERP system unless you move to cloud, like really, really fast. And like these are complex embedded applications. So we're not just copying data from A to B. It's a combination of data and process. It's the migration of the entire app and a mission-critical one. And so we would really kind of disadvantage and torch the customer base, frankly, if we adopt that policy. So we're not doing that. The carrots are what we mentioned before, all the innovation, infused data in the platform better feature investment on cloud relative to the on-prem. And every -- I think, every quarter, every year, the concept of TCO cost pressure. Ironically, the environment we're in kind of works a little bit more in the favor of finding efficiencies over time. So I think that, that could definitely help the progression, the rate of migration because 2% after 1.5 years, 0.5 billion would be at this a long time to get to like 75%. The other things that we're doing to accelerate is we continue to invest in tooling, automation. We're expanding the involvement of SIs. And what we've concluded is that it's not going to be the Deloittes and the Accentures of the world that are really going to jump in to these migration programs. You're going to have to go to like a different class of lower cost, more efficient SI partners who are actually happy to take kind of cloud migrations because these are low-dollar, low-margin work efforts here. And so we're kind of expanding the partner ecosystem working with us on these transactions. And this combination is saying, we will continue to chip away at this. Make no mistake about it, but we expect to see some acceleration over time. We haven't said we'll get to x percent by y year. We've said that we'll do this as quickly as the customer permits because the tail end of this process requires the customer to do testing and user acceptance of the new application. And just like I went through a testing user acceptance process, my cloud ERP system, moving on-prem. I assure you, we spend a lot of time I had to get that right. That's kind of the level of effort that only the customer can put in. And so not all kind of migrations are created equal. This is a very sticky operational high degree of difficulty migration for these reasons.

I know we've got two more minutes. Any questions? Do you have any questions?

I was kind of curious to hear [indiscernible]. So [indiscernible]

Yes, the future of ETL and especially given Kafka and if you could work with Kafka.

Yes. We've supported Kafka streaming data for years. So to us, this is nothing new. It's just another form of flavor of data set, and all the hyperscalers have Kafka. Their flavor of Kafka embed in their platform. So like Kafka is kind of everywhere, it's been for quite some time. And for us, it's just another data source and it coexists along all the other data sources. And so we don't see any kind of change in kind of relative mix. Here we see streaming data sets. We see static data sets. And so we'll continue to support all as we always have, right? We've always been, again, kind of the Switzerland of data management. There's a new data persistence layer or features to feed data into these data persistence layers. We're typically first in line right, working with that third party in advance to kind of understand and optimize for the features. So that's just what we do, and we'll continue to do what we've been doing. And so I would kind of the other part of the question is that the future of ETL. I would just say that it's kind of funny that the phrase ETLs come back in vogue. It's really popular now. 7, 8 years ago, went out of vogue and now it's popular again because people talk about ETL these days in the context of data goes from on-prem called legacy persistence layer A into next-gen cloud into like Snowflake. So like ETL, the acronym is once again cool. But it's like the same basic thing that's been going on for years and years. It's just that there's a different source here that we didn't have in the past.

Wonderful. And that, and I think we're at the end of 40 minutes. Thank you, Eric.

Thanks, Kash.

Thanks for coming down. Appreciate it and spending time with us, and thank you for your questions as well. Drinks on me at 6:00 down the lobby level. We have a client who is actually going to be performing. We heard Max Jellinek perform. Guy is genius. Not because he's a client. Really is awesome. So see you all down there.

Thanks for that, Kash.