a value prop that allows for these products to be a meaningful contributor. So big part of the story as we enter 2027, because that can be a huge contributor on the cross and expansion side.

Got it. Appreciate the feedback.

Thanks, Parker.

And your next question comes from Matt Van Vliet with Cantor Fitzgerald. Your line is open. Please go ahead.

Good afternoon, guys. Thanks for taking the question. I guess following on a couple of these others, but as you look at the cross-sell pipeline ahead of you, and Matt, you mentioned customer engagement continues to improve. I guess if you had to stack rank where the opportunities are going ahead between expanding digital banking footprints, selling in some of this relationship pricing, but then also including things like Innovation Studio and some of these AI SKUs, you know, where should we expect the most movement in terms of revenue growth over the next couple of years? And how might that dynamic change as you look out, you know, five plus years from here?

Well, five years, Matt, I don't know. I don't know if I can do that. But I will tell you that I will say that it's a tough question to answer, which is probably a good question. I was with a bank a couple weeks ago that is running our commercial product, and they're running a legacy product for retail, and they just said it's so frustrating for our commercial customer that uses our retail platform because the commercial product is so much easier to use than the retail one, so we just need to switch and get it all on a single platform. There's opportunities like that all over in our customer base, especially in the upper end of tier one in enterprise where we can go cross-sell those products. Then you get into the fraud conversation, and you talk about what I said earlier about the power of the single platform in the fraud business. Relationship pricing, clearly we're beginning to show a lot of success there. Innovation Studio is an endless opportunity with our customers with more than 200 partners in there. And then you add in our growing confidence in our AI capabilities and our AI products. There's a lot there. I would probably think that fraud, Innovation Studio, and our AI products will certainly have a very long tail on them, as I said earlier. But digital banking, you know, we're adding new deals every quarter, and some of them are buying one aspect of the product rather than the commercial or retail. So there's just a lot of opportunity in the breadth of products that we're rolling out, and, you know, our focus on customer experience seems to be a big differentiator for us.

All right, very helpful. And then, Jonathan, on the margin expansion, obviously, it continues to be very strong here, but you highlighted, you know, OpEx continuing to grow. Obviously, the business is growing, and that's a requirement. But how should we think about the pace of OpEx growth over the next couple of years? Are there any major sort of step function investments you feel like are needed, or have you made a lot of those? and a lot of the internal efficiencies you're growing could see greater upside as growth continues and you just get sort of general leverage in the business?

Yeah, I mean, it's sort of all of the above. I mean, when you think about now what we've done year-to-date in 26 and with lifting our EBITDA guide for the rest of the year, you're seeing some pretty strong outperformance in terms of operating leverage, both from the gross margin step-up in the first half and ongoing OPEX leverage in the back half. As you step forward beyond 2026, I would continue to lean on our long-term framework, which we still have conviction in. I mean, obviously, with the outperformance this year, the bar is higher, but we feel good about the continued opportunities. And, you know, obviously, there is a new, I'll just call it, line item in the P&L when it comes to AI-related infrastructure costs and token costs and the like. And when I think about the guidance we have for the rest of 26 and how we're thinking about planning for 27, we think we have a pretty good handle on that incremental spend relative to obviously a couple of years ago where it didn't exist. So right now it's, you know, we're just focused, heads down, executing and think there continues to be a margin expansion story here. And obviously the investments that you've talked about just now that we continue to make even here heavily in the third and fourth quarter of 26, we think are foundational to continuing this subscription growth trajectory that we're on. So we're kind of trying to do both concurrently, and we feel good about that path. So I wouldn't add anything quantifiably beyond 26 other than what's already out there in terms of the long-term framework, though. All right, great. Thank you.

And your next question comes from Dan Perlin with RBC Capital Markets. Your line is open. Please go ahead.

Hey, guys. Good evening. Great quarter, and congrats. on repaying the convert i'm sure that was a huge burden to get rid of um you know jonathan i actually wanted to follow up on exactly what you just said about token costs um and understanding the dynamics about how that might play through over the course of the next 12 18 maybe 24 months so like maybe can you just talk about how you've gotten comfortable with that what the model looks like i'm assuming it's token plus a margin uh token cost plus some sort of margin but just Anything you could provide there as we think about that part of the business ramping, it feels like it's a sizable blind spot for a lot of clients also. But at the same time, you guys can maybe provide some scale to those clients, and therefore it's a huge benefit to them. So just any of those dynamics would be helpful.

Yeah, it sounds like you're talking about the cost structure of the external products. Most of what I was referring to in the prior answer was really about our cost internally when it comes to everything we're doing with AI, including building those external products. In terms of how we're managing the cost structure, again, we're in EA on the three that we've announced and talked about. We think we have the proper guardrails in place in terms of caps and the ability to throttle token utilization embedded in the product, and we have a pricing scheme that will manage that. Internally, we're very, very focused on model management and enablement of the organization to optimize for things like caching and effective prompting, ultimately to just bend the curve of what we've seen throughout latter part of 25 and through here July of 26, where the run rate is exponential in terms of the spend compared to just a year or two ago. But now we're just figuring out the right mix of what model to use for what team, for what use case, and that gives us confidence that we'll be able to be in the right place when it comes to the amount of spend in totality, both for internal usage and for building external product and for customers as they use those external products that will be in a healthy place. But especially on the external side, we're still very much testing this in the EA phase.

Yeah. No, that's really helpful. Thanks for that clarification. Just a follow-up. It's a little, maybe a little bit more of a one-off in a lot of ways, but, you know, there's, there's so much demand that you you constantly have been talking about so this isn't like the only quarter it's been several quarters now for some time um and i'm just wondering like is the go-to-market motion for the team and like the sales force efficiency are they running hot and and to the point where you need more potentially out of them and therefore you might need to expand or are you comfortable with the the team you got on the field to get you to kind of the growth goals that you've um you've provided to the street but at the same time If you gave a little bit more, would you be able to throttle up that revenue growth and maybe arc it up as we go into 27?

You know, Dan, I think we've got good coverage ratios where they come in. And also, remember, as I talked about earlier, it's not just digital banking. It's relationship pricing, fraud, Innovation Studio, the AI products. I feel good about the coverage we have right now based on the number of deals we're doing and coverage. But if we see, and as you said, this has been multiple quarters and we continue to deliver, but, you know, more sales reps doesn't always solve the problem. It, you know, shrinks commission rates, the amounts of wins. And so I want to make sure, as a salesperson myself, I want to make sure they can make as much money as they possibly can. But I don't see it as a situation where we're understaffed or under-provisioned on the sales and go to market side. The success team and the sales team and the specialists and everybody involved do a great job. I'm happy with what they're doing. They just got to keep it up and continue to make sure we keep winning.

Yep. Excellent. Thanks, guys. I appreciate it.

And the next question is from Michael Infante with Morgan Stanley. Your line is open. Please go ahead.

Yeah. Hi, guys. Thanks for taking my question. I wanted to piggyback on Parker's question earlier, mainly because if I carry forward that low 20s net new subs error range from the first half throughout the balance of the year, I think I get subs error sort of exiting 26 at or above 870, which is around, you know, sub 12% growth. So I guess the question is, just given your reiteration of 27, like, are you implicitly saying and or expect that the back half from a net nearer perspective will accelerate? And how should we be thinking about the key drivers of that?

Well, I mean, when you look at the 2027 subs revenue growth guide of 12.5% to 13%, and you think about where we were in the first half of this year, which was a meaningful premium to that, that actually implies a modest decel in the back half when it comes to those metrics. But again, we feel good about, you've got to remember, in the Q3 and Q4 periods, A, we have tougher comps than we faced, especially in the second quarter, but even all of the first half. But we also expect a higher mix of these larger tier one and enterprise deals. So we definitely think we're going to, assuming we execute on the bookings front, that we're going to be in a good position vis-a-vis that 27 goal. And obviously, all hands on deck to meet or exceed that. But when we think about sort of where we were from a subs ARR perspective, call it this time a year ago, you can kind of see the leading indicator is telling you what that means for subs growth here in the back half of 26. And we'll see where subs ARR comes in based on our bookings. To your right, the incremental subs dollars that we deliver here in the Q3 and Q4 timeframe. but you know the baseline's going up the law of large numbers is is present and so we got to execute at a higher and higher level as time goes on and so that's why we go back to all of these other products we have the cross sale opportunity the continued execution on the net new front because yeah you sort of have to see that success on all those dimensions to continue seeing this business grow at the levels that we've talked about and and higher so hopefully that answers your question, Michael. Let me know if there's any follow-up there.

No, it does, and it makes sense in relation to the slope of net new last. You're going to back up as well. Just the second one, you sort of alluded to it being early in terms of you internally sort of figuring out some of these optimization dynamics on the actual compute cost, but have you sort of learned anything incrementally as it relates to the gross margin profile of some of this ad-delivered functionality? Do you expect it'll be, you know, broadly in line with the aggregate business? Do you think it'll potentially be accretive, depending on what you're able to do internally from a model routing perspective in some of your internal sort of ML use cases? How are you sort of thinking about it early days?

Yeah, I mean, you framed up all of the work we're doing. It's a difficult question to answer today, not just because it's early adopter, but because it may vary wildly by product. And then the question then becomes what products scale up that ultimately become meaningful revenue contributors, and then it may be accretive or dilutive to gross margin accordingly. So it's a pretty difficult question to answer right now, given where we are on all three of them. I can tell you when it comes to all the AI usage internally and managing that, not only optimizing for where it falls on the P&L between cost of goods sold and primarily R&D, but OPEX otherwise, we're doing a lot of work there. And then we're being thoughtful around how do we make sure that we're using the right model and the right level of spend for the right use case? Because in the early days, there wasn't a lot of management around that, admittedly. And so now as these numbers are getting larger and larger, making sure that we're being thoughtful about it and working with the teams around the ROI when we think about how much we spend to deliver a certain outcome, it's just a muscle we're building in real time. And that's going to be a big part of our journey here over the next, for sure, six to 12 months.

Thanks, Jonathan.

Thanks, Michael.

And your next question comes from Chris Kennedy with William Blair. Your line is open. Please go ahead.

Good afternoon. Thanks for taking the question. Last quarter, you talked about a large fraud deal being larger than a digital bank customer. Can you just provide a little bit more perspective on your relationship pricing business, the size of those types of contracts, especially as you have a lot of momentum today.

Yeah, I mean, maybe use the example that was cited in the script. When we think about that top 25 bank, you know, that deal, you know, would look like a very large tier one digital banking deal. I mean, obviously that's a big institution, so it's, you know, a good example of one that should naturally be large. But like Matt said earlier, when you have nine of the top 15 banks for that product, it skews to the enterprise banks and it skews to higher ASP. So when you're asking about relationship pricing, obviously it's a top-heavy product in terms of the customer base, but that also means from an ASP perspective, it's going to be representative of a pretty large tier one digital banking deal or larger.

Got it. Thank you for that. And then now that the balance sheet is in a really good position, can you just remind us of kind of capital allocation priorities and kind of build partner versus buy, especially with Innovation Studio partners?

Yeah, I mean, we feel great about the position we're in. Obviously, alongside this earnings call, we've announced the board authorizing an incremental $350 million share repurchase program or on top of our existing program. So we feel good that when it comes to that opportunity, we can be thoughtful and at the appropriate times exercise that lever. And then at all other times, we now have, we think, the flexibility and the scale to continue to invest back in the business. And you directly see that based on our 27 EBITDA targets. You can see the margin expansion is lighter than what we did in 24 and 25 and what we're now guiding to for full year 26. So that's an example of us reinvesting back in the business to elongate this growth curve. And then when it comes to M&A, we think now with the debt paid off and the free cash flow generation, As you've seen, we haven't done deals, and it's been over five years since we've done anything, so we're not going to do deals for the sake of doing deals, but we certainly have the capacity and the ability to be opportunistic now when the time comes. So long-winded answer, but the reality is it's all three of those things that we're going to be able to optimize for now going forward.

Thank you.

Thank you, Chris.

There are no further questions at this time. This concludes today's call. Thank you for attending. You may now disconnect.